diff --git a/MLQA.ar.ar/xtreme-test.parquet b/MLQA.ar.ar/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..888c193578d08d04d16d5f20c8e0c8304d3207e6
--- /dev/null
+++ b/MLQA.ar.ar/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b0e8c03f6bf7f5c4bdff4539a151738c15bedb2de96cc8908a8d12e6df003828
+size 3691484
diff --git a/MLQA.ar.ar/xtreme-validation.parquet b/MLQA.ar.ar/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..2e6638ede0e8f3a9817d856c194db005e502c740
--- /dev/null
+++ b/MLQA.ar.ar/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e22fbf8f603a28923c9cca10dc3b7b90f33b44631b1b125df8403eb7e4fc8476
+size 356606
diff --git a/MLQA.ar.de/xtreme-test.parquet b/MLQA.ar.de/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..42ff9cc498f1c9c83b965dc8c46a1a37773df5ad
--- /dev/null
+++ b/MLQA.ar.de/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a966feae84a20bd98658a88ff95efe0ab240999e8825734cf5a933e0a86c8774
+size 1041882
diff --git a/MLQA.ar.de/xtreme-validation.parquet b/MLQA.ar.de/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..f67486a24a7bdfbb885dd4d8083c76a60d6b3c0c
--- /dev/null
+++ b/MLQA.ar.de/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f9b3557a1ce13c385ed6dea70f8fe5db86424447a40361b61c304a5431a877fa
+size 150893
diff --git a/MLQA.ar.en/xtreme-test.parquet b/MLQA.ar.en/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..c1c616eef0f2159a9080d4fb11616325f6574c9e
--- /dev/null
+++ b/MLQA.ar.en/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:97cecff3f07542b97e06fb5d90e9859401448a8d6216e4b9343a627f88623396
+size 3645879
diff --git a/MLQA.ar.en/xtreme-validation.parquet b/MLQA.ar.en/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..ed3c7e82f2f5f65310793b9a986ccd36b65c5cd4
--- /dev/null
+++ b/MLQA.ar.en/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e882d0a0c01f69ddfd1dbaea260fcd484a37ea827aa1af283a613bc330a6b3dc
+size 352039
diff --git a/MLQA.ar.es/xtreme-test.parquet b/MLQA.ar.es/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..409148af475091846aae42f985d871bc3281608d
--- /dev/null
+++ b/MLQA.ar.es/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9c4ea47a0f7ac6122672d1aff49c1c17d734de7f99520e9dd811a4deeb9cfdc7
+size 1415604
diff --git a/MLQA.ar.es/xtreme-validation.parquet b/MLQA.ar.es/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..6fa850cae6693b6193a368924fd3618df808c505
--- /dev/null
+++ b/MLQA.ar.es/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a4c3383d85e9ef4262529b97e407cd5e3698993a02fe66a46a2654205e800204
+size 116009
diff --git a/MLQA.ar.hi/xtreme-test.parquet b/MLQA.ar.hi/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..00eaa65654a8d68c84ba4a8855db8995898f26e9
--- /dev/null
+++ b/MLQA.ar.hi/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d1c7ad8aaf14aa3b069789d906d04be379d043e4df6d4ddb5bef207099ed9cc2
+size 1243845
diff --git a/MLQA.ar.hi/xtreme-validation.parquet b/MLQA.ar.hi/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..e928d4259b949a5a4aa0d5c81f3fe58e22fcb6b2
--- /dev/null
+++ b/MLQA.ar.hi/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4bb3c6a67a66bc14384896159e055b2486f852e49a7678ab2e8e36ae15d648f8
+size 125863
diff --git a/MLQA.ar.vi/xtreme-test.parquet b/MLQA.ar.vi/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..653da6486961d65ee067bdaa882fab40776b70a8
--- /dev/null
+++ b/MLQA.ar.vi/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ebd8690a1c1a1f934c45661c088fbfa4a3e3549b2c1aafc1b4b88acebeedb9f5
+size 1525767
diff --git a/MLQA.ar.vi/xtreme-validation.parquet b/MLQA.ar.vi/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..8f10244d809a7d1445ae48553016199c2fde3d23
--- /dev/null
+++ b/MLQA.ar.vi/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e06908b64ead213aab488a1827133d83bbe5e3aa15bdbe6a0dc5b0dd7e3785b1
+size 141413
diff --git a/MLQA.ar.zh/xtreme-test.parquet b/MLQA.ar.zh/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..507c2902b0678ec9d03af062644fa504ff1f4299
--- /dev/null
+++ b/MLQA.ar.zh/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7dbc3d44e9ffa1bfba4e3d0b5c142817b1a90675a7f59c4759a0244a8b3a136a
+size 1449602
diff --git a/MLQA.ar.zh/xtreme-validation.parquet b/MLQA.ar.zh/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..c64bc27ac26cf530a22c3c4b86e7a1f470106d72
--- /dev/null
+++ b/MLQA.ar.zh/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:66661f1dbd95ddeab856e970f676a554d508749ebaec3337bc516ce8393856fa
+size 141795
diff --git a/MLQA.de.ar/xtreme-test.parquet b/MLQA.de.ar/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..7b74a3d13521fb3f2ece728c72d1a19bbbac3e4b
--- /dev/null
+++ b/MLQA.de.ar/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ce16a4fcd92e1e3360823bf1b42566f987ccffeaf8cb11b896dff18666ceb0b0
+size 928693
diff --git a/MLQA.de.ar/xtreme-validation.parquet b/MLQA.de.ar/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..89e22adb2ddf0999dbe5e4cd873d598a635b5297
--- /dev/null
+++ b/MLQA.de.ar/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8b4525133545d009b5509665d87e58bd7addef974f1101a7e9d3ae88d7299572
+size 115740
diff --git a/MLQA.de.de/xtreme-test.parquet b/MLQA.de.de/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..e2089feff78bf24dd574870970a05485833d3e73
--- /dev/null
+++ b/MLQA.de.de/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f6566f4fcea8b806f0832f995a8a4e70ef071acd7b20c47742f3afb32d205abb
+size 2512316
diff --git a/MLQA.de.de/xtreme-validation.parquet b/MLQA.de.de/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..463951eb86544ceb27be08562b544ec8767f1586
--- /dev/null
+++ b/MLQA.de.de/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a23f1dc78d39f3c70c159af93652eaedf2ccc88f459160473e7076622dcd4f9b
+size 285648
diff --git a/MLQA.de.en/xtreme-test.parquet b/MLQA.de.en/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..c9c390bf5297775a661bd6a454a9b31491ef3699
--- /dev/null
+++ b/MLQA.de.en/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8a37001672c3e616709f90b0cb43cff1e814de04929895e0aaa07287ab18575f
+size 2494618
diff --git a/MLQA.de.en/xtreme-validation.parquet b/MLQA.de.en/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..8ac8a8504d38465b725361e71b131a98e1f1aeb9
--- /dev/null
+++ b/MLQA.de.en/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3e15484a475e61a31894926bce92112ccabf81ff13b319b612b668057e89e2bd
+size 283642
diff --git a/MLQA.de.es/xtreme-test.parquet b/MLQA.de.es/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..7e547c18790af7190108437d72aa1782a91949cc
--- /dev/null
+++ b/MLQA.de.es/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aa23aab2e9fe75faf78dd493e8a094ad92d46546390192e941501013fab2ade6
+size 1004357
diff --git a/MLQA.de.es/xtreme-validation.parquet b/MLQA.de.es/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..0e30309653db114a36aa29cf6caa716e48d871d2
--- /dev/null
+++ b/MLQA.de.es/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0df5e241c27b8b375b8f49cb7c744ac5a31015c1c72cb104b3308d4adf0b40ff
+size 114344
diff --git a/MLQA.de.hi/xtreme-test.parquet b/MLQA.de.hi/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..c68928ba5623ee756ec38e3e0036ffd8e88dce71
--- /dev/null
+++ b/MLQA.de.hi/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8fcb0d1c534e37865522d122d566612e24fc3c3e80611390dd93d9e9b6b1b9aa
+size 783301
diff --git a/MLQA.de.hi/xtreme-validation.parquet b/MLQA.de.hi/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..e5be1a2c6a6e9a89f8277a0cfe8ad9d7119c03c9
--- /dev/null
+++ b/MLQA.de.hi/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:266a73a3f94ebaed7a83969635c5b0ea8a31ea3fda7e938c0c0d404283ea0951
+size 97301
diff --git a/MLQA.de.vi/xtreme-test.parquet b/MLQA.de.vi/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..c9d8705a656ebd05c7e56c0c092383376dad4b11
--- /dev/null
+++ b/MLQA.de.vi/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6f851a1b6f48b70c30b45afa8a18a17dc165994c4952a951f59109f0cf37ead6
+size 973109
diff --git a/MLQA.de.vi/xtreme-validation.parquet b/MLQA.de.vi/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..6f92f008ca19a779a143de9fa5c638c486c7c0c3
--- /dev/null
+++ b/MLQA.de.vi/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e135a5f18a06e2dd725e4745a813b51df8547c48f89da2e30b252a990831ff3
+size 135004
diff --git a/MLQA.de.zh/xtreme-test.parquet b/MLQA.de.zh/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..113fd8352da3bb59ea6fc79007910a93a5c18854
--- /dev/null
+++ b/MLQA.de.zh/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:86b4bc1b18997c025e6cf7bfc76e1aab4189f6c5144e6fc4944017026ec3e704
+size 933492
diff --git a/MLQA.de.zh/xtreme-validation.parquet b/MLQA.de.zh/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..6aa5387296c10baefda26262bbcecba18e8df50b
--- /dev/null
+++ b/MLQA.de.zh/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:001e668aa9548e7b559121aa2b0c5946177d66731a5ab131eb3e0302f7f369b7
+size 112319
diff --git a/MLQA.en.ar/xtreme-test.parquet b/MLQA.en.ar/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..31c3a6a782d9e9d551e8fe8f94a5f30e390f54f7
--- /dev/null
+++ b/MLQA.en.ar/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d19238ff0f069c66ba77fa5db36a7aa8c44917bdd21fdd5c788a4004d4a592e0
+size 3595638
diff --git a/MLQA.en.ar/xtreme-validation.parquet b/MLQA.en.ar/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..4788d9df84f581cb1da25b561412f7df4567d67f
--- /dev/null
+++ b/MLQA.en.ar/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:de69c93a862e40f6b103510ce8a5c4a4f9af0a071e0ba3e6dec3a40db8102267
+size 343401
diff --git a/MLQA.en.de/xtreme-test.parquet b/MLQA.en.de/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..d795f0438719c48edfcbfd5c02546dd8089f63f9
--- /dev/null
+++ b/MLQA.en.de/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d03a1c2ff268234c292698e6b30bb82afe55e25b03be777e2843a36d236fcfc8
+size 2887002
diff --git a/MLQA.en.de/xtreme-validation.parquet b/MLQA.en.de/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..c615539dfd8d84a733274c1ddafd4415b77d9f07
--- /dev/null
+++ b/MLQA.en.de/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d700c556bbf09a871c3215760b2ece35d2d3f86cb1fb3a96adff4032e47c432d
+size 336108
diff --git a/MLQA.en.en/xtreme-test.parquet b/MLQA.en.en/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..87b56360278f3e8fbfe7276fc71759f4fc2bdae1
--- /dev/null
+++ b/MLQA.en.en/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a4fe560ffdf12935e8697e4a0ebfc01c2c89e88db155e7ccb3dd6eab005561fb
+size 7493810
diff --git a/MLQA.en.en/xtreme-validation.parquet b/MLQA.en.en/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..84e47ce514c02caa109ede498c0cf95db1353100
--- /dev/null
+++ b/MLQA.en.en/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fc19bda6dbf5fed356ea856a936eaa60a92c44ba75b715203cd220b7726fb5ad
+size 723527
diff --git a/MLQA.en.es/xtreme-test.parquet b/MLQA.en.es/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..87846afde6f8ee8f2321daed06ca09a1c4cd7920
--- /dev/null
+++ b/MLQA.en.es/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:21071390d0f18574ffd2b0312527bf678db973209ca65f4b23b7e9d26f28e0b6
+size 3457233
diff --git a/MLQA.en.es/xtreme-validation.parquet b/MLQA.en.es/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..1f39cbe26c7a2db0a05a5ae5e5c5a3865d59d1d2
--- /dev/null
+++ b/MLQA.en.es/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aab5e4dba0e2e65da9869133b9892d4e57cf520b976352a0a47efdd389183479
+size 319497
diff --git a/MLQA.en.hi/xtreme-test.parquet b/MLQA.en.hi/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..f589ce589bd6956dd3b749cc39600d4e610cbb54
--- /dev/null
+++ b/MLQA.en.hi/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2c5390567a298783d4cb2decf0cd5f085cbfe98f834647716138b9326ff05760
+size 3194270
diff --git a/MLQA.en.hi/xtreme-validation.parquet b/MLQA.en.hi/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..3ebf860a524bf8048bab2dfad021065af3d72127
--- /dev/null
+++ b/MLQA.en.hi/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a87cc6971acf367f1c24ffd939adb7c22b0842ca9bea24087595427d6615aa24
+size 322984
diff --git a/MLQA.en.vi/xtreme-test.parquet b/MLQA.en.vi/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..0a1266529e0888f8b6ccf534d1693be65efeaf20
--- /dev/null
+++ b/MLQA.en.vi/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6d84a37d3f9f411d9b342ee8b9b4662db554e91a03218698c84a1015061a100d
+size 3815324
diff --git a/MLQA.en.vi/xtreme-validation.parquet b/MLQA.en.vi/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..45981988c4af161f074c95264554fe9ada2381d6
--- /dev/null
+++ b/MLQA.en.vi/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:15c50864f2a1c9555cee993395d6b116a18ca004ed9c8a55c214e6677bc13331
+size 355220
diff --git a/MLQA.en.zh/xtreme-test.parquet b/MLQA.en.zh/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..98f4f26161af1877087e33ac5c4ff57bccfd7d17
--- /dev/null
+++ b/MLQA.en.zh/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:810d6bbc743cd89b0fd97cfc68cd6d178944c7242d7cb23b38676563f7eb9d31
+size 3593858
diff --git a/MLQA.en.zh/xtreme-validation.parquet b/MLQA.en.zh/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..18065162408f20b9be7104c0cc8ce02ac9441e82
--- /dev/null
+++ b/MLQA.en.zh/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:59964968e4a1ad29d98f972415834a8160ce01ed6d5da4c7dc7da6f5b1554bec
+size 335168
diff --git a/MLQA.es.ar/xtreme-test.parquet b/MLQA.es.ar/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..afdcee78a92ca1a787c0e4dfb582c9c0a4e76222
--- /dev/null
+++ b/MLQA.es.ar/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:82f40364a5d588858d840bdb3fe19877f157cdd97f4f88ca8b4c9824f5ff8c65
+size 1016550
diff --git a/MLQA.es.ar/xtreme-validation.parquet b/MLQA.es.ar/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..b93297a7a7c950ee8e848ec59da511adc7917282
--- /dev/null
+++ b/MLQA.es.ar/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1d61a3c7f93829c655492b2e0846e3e860b754e1c746f3416085266c06e17311
+size 90835
diff --git a/MLQA.es.de/xtreme-test.parquet b/MLQA.es.de/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..a2444a7c6226f5376629dd9db7a3d54ed44a49a1
--- /dev/null
+++ b/MLQA.es.de/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f234e74060f1b2eaf1a516c4f445b791f00ba181657cb54b4206bcc018e93431
+size 850972
diff --git a/MLQA.es.de/xtreme-validation.parquet b/MLQA.es.de/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..d82527acc75c296a978b479b711ec6e420f10724
--- /dev/null
+++ b/MLQA.es.de/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:25b0612a4acf500cedfe93a1eda34a8f3d785f40f5e9a9c46c7bc18be57a7bd9
+size 99426
diff --git a/MLQA.es.en/xtreme-test.parquet b/MLQA.es.en/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..2fe02cf19429512b0a98c6388cc191005b96cb00
--- /dev/null
+++ b/MLQA.es.en/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:22012a5f62f0a8a1b4ed6dea466b935f4cf1ef37235279980ae490cfad3da20e
+size 2585604
diff --git a/MLQA.es.en/xtreme-validation.parquet b/MLQA.es.en/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..cd835860a2c562e337fd9dad1f4cb7c2274f495e
--- /dev/null
+++ b/MLQA.es.en/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:12fb81349c1e14d76e1c58efd51d512ee171af0280cde23720d8801a8db154e6
+size 257177
diff --git a/MLQA.es.es/xtreme-test.parquet b/MLQA.es.es/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..43489c0a5dc8e8128d5a886bca243717baa26f06
--- /dev/null
+++ b/MLQA.es.es/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bb7f0b9e3f9c39e56cacf51213fddcb70038feded1cf097bdca566a82e305a5c
+size 2598063
diff --git a/MLQA.es.es/xtreme-validation.parquet b/MLQA.es.es/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..be61cd4288b4a0c13db9c4abacafbd20c0916c8d
--- /dev/null
+++ b/MLQA.es.es/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cc81a1696c06df81c0585538ac627a982c37d3c071218e734f1e7506b67180ae
+size 258770
diff --git a/MLQA.es.hi/xtreme-test.parquet b/MLQA.es.hi/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..f7b60da4d5a09de5a72d9992d7f04add95467896
--- /dev/null
+++ b/MLQA.es.hi/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2d82507b4e01ff800d91b28353b8442145e002d3d6cd4db08c1ca58bd8f871fe
+size 844564
diff --git a/MLQA.es.hi/xtreme-validation.parquet b/MLQA.es.hi/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..470864c6ec65a8e6cc982eeeef858cd2cc9f23c0
--- /dev/null
+++ b/MLQA.es.hi/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a37f105837501eabfcae09cd270b1839fb8773cf1061fa2f39ed8d6d3eeb045
+size 109404
diff --git a/MLQA.es.vi/xtreme-test.parquet b/MLQA.es.vi/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..d4f4be82ee359fee642ea8607c614d5c2913a778
--- /dev/null
+++ b/MLQA.es.vi/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:af8b4728854232aabb070cf41ab44e83faa7d1017fa53356f93444b5d2619119
+size 1074029
diff --git a/MLQA.es.vi/xtreme-validation.parquet b/MLQA.es.vi/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..74374727c94ae3e92bebe834a1cfa6af95e2ba62
--- /dev/null
+++ b/MLQA.es.vi/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7a7f97439f4e5178ee81a0322f7d418132dfbaf01143bec28e514dabb1f54c65
+size 113858
diff --git a/MLQA.es.zh/xtreme-test.parquet b/MLQA.es.zh/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..d6c2709234aa4508d37e87d0046b5f96b1f62cbd
--- /dev/null
+++ b/MLQA.es.zh/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:76a40ee408f06026d6e7f171376963032d3e57b450b847c8d1d6351966549141
+size 1012050
diff --git a/MLQA.es.zh/xtreme-validation.parquet b/MLQA.es.zh/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..e1718ad4ae411d33cba9c3eb33540bbf02eb95ac
--- /dev/null
+++ b/MLQA.es.zh/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3203645fca2b5047b85c9ee3ffbca46b407965e27e5f8939dcb2251d0cc1b963
+size 88665
diff --git a/MLQA.hi.ar/xtreme-test.parquet b/MLQA.hi.ar/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..16af6a016194a1677f56025d99c9e956df2e30be
--- /dev/null
+++ b/MLQA.hi.ar/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2d76e5c8a5fc903ea670143809ce26f15e98646cf16f355fdbb7350df7dfb020
+size 1407967
diff --git a/MLQA.hi.ar/xtreme-validation.parquet b/MLQA.hi.ar/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..e9c82d5a024a529d97dcc22d7bca7e5ac919e69b
--- /dev/null
+++ b/MLQA.hi.ar/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f254e2e71563d9a0db4e1799f41b03034eec8df2bdc88480d166e8d936176df3
+size 134755
diff --git a/MLQA.hi.de/xtreme-test.parquet b/MLQA.hi.de/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..6ce660961ad0c99a5ba0ea021027f2b2bfb4e6f4
--- /dev/null
+++ b/MLQA.hi.de/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e71f427a30286004ad43bfbacb6c416812e8f828af29c5170b744dcefd03ad9e
+size 1142055
diff --git a/MLQA.hi.de/xtreme-validation.parquet b/MLQA.hi.de/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..655e6260a8cabbdcd01ce829642a89c33381a94d
--- /dev/null
+++ b/MLQA.hi.de/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9fe524921000d5f53d45f59e2b342b60f70902e5bc5a4648722014b286191001
+size 115743
diff --git a/MLQA.hi.en/xtreme-test.parquet b/MLQA.hi.en/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..e4a4029300e06c413a2e078c0231a8d20764be87
--- /dev/null
+++ b/MLQA.hi.en/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a891676c927caa051f3f3e3f94d4e3f09232c2c87136a1ba8dec9f9b13c5c57
+size 3764579
diff --git a/MLQA.hi.en/xtreme-validation.parquet b/MLQA.hi.en/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..1a7b59b43473e5650f7d9592953ad5866c6998a4
--- /dev/null
+++ b/MLQA.hi.en/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b9a337123dd1d9a1be41988c6c0d040c91b058fbfcbe64c0d72d9fcf253cc2a7
+size 366428
diff --git a/MLQA.hi.es/xtreme-test.parquet b/MLQA.hi.es/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..14f1b7bb163313f7d747ec1e6b8409963543a86e
--- /dev/null
+++ b/MLQA.hi.es/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f566c4bbb8d4f37d0f443ce9c564f15ad0340a084d0a002d62236d876289ec3
+size 1346681
diff --git a/MLQA.hi.es/xtreme-validation.parquet b/MLQA.hi.es/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..1a347d9d05c3acfb7be458db29f8ebc9851225c1
--- /dev/null
+++ b/MLQA.hi.es/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:81ee3a06ca064dafd7c60fdf16a516837cd6bbb0ce4ad21141df25d4aa7fe999
+size 146741
diff --git a/MLQA.hi.hi/xtreme-test.parquet b/MLQA.hi.hi/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..5bde94196afa86199cd4e65b37391d5876f864b3
--- /dev/null
+++ b/MLQA.hi.hi/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:25e53d68c4297e7662b6fd6c75da7fa00bd4cf05d4313dc3de14bf16bfd33925
+size 3859701
diff --git a/MLQA.hi.hi/xtreme-validation.parquet b/MLQA.hi.hi/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..a97a79690cb3d1e7f630943f7405d618f25caeb5
--- /dev/null
+++ b/MLQA.hi.hi/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e45393d9bd2cbf241b671fc0acfa012069da9b54554a8ea5ab121e4627706163
+size 376204
diff --git a/MLQA.hi.vi/xtreme-test.parquet b/MLQA.hi.vi/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..4ed45d543098850b8fdd922128329c0464352394
--- /dev/null
+++ b/MLQA.hi.vi/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:79c0c999eae179754d01d5d7803678a8a5467b16b6b645359c9f18877e85d020
+size 1552293
diff --git a/MLQA.hi.vi/xtreme-validation.parquet b/MLQA.hi.vi/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..3172eabdfce4a5c74934d63e3f68984a8f826a11
--- /dev/null
+++ b/MLQA.hi.vi/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e4371883b1202a0cb9a029814d6772c4376eead0d429bd2019bc39d4f5c1bb6a
+size 152625
diff --git a/MLQA.hi.zh/xtreme-test.parquet b/MLQA.hi.zh/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..b5e2cb81888802d1774fa58e84c6713e02e4ce0c
--- /dev/null
+++ b/MLQA.hi.zh/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0856713c3a862d2fac519b574e928584a507121aca0f9d47eca4469d3b1020ba
+size 1479497
diff --git a/MLQA.hi.zh/xtreme-validation.parquet b/MLQA.hi.zh/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..28508aaf57eadfbfcdc92e4e4654eb24999cd4ac
--- /dev/null
+++ b/MLQA.hi.zh/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:caab206c3918a32166ef085a19621c4db2025cf4b61d8f55da681cf4fca4ba4c
+size 147564
diff --git a/MLQA.vi.ar/xtreme-test.parquet b/MLQA.vi.ar/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..a09fc57c08520816b8e9b5a357230bd4ff03f8f6
--- /dev/null
+++ b/MLQA.vi.ar/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f9a8c5740559715b0d88a5ffa35d870aecff7f5dd171f92731dbe4a86dc77923
+size 1530595
diff --git a/MLQA.vi.ar/xtreme-validation.parquet b/MLQA.vi.ar/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..640be8fb13ab19e1c037f222d52b80cbcd2f5811
--- /dev/null
+++ b/MLQA.vi.ar/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:885d698cf023c754ad38011bba9c4d6b729f5537687882a37d621b9b13539eff
+size 126008
diff --git a/MLQA.vi.de/xtreme-test.parquet b/MLQA.vi.de/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..cb984aa69a77cc6e6f67d1d26ab4adddf55c2b2c
--- /dev/null
+++ b/MLQA.vi.de/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:52de7e6ef9071cc2cd2dfb31125adab831e11e5815f4ba79c72069a3b127503a
+size 1128296
diff --git a/MLQA.vi.de/xtreme-validation.parquet b/MLQA.vi.de/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..5820263d6bb4fbfabf757840fd2b09683893063e
--- /dev/null
+++ b/MLQA.vi.de/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8854663affcf4d726833b97f59b26a8f14cbfa9cf717c4360339e112ec727914
+size 139697
diff --git a/MLQA.vi.en/xtreme-test.parquet b/MLQA.vi.en/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..991a083a6d3201b6e69e1b10951d6858f74f9d33
--- /dev/null
+++ b/MLQA.vi.en/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b923c247ee33aaae1eaa146ad4ea597274e5489465fd45d131e5820170ce1415
+size 3724066
diff --git a/MLQA.vi.en/xtreme-validation.parquet b/MLQA.vi.en/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..4298799eec4165ad2ca613c139e18e5d2c096441
--- /dev/null
+++ b/MLQA.vi.en/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b049a5e360925cfc075597ada557710996fe4fe1ce8e608e9bcfe7a31f743b98
+size 347543
diff --git a/MLQA.vi.es/xtreme-test.parquet b/MLQA.vi.es/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..2e063dc29a68cd72d5f7d4a4b3d7275536e9f332
--- /dev/null
+++ b/MLQA.vi.es/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:73f406c27cfeaee9cde758f2213da04740580e03f4762b6353ae95861a3866f2
+size 1465977
diff --git a/MLQA.vi.es/xtreme-validation.parquet b/MLQA.vi.es/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..3f346331be3d4142a86155bb5f0b9b0904692e50
--- /dev/null
+++ b/MLQA.vi.es/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:95893a682f110578a77c712d187e25a0ea63fdbe6e404173acc8a23ca8bf47f7
+size 141889
diff --git a/MLQA.vi.hi/xtreme-test.parquet b/MLQA.vi.hi/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..d7ecbb9a41c44232b6aab3e3b9d95bd61394d7ee
--- /dev/null
+++ b/MLQA.vi.hi/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1c89d662dfeaff88848f010bdfa21eacfc3d5013884b95175811a0f1442e52d4
+size 1236388
diff --git a/MLQA.vi.hi/xtreme-validation.parquet b/MLQA.vi.hi/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..c484c716b7ce0ba210e5f7b5415ab42a43152ae0
--- /dev/null
+++ b/MLQA.vi.hi/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0001f16ab945bf7f736ff493348f96c84bda75053cfdd4bf0db7b428e8f2c523
+size 129619
diff --git a/MLQA.vi.vi/xtreme-test.parquet b/MLQA.vi.vi/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..5b1af5bb9b6a2871a5bfc7f1dc658eb10b8d3fff
--- /dev/null
+++ b/MLQA.vi.vi/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a1fe6de0d5984f96c3115c8f63c4358c8c8ebbcd7375c3ab24087a5eb0a090bb
+size 3754995
diff --git a/MLQA.vi.vi/xtreme-validation.parquet b/MLQA.vi.vi/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..9ad1065001f2db2f5c9394133dc677ee22e50d60
--- /dev/null
+++ b/MLQA.vi.vi/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:08c58bc7a314b1764135911e7906f8172913f66cfa8574164f7a4c887aa1e1bd
+size 350299
diff --git a/MLQA.vi.zh/xtreme-test.parquet b/MLQA.vi.zh/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..7c8ffb5482e0e84b6d3092033cc9f6d20db81671
--- /dev/null
+++ b/MLQA.vi.zh/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1b9bada7330dc43a49213a4f429259a22db3f88d4121f3919721bb733edf49a8
+size 1432327
diff --git a/MLQA.vi.zh/xtreme-validation.parquet b/MLQA.vi.zh/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..201662bf1b1ca50a6f6f617eae6786168c50b765
--- /dev/null
+++ b/MLQA.vi.zh/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bea972e15725c41745bb00ec1c660be987a49bb3a25b6ba334939961e09c1f97
+size 138018
diff --git a/MLQA.zh.ar/xtreme-test.parquet b/MLQA.zh.ar/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..a4b84a7322d21dd75430bfd83ca729c99f53adcc
--- /dev/null
+++ b/MLQA.zh.ar/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f098c9f6e63b26713be983f5d6aa04d94eb5c9000139ee59491d658fb9d32f69
+size 1117028
diff --git a/MLQA.zh.ar/xtreme-validation.parquet b/MLQA.zh.ar/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..d214df6f4a9e4350d5bde9c3824ad6edbddfe9c7
--- /dev/null
+++ b/MLQA.zh.ar/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d33006f7c0228e4e5c93fee491fe16d331ff2976146576061ce4581703aec43a
+size 106785
diff --git a/MLQA.zh.de/xtreme-test.parquet b/MLQA.zh.de/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..e3f25c1999ee81ba1550fe864aa041037452b6b7
--- /dev/null
+++ b/MLQA.zh.de/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:666e0f40216f962f0e11c55fe5856b49248560a59eb327498eb9c45017dbd71d
+size 892824
diff --git a/MLQA.zh.de/xtreme-validation.parquet b/MLQA.zh.de/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..239779f7748f210ea01a9c884bc7f5f8ccb17058
--- /dev/null
+++ b/MLQA.zh.de/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e62439b907a9bf153cf66abb384bbcc49f04630fe9f8d3cf750c8b8bd4abd6b0
+size 113955
diff --git a/MLQA.zh.en/xtreme-test.parquet b/MLQA.zh.en/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..03dc03c62d3e0f0777887319b2809462a823be2e
--- /dev/null
+++ b/MLQA.zh.en/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4b96168bdc82c4b2c97c29d240def7ebdc026baa2d021f68f8909c6b94355a18
+size 2830195
diff --git a/MLQA.zh.en/xtreme-validation.parquet b/MLQA.zh.en/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..253264bba1db9473bed3d2fb1ddce7ec262cc4d8
--- /dev/null
+++ b/MLQA.zh.en/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f78104715fdce02e8ce3467450940f27acf1583ebb8d5570b0d29a0a104b2794
+size 278140
diff --git a/MLQA.zh.es/xtreme-test.parquet b/MLQA.zh.es/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..bb8db75d0680c3bb55ea49e7587f582490303648
--- /dev/null
+++ b/MLQA.zh.es/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2980a621caa0bcc766bf767a9c93453fe4bf0106abf6a5832f2b8277c666c29d
+size 1124001
diff --git a/MLQA.zh.es/xtreme-validation.parquet b/MLQA.zh.es/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..227a66f739f5f09f0e11be22ef72e0d183778a64
--- /dev/null
+++ b/MLQA.zh.es/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4db8b16b903e2e1d3f3cfc92152610e1c9172130799ff83c581fd88097cf0f4b
+size 99416
diff --git a/MLQA.zh.hi/xtreme-test.parquet b/MLQA.zh.hi/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..29090d439590ac1757cebee1ee4b8bed7381530c
--- /dev/null
+++ b/MLQA.zh.hi/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b0a5f7e828c38de14c03a45069a1354f6b0e36ffe9e05fe7a54f6b6d1e4637a0
+size 933785
diff --git a/MLQA.zh.hi/xtreme-validation.parquet b/MLQA.zh.hi/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..5b39599ecfdbec887d4516237d81c59042271dea
--- /dev/null
+++ b/MLQA.zh.hi/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d3d7a0d0b6832789125dea35e9eb4e66672dc55a0e7027e7544d908cc9eaa5a5
+size 110764
diff --git a/MLQA.zh.vi/xtreme-test.parquet b/MLQA.zh.vi/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..0b800878bb65215739116120742fa2e7efe42664
--- /dev/null
+++ b/MLQA.zh.vi/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0e9622cda7f28474d6a9e539a2a394744ce55dfb0d5d8a46c9bb02b9de221e85
+size 1147708
diff --git a/MLQA.zh.vi/xtreme-validation.parquet b/MLQA.zh.vi/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..12e0c410b54fd090d849e44bef588bb92c96a917
--- /dev/null
+++ b/MLQA.zh.vi/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a88be0c66722663375eeb50294336026dc716d9cbbd674d2d5efad350a60c06f
+size 120455
diff --git a/MLQA.zh.zh/xtreme-test.parquet b/MLQA.zh.zh/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..3fa8538b2c07a15e1f2e5bd3b32027ca6e6d672e
--- /dev/null
+++ b/MLQA.zh.zh/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fb842074822d5fc4e88f0f868b370f7ebe87b62b070dbcaac2a35cb20972c315
+size 2827503
diff --git a/MLQA.zh.zh/xtreme-validation.parquet b/MLQA.zh.zh/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..f64709c8088b7f97bde451d49ac4390ceaa46b61
--- /dev/null
+++ b/MLQA.zh.zh/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ffcde2557aad059c028e167ee9ad868e93fa41b2b4f3ed7740b3ab83efaaf3be
+size 277761
diff --git a/PAN-X.af/xtreme-test.parquet b/PAN-X.af/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..639be0c8bcc6383863eb72d1785f7040e874090b
--- /dev/null
+++ b/PAN-X.af/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5fd377b5158114021484d5d20f89a5694ae1721399461a27468302ec38e33ff7
+size 55730
diff --git a/PAN-X.af/xtreme-train.parquet b/PAN-X.af/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..777e5046a84fd205093f0fca379f3aea70a3fc9b
--- /dev/null
+++ b/PAN-X.af/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:66d35d2bd9a05d394306b4f7418d425f654362fdc45a48c99a5abc783aa4bb02
+size 277444
diff --git a/PAN-X.af/xtreme-validation.parquet b/PAN-X.af/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..91918c50fca19468a726adc0c1dabe215fe30b14
--- /dev/null
+++ b/PAN-X.af/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5242f907575c95f7763a6ed78ebf1a77a51c5fc6b4b88fa6709625e972b2c185
+size 55685
diff --git a/PAN-X.ar/xtreme-test.parquet b/PAN-X.ar/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..51dd0e25a96bb3d0d3af23519a479b0c5f38952f
--- /dev/null
+++ b/PAN-X.ar/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:236786b23af2e42249e60dd4b5dcf80595e839193c5ef0b9ea9cf3409b463b31
+size 392157
diff --git a/PAN-X.ar/xtreme-train.parquet b/PAN-X.ar/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..a194456922319c5128bc09ddbc398d8848916f4c
--- /dev/null
+++ b/PAN-X.ar/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f73d6651257ae7a522db1f92af400c7314d4f27db4cf93c5806ca5aa8c913c46
+size 784505
diff --git a/PAN-X.ar/xtreme-validation.parquet b/PAN-X.ar/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..6db731b40a2d4ad4d56bca9d1199dffa6376a601
--- /dev/null
+++ b/PAN-X.ar/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:13651c02d643ce682f3aa8bc3af5acfc296c0477260389c88db7956d0fd43282
+size 390058
diff --git a/PAN-X.bg/xtreme-test.parquet b/PAN-X.bg/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..245d4e3a03aa1d22d11973155936915234f33bae
--- /dev/null
+++ b/PAN-X.bg/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b3bb6b5ec927e8fe32323930c975589d1d6e09ab3a4185a3f514fde85e2776aa
+size 509807
diff --git a/PAN-X.bg/xtreme-train.parquet b/PAN-X.bg/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..e0e1ad4851f1c5a5bd9881e99da89e5e10ab24a6
--- /dev/null
+++ b/PAN-X.bg/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:be730a00426ce90f7ad620aefbff515df66cffc0cca97e09ed58cba756064b57
+size 1010917
diff --git a/PAN-X.bg/xtreme-validation.parquet b/PAN-X.bg/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..21c46903c2079d8cccf0f514931b2372cf8b9eda
--- /dev/null
+++ b/PAN-X.bg/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:434eb44e6329bdb0ca6eb734155dbe2f4409e878f9dbad061d8056a4678261fb
+size 509195
diff --git a/PAN-X.bn/xtreme-test.parquet b/PAN-X.bn/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..aafafcd9c56b491d33e2a204e816a205800c2a7e
--- /dev/null
+++ b/PAN-X.bn/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ca7d0ac838f228c65dc388c6a4886042b3c86bd922720fed77b232162a62c950
+size 32079
diff --git a/PAN-X.bn/xtreme-train.parquet b/PAN-X.bn/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..8673a68ab3b48626743b7d58d369524154576bfc
--- /dev/null
+++ b/PAN-X.bn/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d2553b75df388a6b513c5d979091556d27d6abd8941eb8211024a196bd8a99c5
+size 300810
diff --git a/PAN-X.bn/xtreme-validation.parquet b/PAN-X.bn/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..c4169db9e9552747924360e5bf3c31a497d17471
--- /dev/null
+++ b/PAN-X.bn/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f0958df135e141b46726c8ba82a4c92a7bb07aa365c0701b361431a5900b4e48
+size 30889
diff --git a/PAN-X.de/xtreme-test.parquet b/PAN-X.de/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..ddce33c3d2faf790a57fe1d7c66e35c1d2b99f80
--- /dev/null
+++ b/PAN-X.de/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:800d4d3c1177f6f47aeda6961dd3561ee02cf23f0d503265cbbc25c25283b5e3
+size 587403
diff --git a/PAN-X.de/xtreme-train.parquet b/PAN-X.de/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..0a8ac130f95f2f387d42d408d4fbb0719334fe33
--- /dev/null
+++ b/PAN-X.de/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:89e68143bca6dc1bbb5c7a35a8a27caa3f46b9a42c25bef2db676753134296ef
+size 1181996
diff --git a/PAN-X.de/xtreme-validation.parquet b/PAN-X.de/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..62befccf3d5dfe4260f5baff94075917be4c2dff
--- /dev/null
+++ b/PAN-X.de/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4fd48ccee838fc37920bbce6d4a94424b606f834ff95b9ee5a8d3725877591bf
+size 590093
diff --git a/PAN-X.el/xtreme-test.parquet b/PAN-X.el/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..9283767a8583f94bd70a0ba29e6b7e2462076b01
--- /dev/null
+++ b/PAN-X.el/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:18af2cec9517e3d9f4340fde7461ff68656751e5999704670261f734c3fc5ae5
+size 570430
diff --git a/PAN-X.el/xtreme-train.parquet b/PAN-X.el/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..4ade6a1919c50d330c26e34235c759e82bf135c5
--- /dev/null
+++ b/PAN-X.el/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e188e48114eea8cea46e166d52f878021907e8536db4469a1aae263beafb1248
+size 1131914
diff --git a/PAN-X.el/xtreme-validation.parquet b/PAN-X.el/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..4a101feb329c4fbcbfab10138eb82f1ff39d1739
--- /dev/null
+++ b/PAN-X.el/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05418769d04e4b58e3911dc5979b64a9a547a18bbf8d07bd93b5c979b9ef43b3
+size 568632
diff --git a/PAN-X.en/xtreme-test.parquet b/PAN-X.en/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..e6f1684e9896c858784c5213030abd6e1bca33b3
--- /dev/null
+++ b/PAN-X.en/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8bcb38f87a2d5677ff326827d16b0ac13880d0d798f98f5be1e45860326aca44
+size 472114
diff --git a/PAN-X.en/xtreme-train.parquet b/PAN-X.en/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..8b7a1928f09cacd71be55e444bb1859c18574e76
--- /dev/null
+++ b/PAN-X.en/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4d690d08eff1a4ad2cdc69e37474b0aa0c231f476376dc53b3109fc4d9390232
+size 941856
diff --git a/PAN-X.en/xtreme-validation.parquet b/PAN-X.en/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..26fad8b0610918e715cadecc404db765f81bd1d7
--- /dev/null
+++ b/PAN-X.en/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:73141b934c7cef56c14a58402159850c4a916980afa43e188e58811302f571d7
+size 471564
diff --git a/PAN-X.es/xtreme-test.parquet b/PAN-X.es/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..558a6c389c25f65f39963bff1812b85036c03111
--- /dev/null
+++ b/PAN-X.es/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:12a66b9ea3e6f3e77f2f2731e57712c0aee3464ab094a109fd6a99fc7ac50903
+size 373196
diff --git a/PAN-X.es/xtreme-train.parquet b/PAN-X.es/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..71a694a6f16e11e762825731e5925d440d7c0e9c
--- /dev/null
+++ b/PAN-X.es/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d3f4d03563460b3b3e671e17b42d6d87ec0cc0d22b950db51a734f20f625f2f5
+size 743953
diff --git a/PAN-X.es/xtreme-validation.parquet b/PAN-X.es/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..33864ad6433872bd6bdc80eddb3d4d1d62fb8d3b
--- /dev/null
+++ b/PAN-X.es/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2c49f8010892fac79f2a5b3bd90b0385e7189fe7de8207530d781a3fb4cc9f5a
+size 371663
diff --git a/PAN-X.et/xtreme-test.parquet b/PAN-X.et/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..47494f46a1b1ec97c5eed6085d07cc3af959d35b
--- /dev/null
+++ b/PAN-X.et/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ad6b747f4891dccb5b99e312f31a55e56fbede93c3108e5b3a53519a2d305bed
+size 544828
diff --git a/PAN-X.et/xtreme-train.parquet b/PAN-X.et/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..836fe2d68de48c194ee8ee2a7f4d27a40279c7df
--- /dev/null
+++ b/PAN-X.et/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e2186dc58e8069a68851421ee71c57035a454b1fe23c4a67084456599348ab20
+size 818360
diff --git a/PAN-X.et/xtreme-validation.parquet b/PAN-X.et/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..4c3400ca814410b21d15473685838f2b3bed2c8d
--- /dev/null
+++ b/PAN-X.et/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2b3536136d01ac00a4e202ca79ce57edfae55cf77038bfc18cd36c425e277c3f
+size 551776
diff --git a/PAN-X.eu/xtreme-test.parquet b/PAN-X.eu/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..b0750bf3e1d6cf32eb0951bf4d8b8358a1097663
--- /dev/null
+++ b/PAN-X.eu/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2a6f9838f7538ec8c6e353586dd16296fc04755401f5fedba9a7acc0e4e56397
+size 460436
diff --git a/PAN-X.eu/xtreme-train.parquet b/PAN-X.eu/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..0df53115d1b1673720b547157a37d155a0da778f
--- /dev/null
+++ b/PAN-X.eu/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a86a9240a86f4b9d4c1e54b2f9047587c758dd8782a965f42521ca74b8b3fb1f
+size 466382
diff --git a/PAN-X.eu/xtreme-validation.parquet b/PAN-X.eu/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..1ab272760169257804319ae3a94ffe6f2c06e757
--- /dev/null
+++ b/PAN-X.eu/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:74430f7336b679eaa6fdba0ef4f9f2154244024b76636896154394ae8993d8e7
+size 465791
diff --git a/PAN-X.fa/xtreme-test.parquet b/PAN-X.fa/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..64dfc87a9c80b4ed851575665ee11cba3b3142f6
--- /dev/null
+++ b/PAN-X.fa/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:21e7b5b784d457e2aff51162e58ae25ca98a799bd469e6b1d1e7410f7e52065f
+size 350804
diff --git a/PAN-X.fa/xtreme-train.parquet b/PAN-X.fa/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..4c99008d1eff8841a184c6628a6a52c8ef365832
--- /dev/null
+++ b/PAN-X.fa/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7eff7cf0a3290249bc8c4d5ceebb459ef44a70da17ecc4b2e9130b809389539d
+size 694478
diff --git a/PAN-X.fa/xtreme-validation.parquet b/PAN-X.fa/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..c5174e9b36cfd27c22017a46a20cbc13c5be4dee
--- /dev/null
+++ b/PAN-X.fa/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ac4fb2bf1746bc35c5c9c5987ba96adb694b8f345df94259ae2d55ba8dce5c5
+size 355176
diff --git a/PAN-X.fi/xtreme-test.parquet b/PAN-X.fi/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..e112c023461aceb98bffd58e4e8ca0f4e2b86e86
--- /dev/null
+++ b/PAN-X.fi/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a719cb78e12d9d103883b6dd4e22da82f0c39553f16f0228d945bf176e2d4e70
+size 613591
diff --git a/PAN-X.fi/xtreme-train.parquet b/PAN-X.fi/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..2e51741b2f25da364fcca0cf8ef4b5cd0a050a23
--- /dev/null
+++ b/PAN-X.fi/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0969e0fa9da5eca1e5279d41d679d033f5368536109c7741aff2ed8fde010056
+size 1229563
diff --git a/PAN-X.fi/xtreme-validation.parquet b/PAN-X.fi/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..2603297a4f5a8da40b2275b74943854efe704a42
--- /dev/null
+++ b/PAN-X.fi/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cba40fc16026b585fac2172db9cb27ca65383b22c285f5714d5fa9b3011eea2a
+size 615245
diff --git a/PAN-X.fr/xtreme-test.parquet b/PAN-X.fr/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..960d89c66ae46b7c76114104ed8808ea0b208071
--- /dev/null
+++ b/PAN-X.fr/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e02f709ee064eced8cf38c193095deafa638499879f65051efebcfe162c1be7d
+size 422416
diff --git a/PAN-X.fr/xtreme-train.parquet b/PAN-X.fr/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..b73b16b76893c557cf6966ec5177423dabfabd85
--- /dev/null
+++ b/PAN-X.fr/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2090c91c9f036ab8b191d47214efdb86db56ef3230233308f81f48f3ee1f88cb
+size 837067
diff --git a/PAN-X.fr/xtreme-validation.parquet b/PAN-X.fr/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..8fad18fe959f7657ac2191e3842b8c8ec2f591b1
--- /dev/null
+++ b/PAN-X.fr/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6e6a27d4784474f38daea062e9a12b6cdfab228f7a0a945572af5aba04fdc5e3
+size 419050
diff --git a/PAN-X.he/xtreme-test.parquet b/PAN-X.he/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..f2ada2ec2662975c269042e261bad5d2cc2bc270
--- /dev/null
+++ b/PAN-X.he/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c40513e2414b08c7bb60ee666c4ef742e44c3bd3ee0b3ca18737a89e3f8651e0
+size 543588
diff --git a/PAN-X.he/xtreme-train.parquet b/PAN-X.he/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..6f700607c3e519e3e4ee6da22c0f84d9301b6e51
--- /dev/null
+++ b/PAN-X.he/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a338b19507889346dea4489476b2e39ee27f13acb49a30861a36b6385be04f2a
+size 1091744
diff --git a/PAN-X.he/xtreme-validation.parquet b/PAN-X.he/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..77223e949f171d0ef94acad01eb0b8cc08199724
--- /dev/null
+++ b/PAN-X.he/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:704379c62c72e7cd972b0f499f116d0f56495085382644daed33e67f83fc667b
+size 550381
diff --git a/PAN-X.hi/xtreme-test.parquet b/PAN-X.hi/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..2d11facd62ad385d530c337a7bfcee99f702ed12
--- /dev/null
+++ b/PAN-X.hi/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:23ccdb08feb1b0fb416a30a2ab1c27968bb4f27ad41566434f64bca8ea36960a
+size 39627
diff --git a/PAN-X.hi/xtreme-train.parquet b/PAN-X.hi/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..fd957d3c4b1929739f8f3b793a2cc79a5449b054
--- /dev/null
+++ b/PAN-X.hi/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8db3ec2a7cf223127aa27b78443f809d6424730b251890a0402359335ac4f96c
+size 187446
diff --git a/PAN-X.hi/xtreme-validation.parquet b/PAN-X.hi/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..acfa69892bb4317c29a60e201e9d5d9cf7210d92
--- /dev/null
+++ b/PAN-X.hi/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7d0bd14d9176f646be0a76a1f707725d6fbb98da8e92af6fe1cfcf1527ff989a
+size 38857
diff --git a/PAN-X.hu/xtreme-test.parquet b/PAN-X.hu/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..b63629649231b7794eba4e992a3f26d81f489411
--- /dev/null
+++ b/PAN-X.hu/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e0ebcbb7a17c519412b130d2395235aea6874afd4e7f3872b4301930db99ea42
+size 603586
diff --git a/PAN-X.hu/xtreme-train.parquet b/PAN-X.hu/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..f51fe6996ae44de3326e9051d69415ae0d0ca3f0
--- /dev/null
+++ b/PAN-X.hu/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ea4db554028d8b5dccc653c6ac99591521eac86edd3b3ec72ed97f3c0a60a929
+size 1208218
diff --git a/PAN-X.hu/xtreme-validation.parquet b/PAN-X.hu/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..4747ee784ad853be452ecf44758d25054e58b16c
--- /dev/null
+++ b/PAN-X.hu/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80691316f067fb54b8b8c336c89ddbda26daa40e5d646a0cee23177d2b8bfa60
+size 586836
diff --git a/PAN-X.id/xtreme-test.parquet b/PAN-X.id/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..90ed6e8b4ef08703ef5020460a333d8edc9997b8
--- /dev/null
+++ b/PAN-X.id/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6f269d7f274f1b064126b06f22d155b6fb4ea7e925893503d8a94664d0a32bec
+size 350798
diff --git a/PAN-X.id/xtreme-train.parquet b/PAN-X.id/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..6fc52175d19ce1a786fd75896457a142655319f7
--- /dev/null
+++ b/PAN-X.id/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6801b9c1def539f0ca088189ba8a2bc7dfa0b22b8ddc3fb8b2e4f6a3b2760b20
+size 707547
diff --git a/PAN-X.id/xtreme-validation.parquet b/PAN-X.id/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..0852030d5d33fa7428fc675a3f5d109d49e2fdd8
--- /dev/null
+++ b/PAN-X.id/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:09fe9e8abcf26e1d6047777412f6287bd4532920994351a73bd24b261f33bcca
+size 352954
diff --git a/PAN-X.it/xtreme-test.parquet b/PAN-X.it/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..18fe61efe1de9016e6346cff5e6a64b53aca134a
--- /dev/null
+++ b/PAN-X.it/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:59dd7f8ed7e3c7fe03cae4f59c7b93ba9da13efd4034c7fb190c1e34bbd3cc86
+size 463735
diff --git a/PAN-X.it/xtreme-train.parquet b/PAN-X.it/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..99afdb470d8c8b6a27814b0411396136c4c2f053
--- /dev/null
+++ b/PAN-X.it/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7d3d804906a296643c81959914b803dada873c766a1223ab1af58a2e58fe2e66
+size 932120
diff --git a/PAN-X.it/xtreme-validation.parquet b/PAN-X.it/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..addfef627a870dba069079e8b6c48a7dc17bc0cb
--- /dev/null
+++ b/PAN-X.it/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d35a0892b4c50c384750e3f2d07804c939ebf884ddc5c9c76ca534ae95a4bfd6
+size 459193
diff --git a/PAN-X.ja/xtreme-test.parquet b/PAN-X.ja/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..e26854ef9f285183a83408f90ef94f58f70f0c49
--- /dev/null
+++ b/PAN-X.ja/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f07efa9529b80141998ff7b5804841f14c96da88700cb45f2547520060889a3
+size 623415
diff --git a/PAN-X.ja/xtreme-train.parquet b/PAN-X.ja/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..5f9160f239f9b50203dee797c426627cf3464ffe
--- /dev/null
+++ b/PAN-X.ja/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a24671885627a0f2ecd4fc8bdb3ed63672b736a837f7299b611549581ecdef09
+size 1227283
diff --git a/PAN-X.ja/xtreme-validation.parquet b/PAN-X.ja/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..f564024451a7dbb2c4698dd31fb4ef0aaf7df9a6
--- /dev/null
+++ b/PAN-X.ja/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91c96d2405f3ebc5c14d0693bcaa99eb90960f6a4f1878797c67961d2e21489d
+size 614226
diff --git a/PAN-X.jv/xtreme-test.parquet b/PAN-X.jv/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..89e02d72eded80d8ce670660e0b0b5ac5ae40cac
--- /dev/null
+++ b/PAN-X.jv/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:69af6f2f7776f87376992f4fd29565e1f33f97b392c1be0c55c947a4a76c98b3
+size 7010
diff --git a/PAN-X.jv/xtreme-train.parquet b/PAN-X.jv/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..3345ef700c24f6aba1956d2bfcd0b75e4b4eb4cd
--- /dev/null
+++ b/PAN-X.jv/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0ab7d9a0344cfa715216d2a2f76f32a29fb9dab1e7303cfddcfe453a6dd8459
+size 6944
diff --git a/PAN-X.jv/xtreme-validation.parquet b/PAN-X.jv/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..2fc54d271258e11d0e6d7c9ac9d8364a9bfc9ae1
--- /dev/null
+++ b/PAN-X.jv/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b48aaadb374d7d499b58f9cce0d988dfd688a2560a03f51e536ca92a2150e120
+size 6437
diff --git a/PAN-X.ka/xtreme-test.parquet b/PAN-X.ka/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..fbecd88918dc7cd3fec8e9e4d4e4a04c7c0bce61
--- /dev/null
+++ b/PAN-X.ka/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6557986416338519ee281aeb73526b33b5dc583029e989e4fab8493757b6c594
+size 608237
diff --git a/PAN-X.ka/xtreme-train.parquet b/PAN-X.ka/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..c3edd9f3b8d4a9b39795314af513308958a1d0c8
--- /dev/null
+++ b/PAN-X.ka/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a8efe599a4d632add4bc41e32ff63f77df978931e6e1ecef6e53266b7ce33d44
+size 601934
diff --git a/PAN-X.ka/xtreme-validation.parquet b/PAN-X.ka/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..80dd4c6277e83873f78bd9fe13c3f9357dfda32d
--- /dev/null
+++ b/PAN-X.ka/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2826ebb963102534bb3b0b74b2406d62eaaf3132cf129401d3da8692c116728f
+size 606539
diff --git a/PAN-X.kk/xtreme-test.parquet b/PAN-X.kk/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..74f6d3097dae836da6925dd39d3aefe2b71f5dd7
--- /dev/null
+++ b/PAN-X.kk/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7c6e56543b9482ec611ad572158ed6f32429bb7f97d0239e952b1cf8f0986a42
+size 52415
diff --git a/PAN-X.kk/xtreme-train.parquet b/PAN-X.kk/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..8c0268185d266abdef97ab2ac1d3dcf6f6441879
--- /dev/null
+++ b/PAN-X.kk/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:38136593e1e4447feb46620d440e2e88d448c8f924eebf03a432e4c0ac434d9b
+size 53182
diff --git a/PAN-X.kk/xtreme-validation.parquet b/PAN-X.kk/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..94c85b94e66eec26678c96d1f71c05b12c1c7e81
--- /dev/null
+++ b/PAN-X.kk/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:192ca0c03af4ff4cf0d43bf1ce1d0ed519a83a38e4a5496df95f32784a0a1be2
+size 54873
diff --git a/PAN-X.ko/xtreme-test.parquet b/PAN-X.ko/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..e1d7fb2e712d6fb81e5d3ec45febedcc653e7655
--- /dev/null
+++ b/PAN-X.ko/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d42b77cb34f1b494f745af332dcc4226f109dfccf8c7c93f9166edbcf23f76fa
+size 636346
diff --git a/PAN-X.ko/xtreme-train.parquet b/PAN-X.ko/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..b18eed500f630317d1197a7b4f1961dcfecfefd4
--- /dev/null
+++ b/PAN-X.ko/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ee124317e80bb166630baa1e014d891a715d43530d73ece2436fb53c85e4552b
+size 1270039
diff --git a/PAN-X.ko/xtreme-validation.parquet b/PAN-X.ko/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..524ea8a117efcf59ae74a5a921ee75d259d94346
--- /dev/null
+++ b/PAN-X.ko/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:43bbec82c26e51ae19bc73bb7286f37b8921d4ce974818ff3002e152064ea41a
+size 632456
diff --git a/PAN-X.ml/xtreme-test.parquet b/PAN-X.ml/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..6160e67de2bf08cccec9006087dce2b736e9df94
--- /dev/null
+++ b/PAN-X.ml/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8df65e4e2cdcf201f8c278f7a8343ba2e0eeba0f99a14f6a5a4bf5055ff2ddf4
+size 69916
diff --git a/PAN-X.ml/xtreme-train.parquet b/PAN-X.ml/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..295b096edf32d102cfa5b5dcac1c7865129c67c3
--- /dev/null
+++ b/PAN-X.ml/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:df3aecd27cd76b507ef076b53ca6376ab8284904993d48233ba5f26a29cfb8ec
+size 708402
diff --git a/PAN-X.ml/xtreme-validation.parquet b/PAN-X.ml/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..19c2864c4130e73e0e0db14fc1e50d7fb5cd8e6a
--- /dev/null
+++ b/PAN-X.ml/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0b281eeaae198b4e3684cb6fd50d0efc70ae7e91dbeb672074bac249eed89a93
+size 74391
diff --git a/PAN-X.mr/xtreme-test.parquet b/PAN-X.mr/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..5482a56e254453d1d0e153c53720f62a0173afc1
--- /dev/null
+++ b/PAN-X.mr/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f9b02a2a0da4c4a2c976e54d0e117b76bad15f5163f2110e493b4f23aa76cb5d
+size 50625
diff --git a/PAN-X.mr/xtreme-train.parquet b/PAN-X.mr/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..a8e0ba49cd4a611e6799a803fb49f0271abd23f2
--- /dev/null
+++ b/PAN-X.mr/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:15104f6d152a576148f1167e5d21143cb38933d52ca60cc6e5d51b3f26b8985f
+size 246939
diff --git a/PAN-X.mr/xtreme-validation.parquet b/PAN-X.mr/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..41a77cb96012403cb2a354a671530a882a009938
--- /dev/null
+++ b/PAN-X.mr/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a4198a28345b3de2b80bf8331305de38f4192dab4987d29a8cfcc02a39ba9e8f
+size 49495
diff --git a/PAN-X.ms/xtreme-test.parquet b/PAN-X.ms/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..00da0a05c706145e9f16abe2359e7772cd68567f
--- /dev/null
+++ b/PAN-X.ms/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:240839070f02565af28693e728ae174f9decf81950b083bb37fb8912a71c3f77
+size 33898
diff --git a/PAN-X.ms/xtreme-train.parquet b/PAN-X.ms/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..2a0b9ef0c1838bef110a27222df43774fa5df541
--- /dev/null
+++ b/PAN-X.ms/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a002a8e5b8900da2a8c91ebe7bed268636e3ab9843653267f92459c8dc9a680f
+size 640803
diff --git a/PAN-X.ms/xtreme-validation.parquet b/PAN-X.ms/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..31376aa537105fb978f9f749680cef77a27b9c61
--- /dev/null
+++ b/PAN-X.ms/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9b3278c86151262f7c61ef05df24f0cbaf8b256ac4fb0d8dc51a8c4bd7c73334
+size 33668
diff --git a/PAN-X.my/xtreme-test.parquet b/PAN-X.my/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..24c57bf24e1df0f0fa0bb218a295533bfadac01a
--- /dev/null
+++ b/PAN-X.my/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b0318c8d0214c8a2421f14201b2ce405ba97a0df1d17cb7bc1f671094a3b8c6a
+size 13434
diff --git a/PAN-X.my/xtreme-train.parquet b/PAN-X.my/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..5a5d0885a0c295e7f4765c15b4012684794036ce
--- /dev/null
+++ b/PAN-X.my/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f08f5dbb687ceb248f62d1e6159a6e3f1a51364bac5c95e91d103aeb7c322438
+size 11406
diff --git a/PAN-X.my/xtreme-validation.parquet b/PAN-X.my/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..94d237566c1fb82d2bc37ffdae826447c65bf3da
--- /dev/null
+++ b/PAN-X.my/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f110561a12f2b14bed312aa034f99af523e4962a8d885fbf9e7f4ea0ac3b5d17
+size 14084
diff --git a/PAN-X.nl/xtreme-test.parquet b/PAN-X.nl/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..6296b7728e415b1ca7e7815faf035f96f4c94f22
--- /dev/null
+++ b/PAN-X.nl/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:73a23d85759944abdf13451274dfe1e0d08cc98ad217d0eb5fc45792e3b4663e
+size 485859
diff --git a/PAN-X.nl/xtreme-train.parquet b/PAN-X.nl/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..7789c2ca771c45a843e77e9c3516a0af0bea672a
--- /dev/null
+++ b/PAN-X.nl/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d4ccc41e755207fa3bd67305d23c96cee978769edbfd011640005a7068870bc2
+size 974922
diff --git a/PAN-X.nl/xtreme-validation.parquet b/PAN-X.nl/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..571179e578989f8a66ce357cfbf19c28f6bc7413
--- /dev/null
+++ b/PAN-X.nl/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1d8c973385ab1fceb9996e8870ead8501fee2379441cba1257423c5fbda4b2cd
+size 482362
diff --git a/PAN-X.pt/xtreme-test.parquet b/PAN-X.pt/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..162211d1ce6fd0c8f9186788c529d45f06da9d60
--- /dev/null
+++ b/PAN-X.pt/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:769c0cda0f66b40ec1bc9779582f11bb7f52d7ca23b81e42c12d530dbb6ebc65
+size 382498
diff --git a/PAN-X.pt/xtreme-train.parquet b/PAN-X.pt/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..3cf1edca9ba9b762ff006a48d86357835afc30e0
--- /dev/null
+++ b/PAN-X.pt/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:822e12d0543ba1b09d0e5ed19198feb888399bf6aeb16dd7c59e4341e94d9ba4
+size 769449
diff --git a/PAN-X.pt/xtreme-validation.parquet b/PAN-X.pt/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..8b7d947e3613d5b37d5f9fc41e67c57a73ee2282
--- /dev/null
+++ b/PAN-X.pt/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:50bf8ff98318618765f0472d4a5b50147ea78a1b7b8b85ff086361c9a2a7f796
+size 387781
diff --git a/PAN-X.ru/xtreme-test.parquet b/PAN-X.ru/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..90f30ec2186ff5b8efd0bacabb7a2fa306496f3d
--- /dev/null
+++ b/PAN-X.ru/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fc2fdadeb0b004ad4a5ee3c84fda79af0cf945268789df2dc73c2f52737c56c0
+size 535642
diff --git a/PAN-X.ru/xtreme-train.parquet b/PAN-X.ru/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..e763b6f41f6cac4e790d2be74cb08429906e9474
--- /dev/null
+++ b/PAN-X.ru/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1382f3238487cdae685f7eb7525d35f7fae63caa603ceb2376741a6e312ddaf6
+size 1061482
diff --git a/PAN-X.ru/xtreme-validation.parquet b/PAN-X.ru/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..0dee50954b0ee026d20dfe646c53934e7a41a338
--- /dev/null
+++ b/PAN-X.ru/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:89a0d4209e5bdd6a11e6ecb52d18acca28253b6142b8fd67637270580940185c
+size 529856
diff --git a/PAN-X.sw/xtreme-test.parquet b/PAN-X.sw/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..d6cb4795ca22f990240f7488fcdaf26c698f198d
--- /dev/null
+++ b/PAN-X.sw/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bb9636fb91a3a607c0f8e9d17e64325459fb2230d37d48cb11b6b60b17b08c5c
+size 29489
diff --git a/PAN-X.sw/xtreme-train.parquet b/PAN-X.sw/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..e2c516f8033bd7fcfa38b1ff2b6476ffa22be305
--- /dev/null
+++ b/PAN-X.sw/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f8b0de6b0d18f514b42acc10a4cd9b5f2994fb39190235512dec12b852eec414
+size 28890
diff --git a/PAN-X.sw/xtreme-validation.parquet b/PAN-X.sw/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..b64c74d96d16b5ce65d01236608f29b271601251
--- /dev/null
+++ b/PAN-X.sw/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a80499a911f5dcbf0609e37744bb26d064a23702e8199db15de9aae425ac6cb6
+size 28972
diff --git a/PAN-X.ta/xtreme-test.parquet b/PAN-X.ta/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..175c1fa73d0469b6ae513ef3727af000196513a6
--- /dev/null
+++ b/PAN-X.ta/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8329eaf65c280fce6c295c1b3bc4d86d761d04c5a6de7ee5be031c6a7312df92
+size 62663
diff --git a/PAN-X.ta/xtreme-train.parquet b/PAN-X.ta/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..1750e891813934e68cd59600e64d99f478cacc4d
--- /dev/null
+++ b/PAN-X.ta/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:df38ca2b5b554067ffbfe1e5ce78fef9e248ea819f06aede880e9f60d2e91af1
+size 918467
diff --git a/PAN-X.ta/xtreme-validation.parquet b/PAN-X.ta/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..7a0f68e01e86eafd8875dcd7d0f1783198237119
--- /dev/null
+++ b/PAN-X.ta/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1d8a0a9053a598eafd3f1e58cf4bf9d1c5cdbbfb49012c6dceecf5172f760780
+size 63263
diff --git a/PAN-X.te/xtreme-test.parquet b/PAN-X.te/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..bdb49a053b0a058013efc716b62aca9f5dd9a534
--- /dev/null
+++ b/PAN-X.te/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f58186a4c90e41f5b94e3ea6a002bb7dd59d91420ca4ee014997d7dce7802d85
+size 66993
diff --git a/PAN-X.te/xtreme-train.parquet b/PAN-X.te/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..44f3d5d540442dd82c2c883e12d341e0ec0a0821
--- /dev/null
+++ b/PAN-X.te/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b34d2a835546134172835b86d1e44bc5321c815554dab16f017d13786f3e9b83
+size 67658
diff --git a/PAN-X.te/xtreme-validation.parquet b/PAN-X.te/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..ab1b9794bb6c3b1be0d025aa3377b78924c08d8b
--- /dev/null
+++ b/PAN-X.te/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b0ca8eebd1675371978d1f9f1566ea9b142c80eb94de153e743e2a4d43266a48
+size 65781
diff --git a/PAN-X.th/xtreme-test.parquet b/PAN-X.th/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..0aadeb922fcfbc234958e0959d52d77022b89ad1
--- /dev/null
+++ b/PAN-X.th/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:470ff32baf8881f3a3c048a23c3f077dd5b2ba8979cebf0c65b721e6fab0d369
+size 645165
diff --git a/PAN-X.th/xtreme-train.parquet b/PAN-X.th/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..88123bc9b81c29caaffd8d05542c187a487e0b97
--- /dev/null
+++ b/PAN-X.th/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:08092235aca49f8f8a79ec483695560afc1565e0f450e41dffde15745216abeb
+size 1289705
diff --git a/PAN-X.th/xtreme-validation.parquet b/PAN-X.th/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..b3eb9ef0482e7c56cb47682b5d907b5305a08adc
--- /dev/null
+++ b/PAN-X.th/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1dcf9a1027791c363a0646ee98a00bbee25e78375b4ec22580a96818a22d2f7e
+size 633946
diff --git a/PAN-X.tl/xtreme-test.parquet b/PAN-X.tl/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..fc91edc0b685d5032ee79b32c263d85e57e940a5
--- /dev/null
+++ b/PAN-X.tl/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3128c754068ae6b682c090ff3a90a3af11ff4e85a95c9925b208113568c13da2
+size 27327
diff --git a/PAN-X.tl/xtreme-train.parquet b/PAN-X.tl/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..5b541aed57d2bf3de68cffb1401ce6807da0eeeb
--- /dev/null
+++ b/PAN-X.tl/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d0711a572c41e1cb3480538cf6e107a9fcc034bb8db32fb2132c11d624eac12c
+size 254135
diff --git a/PAN-X.tl/xtreme-validation.parquet b/PAN-X.tl/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..2afed0e29f1c465d8977ebc6c64570776f072e66
--- /dev/null
+++ b/PAN-X.tl/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d3a894171fabe3c6a09d2d0e887af2e3cc2382c667948c45988d2030f8375542
+size 26452
diff --git a/PAN-X.tr/xtreme-test.parquet b/PAN-X.tr/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..cf47bbbec262d2c4ec36283463ff8ba9e2710b44
--- /dev/null
+++ b/PAN-X.tr/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:15e24148ea8e4bb22dd7dfb4be381773151a103192718cadd0fdbe72a68b3fa9
+size 504026
diff --git a/PAN-X.tr/xtreme-train.parquet b/PAN-X.tr/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..33f5ddc76546404b7aa9ccec6bef22ca3e2640d8
--- /dev/null
+++ b/PAN-X.tr/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:243050b29edb8c144df257440bdf28ef563e46975c7efa59a482ee04d4434b02
+size 993483
diff --git a/PAN-X.tr/xtreme-validation.parquet b/PAN-X.tr/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..acc36cfbf99e349b5ad299003ba9377370a41544
--- /dev/null
+++ b/PAN-X.tr/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bed87378f9721c4a9659fbf6a6db17973bbc783f05ec6b3624a1076ad13c7efd
+size 502440
diff --git a/PAN-X.ur/xtreme-test.parquet b/PAN-X.ur/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..a83afbe8cec984b7a8ac2f8b09e4ea494bc89353
--- /dev/null
+++ b/PAN-X.ur/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ce03321cb231df698b5dc1efe72a0754640235c8989012edcaa13257bf3f4496
+size 28933
diff --git a/PAN-X.ur/xtreme-train.parquet b/PAN-X.ur/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..18689b029833655a04f9793e2b6ff083e925519a
--- /dev/null
+++ b/PAN-X.ur/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:36ce76c3a8947503426e7484ef3a684249bc11abc6f70c1cd7ebd91bcca280c7
+size 553003
diff --git a/PAN-X.ur/xtreme-validation.parquet b/PAN-X.ur/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..21d3ab47e692c669ffcff3df2f935ddbcf340425
--- /dev/null
+++ b/PAN-X.ur/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2b31b9d4082fe4d50b4c4f78fc3655a85d01c8bbabe42b57f4af5fad0649f1e5
+size 28507
diff --git a/PAN-X.vi/xtreme-test.parquet b/PAN-X.vi/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..73d02bfdedfc04cb7ffe4e6f29b7a4acacfe5beb
--- /dev/null
+++ b/PAN-X.vi/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a157e8845350876272679bfd79ca1446bd03af6b7d5466f63c9e24b7e607415
+size 345230
diff --git a/PAN-X.vi/xtreme-train.parquet b/PAN-X.vi/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..5c1cd64b03c3b304ecb6cac83de58a35b3107631
--- /dev/null
+++ b/PAN-X.vi/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a5b6451e2565ce9816774143a6154fb2b1898baff325bab967d4cd5fe6f0850c
+size 688362
diff --git a/PAN-X.vi/xtreme-validation.parquet b/PAN-X.vi/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..1e8cf3d31e1c0610e71dc96cfbce095626ce96cf
--- /dev/null
+++ b/PAN-X.vi/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b0314c486148dd01d7de3a865bbd0c6be956f9d81833ab28858f8bfe4607a9e6
+size 341289
diff --git a/PAN-X.yo/xtreme-test.parquet b/PAN-X.yo/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..2c7998c98ead1fd4ae01c7c9dbc8e105ebac09ee
--- /dev/null
+++ b/PAN-X.yo/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b1d3fba86cd24d20a8c0b928b2e7c4788a878cd2f3fdb37e47c0fe30c91899c6
+size 5526
diff --git a/PAN-X.yo/xtreme-train.parquet b/PAN-X.yo/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..ee18f3c70e34a339f738a3f6b4d65e30c3a43ef4
--- /dev/null
+++ b/PAN-X.yo/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:73cb3398eda7f86db437376505bc917aea91f52dacca1b8558de9b42999b314d
+size 6052
diff --git a/PAN-X.yo/xtreme-validation.parquet b/PAN-X.yo/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..53f64032f3b2c3b8cce5532197b271a5c54698f6
--- /dev/null
+++ b/PAN-X.yo/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b1c55bbee7e6ac36870d690c93dceadd1e22d7cc1bbc694e651509a76390c56d
+size 5675
diff --git a/PAN-X.zh/xtreme-test.parquet b/PAN-X.zh/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..b8d118ba8d970fb14f1a457e66b17a081bb390a7
--- /dev/null
+++ b/PAN-X.zh/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cd1914b35dc6f94800d269afeb82160389fe4573ea7e19189fa6db4b5a5b9d8b
+size 511198
diff --git a/PAN-X.zh/xtreme-train.parquet b/PAN-X.zh/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..9abc46ad056323d7956b1a3a175901e8df2ab8a0
--- /dev/null
+++ b/PAN-X.zh/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:697fcbe0f1d406978cd2e8a2426889e0e9fdb4dd99d5562b03543ba000a74a08
+size 1043262
diff --git a/PAN-X.zh/xtreme-validation.parquet b/PAN-X.zh/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..7f21ce04530ac19d7a825a67c709769888073668
--- /dev/null
+++ b/PAN-X.zh/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1fb9c805732a6e63d3816e71a9e8e51e3443f42ac37184d959bc47a81911ea14
+size 527988
diff --git a/PAWS-X.de/xtreme-test.parquet b/PAWS-X.de/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..a4967445c79217fee891845ae7ef95ccb879d002
--- /dev/null
+++ b/PAWS-X.de/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c5c65411803f798af2e06305b76efcb652add080770e0fad8392cc1b5f88394
+size 340530
diff --git a/PAWS-X.de/xtreme-train.parquet b/PAWS-X.de/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..93ea0e5446d7c0c58cb399c93500ab7d8972924c
--- /dev/null
+++ b/PAWS-X.de/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d77b3d9129511488d7e5ee62e668de453edd1644e0d5045169ecce14cf7a54a0
+size 8620465
diff --git a/PAWS-X.de/xtreme-validation.parquet b/PAWS-X.de/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..0bbe8ad8b6fc9557fb2f14f90b70818329cf8adc
--- /dev/null
+++ b/PAWS-X.de/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4e44e5a71daa3c24a3fe21e0758280bec4a16df74f5996a2270c4e0d6e8f3ed1
+size 333036
diff --git a/PAWS-X.en/xtreme-test.parquet b/PAWS-X.en/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..5b748496ecfd59d1c7696026d5239d384d551532
--- /dev/null
+++ b/PAWS-X.en/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72d9401181654d78f8bd679bd743190013bcd306c6d03aa846245848cb390ab0
+size 298431
diff --git a/PAWS-X.en/xtreme-train.parquet b/PAWS-X.en/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..660607c74adfb616f0b0c0d2d2d9e8a7f2584381
--- /dev/null
+++ b/PAWS-X.en/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:78f3495d24b92eb60d1d38b0c0392702cc5d5f18482464a1d58d956b25080e49
+size 8123609
diff --git a/PAWS-X.en/xtreme-validation.parquet b/PAWS-X.en/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..ba32691180e32b7b2d2fc8fe484866d8b460f3da
--- /dev/null
+++ b/PAWS-X.en/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bda1e57fa2e5c5127764854bf18cbdc6797c392e96fcbc0cb6bb734ca33a83ac
+size 295596
diff --git a/PAWS-X.es/xtreme-test.parquet b/PAWS-X.es/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..fe7375ed9efc6468a130a9da15eb1eca321849b1
--- /dev/null
+++ b/PAWS-X.es/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6fb0454122332943fed808d1da5bfbd24c08e5f3fda628338af8a20e02437389
+size 332390
diff --git a/PAWS-X.es/xtreme-train.parquet b/PAWS-X.es/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..87aaf80dff5c62ba4dbb0bee567f32f25e7cfc93
--- /dev/null
+++ b/PAWS-X.es/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9d7d6a626db10e38f64137428817e0c95498cd4e6ee3d92ef5178e28208462d9
+size 8571631
diff --git a/PAWS-X.es/xtreme-validation.parquet b/PAWS-X.es/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..4ac286b4b371c3d75378f1f04dd90cd7590ec20b
--- /dev/null
+++ b/PAWS-X.es/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:13de7181e08fba9d1b63022a4f5658b3e24445facb4e01b18a5819637d1fdaf2
+size 325894
diff --git a/PAWS-X.fr/xtreme-test.parquet b/PAWS-X.fr/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..3f2583ded83c2a10f3d45826395e2b8ab8772d35
--- /dev/null
+++ b/PAWS-X.fr/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:193eaf2399304d2058956715a213275113d22ffe8785b2aef1343ee6c45cee15
+size 340001
diff --git a/PAWS-X.fr/xtreme-train.parquet b/PAWS-X.fr/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..e28f547af2da739b0af205155d8f731d29f275b3
--- /dev/null
+++ b/PAWS-X.fr/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9012e054454553d81a5c2334cc34bd88fe75db83bc6d2ff5bd3cbacceffa9e40
+size 8792159
diff --git a/PAWS-X.fr/xtreme-validation.parquet b/PAWS-X.fr/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..6833163afa51b48c0b847aec0ec8b606d9b427bd
--- /dev/null
+++ b/PAWS-X.fr/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9376285e8e16e455602e2f3f57b1d0025b18ace97cacc2a55839a9821b453ae8
+size 332824
diff --git a/PAWS-X.ja/xtreme-test.parquet b/PAWS-X.ja/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..80dc60523879fa2070f44a3763c2cde5444a6344
--- /dev/null
+++ b/PAWS-X.ja/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3cb8978359416bd646ab9f45eac5c7acf0abcc7eb9a0baf590f05d26cfb78b0a
+size 398401
diff --git a/PAWS-X.ja/xtreme-train.parquet b/PAWS-X.ja/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..d3e1fa09924b869ddb54703cfde2839fe5f454f1
--- /dev/null
+++ b/PAWS-X.ja/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6cf5ab492a7197efe88cef85bda7f5f0df159520b7d35fd8ef1467baf60feb42
+size 9347910
diff --git a/PAWS-X.ja/xtreme-validation.parquet b/PAWS-X.ja/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..029bc8b95ad4c70dddc2cf4f68604ea9d33bc9f2
--- /dev/null
+++ b/PAWS-X.ja/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3964eb2c668e3a4d9912c931503626baef5dc96c0d8ae6f26657414820e3d58a
+size 389914
diff --git a/PAWS-X.ko/xtreme-test.parquet b/PAWS-X.ko/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..97079edfe4ebfe3eed6315ffdb7e5b88b7e6c68c
--- /dev/null
+++ b/PAWS-X.ko/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cea6d7c64ebfb3249693d87592c4adbd20a96f80b76be3c9149665907c087f07
+size 360975
diff --git a/PAWS-X.ko/xtreme-train.parquet b/PAWS-X.ko/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..bf3b81276ed6d1493df82cf597d5aaf2f29b116d
--- /dev/null
+++ b/PAWS-X.ko/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:51d207ba5f009802c6fabc4b29e4134de3b327f24ac699e0d1333e244c243275
+size 9214429
diff --git a/PAWS-X.ko/xtreme-validation.parquet b/PAWS-X.ko/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..31c16bba5e140f0e31a0fd396f86eca8ad7d0dbc
--- /dev/null
+++ b/PAWS-X.ko/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:50bb6ce2285a1b7da0a63103ed652d01fceddb1f3e7d7de1fddc9e7963810465
+size 350885
diff --git a/PAWS-X.zh/xtreme-test.parquet b/PAWS-X.zh/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..2dfe68d2cdb3e8a18bf9a636721c678919dbe656
--- /dev/null
+++ b/PAWS-X.zh/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5bfd2fd3ffcbc4ba6dfc802cd95fdc46030181c0f596e6966b50815af709d29f
+size 333442
diff --git a/PAWS-X.zh/xtreme-train.parquet b/PAWS-X.zh/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..009c341e36a1f49e4db779825e10a687490619cc
--- /dev/null
+++ b/PAWS-X.zh/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4e6328cee44d1ca0e35d1d47a50ac18eee16fea6f48be70d0f22cc4a1092b414
+size 8209894
diff --git a/PAWS-X.zh/xtreme-validation.parquet b/PAWS-X.zh/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..fe09662d39d61d83630c03bc3caa17e91f9144cf
--- /dev/null
+++ b/PAWS-X.zh/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bc100b878435150c225fc1a31649d58010ad7dfde7a2bdffe848962e519da972
+size 335516
diff --git a/README.md b/README.md
deleted file mode 100644
index 69ba203be02f33051ba93bce7567b3417a07bb62..0000000000000000000000000000000000000000
--- a/README.md
+++ /dev/null
@@ -1,5203 +0,0 @@
----
-annotations_creators:
-- found
-language_creators:
-- found
-language:
-- af
-- ar
-- bg
-- bn
-- de
-- el
-- en
-- es
-- et
-- eu
-- fa
-- fi
-- fr
-- he
-- hi
-- hu
-- id
-- it
-- ja
-- jv
-- ka
-- kk
-- ko
-- ml
-- mr
-- ms
-- my
-- nl
-- pt
-- ru
-- sw
-- ta
-- te
-- th
-- tl
-- tr
-- ur
-- vi
-- yo
-- zh
-language_bcp47:
-- fa-IR
-license:
-- apache-2.0
-- cc-by-4.0
-- cc-by-2.0
-- cc-by-sa-4.0
-- other
-- cc-by-nc-4.0
-license_details: Licence Universal Dependencies v2.5
-multilinguality:
-- multilingual
-- translation
-pretty_name: XTREME
-size_categories:
-- n<1K
-- 1K<n<10K
-- 10K<n<100K
-- 100K<n<1M
-source_datasets:
-- extended|xnli
-- extended|paws-x
-- extended|wikiann
-- extended|xquad
-- extended|mlqa
-- extended|tydiqa
-- extended|tatoeba
-- extended|squad
-task_categories:
-- multiple-choice
-- question-answering
-- token-classification
-- text-classification
-- text-retrieval
-- token-classification
-task_ids:
-- multiple-choice-qa
-- extractive-qa
-- open-domain-qa
-- natural-language-inference
-- named-entity-recognition
-- part-of-speech
-paperswithcode_id: xtreme
-configs:
-- MLQA.ar.ar
-- MLQA.ar.de
-- MLQA.ar.en
-- MLQA.ar.es
-- MLQA.ar.hi
-- MLQA.ar.vi
-- MLQA.ar.zh
-- MLQA.de.ar
-- MLQA.de.de
-- MLQA.de.en
-- MLQA.de.es
-- MLQA.de.hi
-- MLQA.de.vi
-- MLQA.de.zh
-- MLQA.en.ar
-- MLQA.en.de
-- MLQA.en.en
-- MLQA.en.es
-- MLQA.en.hi
-- MLQA.en.vi
-- MLQA.en.zh
-- MLQA.es.ar
-- MLQA.es.de
-- MLQA.es.en
-- MLQA.es.es
-- MLQA.es.hi
-- MLQA.es.vi
-- MLQA.es.zh
-- MLQA.hi.ar
-- MLQA.hi.de
-- MLQA.hi.en
-- MLQA.hi.es
-- MLQA.hi.hi
-- MLQA.hi.vi
-- MLQA.hi.zh
-- MLQA.vi.ar
-- MLQA.vi.de
-- MLQA.vi.en
-- MLQA.vi.es
-- MLQA.vi.hi
-- MLQA.vi.vi
-- MLQA.vi.zh
-- MLQA.zh.ar
-- MLQA.zh.de
-- MLQA.zh.en
-- MLQA.zh.es
-- MLQA.zh.hi
-- MLQA.zh.vi
-- MLQA.zh.zh
-- PAN-X.af
-- PAN-X.ar
-- PAN-X.bg
-- PAN-X.bn
-- PAN-X.de
-- PAN-X.el
-- PAN-X.en
-- PAN-X.es
-- PAN-X.et
-- PAN-X.eu
-- PAN-X.fa
-- PAN-X.fi
-- PAN-X.fr
-- PAN-X.he
-- PAN-X.hi
-- PAN-X.hu
-- PAN-X.id
-- PAN-X.it
-- PAN-X.ja
-- PAN-X.jv
-- PAN-X.ka
-- PAN-X.kk
-- PAN-X.ko
-- PAN-X.ml
-- PAN-X.mr
-- PAN-X.ms
-- PAN-X.my
-- PAN-X.nl
-- PAN-X.pt
-- PAN-X.ru
-- PAN-X.sw
-- PAN-X.ta
-- PAN-X.te
-- PAN-X.th
-- PAN-X.tl
-- PAN-X.tr
-- PAN-X.ur
-- PAN-X.vi
-- PAN-X.yo
-- PAN-X.zh
-- PAWS-X.de
-- PAWS-X.en
-- PAWS-X.es
-- PAWS-X.fr
-- PAWS-X.ja
-- PAWS-X.ko
-- PAWS-X.zh
-- SQuAD
-- XNLI
-- XQuAD
-- bucc18.de
-- bucc18.fr
-- bucc18.ru
-- bucc18.zh
-- tatoeba.afr
-- tatoeba.ara
-- tatoeba.ben
-- tatoeba.bul
-- tatoeba.cmn
-- tatoeba.deu
-- tatoeba.ell
-- tatoeba.est
-- tatoeba.eus
-- tatoeba.fin
-- tatoeba.fra
-- tatoeba.heb
-- tatoeba.hin
-- tatoeba.hun
-- tatoeba.ind
-- tatoeba.ita
-- tatoeba.jav
-- tatoeba.jpn
-- tatoeba.kat
-- tatoeba.kaz
-- tatoeba.kor
-- tatoeba.mal
-- tatoeba.mar
-- tatoeba.nld
-- tatoeba.pes
-- tatoeba.por
-- tatoeba.rus
-- tatoeba.spa
-- tatoeba.swh
-- tatoeba.tam
-- tatoeba.tel
-- tatoeba.tgl
-- tatoeba.tha
-- tatoeba.tur
-- tatoeba.urd
-- tatoeba.vie
-- tydiqa
-- udpos.Afrikans
-- udpos.Arabic
-- udpos.Basque
-- udpos.Bulgarian
-- udpos.Chinese
-- udpos.Dutch
-- udpos.English
-- udpos.Estonian
-- udpos.Finnish
-- udpos.French
-- udpos.German
-- udpos.Greek
-- udpos.Hebrew
-- udpos.Hindi
-- udpos.Hungarian
-- udpos.Indonesian
-- udpos.Italian
-- udpos.Japanese
-- udpos.Kazakh
-- udpos.Korean
-- udpos.Marathi
-- udpos.Persian
-- udpos.Portuguese
-- udpos.Russian
-- udpos.Spanish
-- udpos.Tagalog
-- udpos.Tamil
-- udpos.Telugu
-- udpos.Thai
-- udpos.Turkish
-- udpos.Urdu
-- udpos.Vietnamese
-- udpos.Yoruba
-tags:
-- parallel-sentence-retrieval
-- paraphrase-identification
-dataset_info:
-- config_name: XNLI
-  features:
-  - name: language
-    dtype: string
-  - name: sentence1
-    dtype: string
-  - name: sentence2
-    dtype: string
-  - name: gold_label
-    dtype: string
-  splits:
-  - name: test
-    num_bytes: 20359500
-    num_examples: 75150
-  - name: validation
-    num_bytes: 10049303
-    num_examples: 37350
-  download_size: 17865352
-  dataset_size: 30408803
-- config_name: tydiqa
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: train
-    num_bytes: 52948607
-    num_examples: 49881
-  - name: validation
-    num_bytes: 5006461
-    num_examples: 5077
-  download_size: 63621485
-  dataset_size: 57955068
-- config_name: SQuAD
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: train
-    num_bytes: 79317110
-    num_examples: 87599
-  - name: validation
-    num_bytes: 10472653
-    num_examples: 10570
-  download_size: 35142551
-  dataset_size: 89789763
-- config_name: PAN-X.af
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 259709
-    num_examples: 1000
-  - name: test
-    num_bytes: 257204
-    num_examples: 1000
-  - name: train
-    num_bytes: 1321396
-    num_examples: 5000
-  download_size: 234008884
-  dataset_size: 1838309
-- config_name: PAN-X.ar
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 1808303
-    num_examples: 10000
-  - name: test
-    num_bytes: 1811983
-    num_examples: 10000
-  - name: train
-    num_bytes: 3634136
-    num_examples: 20000
-  download_size: 234008884
-  dataset_size: 7254422
-- config_name: PAN-X.bg
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 2310314
-    num_examples: 10000
-  - name: test
-    num_bytes: 2306158
-    num_examples: 10000
-  - name: train
-    num_bytes: 4600773
-    num_examples: 20000
-  download_size: 234008884
-  dataset_size: 9217245
-- config_name: PAN-X.bn
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 159088
-    num_examples: 1000
-  - name: test
-    num_bytes: 159282
-    num_examples: 1000
-  - name: train
-    num_bytes: 1568845
-    num_examples: 10000
-  download_size: 234008884
-  dataset_size: 1887215
-- config_name: PAN-X.de
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 2381565
-    num_examples: 10000
-  - name: test
-    num_bytes: 2377639
-    num_examples: 10000
-  - name: train
-    num_bytes: 4762352
-    num_examples: 20000
-  download_size: 234008884
-  dataset_size: 9521556
-- config_name: PAN-X.el
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 2533806
-    num_examples: 10000
-  - name: test
-    num_bytes: 2547594
-    num_examples: 10000
-  - name: train
-    num_bytes: 5063176
-    num_examples: 20000
-  download_size: 234008884
-  dataset_size: 10144576
-- config_name: PAN-X.en
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 1920069
-    num_examples: 10000
-  - name: test
-    num_bytes: 1916220
-    num_examples: 10000
-  - name: train
-    num_bytes: 3823474
-    num_examples: 20000
-  download_size: 234008884
-  dataset_size: 7659763
-- config_name: PAN-X.es
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 1592525
-    num_examples: 10000
-  - name: test
-    num_bytes: 1602291
-    num_examples: 10000
-  - name: train
-    num_bytes: 3199161
-    num_examples: 20000
-  download_size: 234008884
-  dataset_size: 6393977
-- config_name: PAN-X.et
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 2030160
-    num_examples: 10000
-  - name: test
-    num_bytes: 2021409
-    num_examples: 10000
-  - name: train
-    num_bytes: 3023211
-    num_examples: 15000
-  download_size: 234008884
-  dataset_size: 7074780
-- config_name: PAN-X.eu
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 2296335
-    num_examples: 10000
-  - name: test
-    num_bytes: 2249835
-    num_examples: 10000
-  - name: train
-    num_bytes: 2292327
-    num_examples: 10000
-  download_size: 234008884
-  dataset_size: 6838497
-- config_name: PAN-X.fa
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 1782306
-    num_examples: 10000
-  - name: test
-    num_bytes: 1770284
-    num_examples: 10000
-  - name: train
-    num_bytes: 3529354
-    num_examples: 20000
-  download_size: 234008884
-  dataset_size: 7081944
-- config_name: PAN-X.fi
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 2131769
-    num_examples: 10000
-  - name: test
-    num_bytes: 2130665
-    num_examples: 10000
-  - name: train
-    num_bytes: 4273793
-    num_examples: 20000
-  download_size: 234008884
-  dataset_size: 8536227
-- config_name: PAN-X.fr
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 1664190
-    num_examples: 10000
-  - name: test
-    num_bytes: 1675785
-    num_examples: 10000
-  - name: train
-    num_bytes: 3335424
-    num_examples: 20000
-  download_size: 234008884
-  dataset_size: 6675399
-- config_name: PAN-X.he
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 2332760
-    num_examples: 10000
-  - name: test
-    num_bytes: 2318756
-    num_examples: 10000
-  - name: train
-    num_bytes: 4667100
-    num_examples: 20000
-  download_size: 234008884
-  dataset_size: 9318616
-- config_name: PAN-X.hi
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 190671
-    num_examples: 1000
-  - name: test
-    num_bytes: 196190
-    num_examples: 1000
-  - name: train
-    num_bytes: 964212
-    num_examples: 5000
-  download_size: 234008884
-  dataset_size: 1351073
-- config_name: PAN-X.hu
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 2211851
-    num_examples: 10000
-  - name: test
-    num_bytes: 2249779
-    num_examples: 10000
-  - name: train
-    num_bytes: 4499914
-    num_examples: 20000
-  download_size: 234008884
-  dataset_size: 8961544
-- config_name: PAN-X.id
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 1537979
-    num_examples: 10000
-  - name: test
-    num_bytes: 1536879
-    num_examples: 10000
-  - name: train
-    num_bytes: 3084007
-    num_examples: 20000
-  download_size: 234008884
-  dataset_size: 6158865
-- config_name: PAN-X.it
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 1908529
-    num_examples: 10000
-  - name: test
-    num_bytes: 1928408
-    num_examples: 10000
-  - name: train
-    num_bytes: 3874663
-    num_examples: 20000
-  download_size: 234008884
-  dataset_size: 7711600
-- config_name: PAN-X.ja
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 6323003
-    num_examples: 10000
-  - name: test
-    num_bytes: 6448960
-    num_examples: 10000
-  - name: train
-    num_bytes: 12670401
-    num_examples: 20000
-  download_size: 234008884
-  dataset_size: 25442364
-- config_name: PAN-X.jv
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 14600
-    num_examples: 100
-  - name: test
-    num_bytes: 16917
-    num_examples: 100
-  - name: train
-    num_bytes: 16106
-    num_examples: 100
-  download_size: 234008884
-  dataset_size: 47623
-- config_name: PAN-X.ka
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 2806901
-    num_examples: 10000
-  - name: test
-    num_bytes: 2824641
-    num_examples: 10000
-  - name: train
-    num_bytes: 2777362
-    num_examples: 10000
-  download_size: 234008884
-  dataset_size: 8408904
-- config_name: PAN-X.kk
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 238109
-    num_examples: 1000
-  - name: test
-    num_bytes: 236724
-    num_examples: 1000
-  - name: train
-    num_bytes: 240276
-    num_examples: 1000
-  download_size: 234008884
-  dataset_size: 715109
-- config_name: PAN-X.ko
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 2138167
-    num_examples: 10000
-  - name: test
-    num_bytes: 2138294
-    num_examples: 10000
-  - name: train
-    num_bytes: 4284733
-    num_examples: 20000
-  download_size: 234008884
-  dataset_size: 8561194
-- config_name: PAN-X.ml
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 290755
-    num_examples: 1000
-  - name: test
-    num_bytes: 276926
-    num_examples: 1000
-  - name: train
-    num_bytes: 2865204
-    num_examples: 10000
-  download_size: 234008884
-  dataset_size: 3432885
-- config_name: PAN-X.mr
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 245358
-    num_examples: 1000
-  - name: test
-    num_bytes: 255904
-    num_examples: 1000
-  - name: train
-    num_bytes: 1248259
-    num_examples: 5000
-  download_size: 234008884
-  dataset_size: 1749521
-- config_name: PAN-X.ms
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 147515
-    num_examples: 1000
-  - name: test
-    num_bytes: 147168
-    num_examples: 1000
-  - name: train
-    num_bytes: 2965048
-    num_examples: 20000
-  download_size: 234008884
-  dataset_size: 3259731
-- config_name: PAN-X.my
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 40428
-    num_examples: 100
-  - name: test
-    num_bytes: 37366
-    num_examples: 100
-  - name: train
-    num_bytes: 32735
-    num_examples: 100
-  download_size: 234008884
-  dataset_size: 110529
-- config_name: PAN-X.nl
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 2016856
-    num_examples: 10000
-  - name: test
-    num_bytes: 2038638
-    num_examples: 10000
-  - name: train
-    num_bytes: 4062189
-    num_examples: 20000
-  download_size: 234008884
-  dataset_size: 8117683
-- config_name: PAN-X.pt
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 1575141
-    num_examples: 10000
-  - name: test
-    num_bytes: 1562625
-    num_examples: 10000
-  - name: train
-    num_bytes: 3149283
-    num_examples: 20000
-  download_size: 234008884
-  dataset_size: 6287049
-- config_name: PAN-X.ru
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 2053169
-    num_examples: 10000
-  - name: test
-    num_bytes: 2074145
-    num_examples: 10000
-  - name: train
-    num_bytes: 4121791
-    num_examples: 20000
-  download_size: 234008884
-  dataset_size: 8249105
-- config_name: PAN-X.sw
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 136368
-    num_examples: 1000
-  - name: test
-    num_bytes: 140231
-    num_examples: 1000
-  - name: train
-    num_bytes: 135911
-    num_examples: 1000
-  download_size: 234008884
-  dataset_size: 412510
-- config_name: PAN-X.ta
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 277625
-    num_examples: 1000
-  - name: test
-    num_bytes: 278114
-    num_examples: 1000
-  - name: train
-    num_bytes: 4122130
-    num_examples: 15000
-  download_size: 234008884
-  dataset_size: 4677869
-- config_name: PAN-X.te
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 293281
-    num_examples: 1000
-  - name: test
-    num_bytes: 296963
-    num_examples: 1000
-  - name: train
-    num_bytes: 295410
-    num_examples: 1000
-  download_size: 234008884
-  dataset_size: 885654
-- config_name: PAN-X.th
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 13262737
-    num_examples: 10000
-  - name: test
-    num_bytes: 13586928
-    num_examples: 10000
-  - name: train
-    num_bytes: 27133029
-    num_examples: 20000
-  download_size: 234008884
-  dataset_size: 53982694
-- config_name: PAN-X.tl
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 114156
-    num_examples: 1000
-  - name: test
-    num_bytes: 117904
-    num_examples: 1000
-  - name: train
-    num_bytes: 1168717
-    num_examples: 10000
-  download_size: 234008884
-  dataset_size: 1400777
-- config_name: PAN-X.tr
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 1915352
-    num_examples: 10000
-  - name: test
-    num_bytes: 1911503
-    num_examples: 10000
-  - name: train
-    num_bytes: 3779170
-    num_examples: 20000
-  download_size: 234008884
-  dataset_size: 7606025
-- config_name: PAN-X.ur
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 152148
-    num_examples: 1000
-  - name: test
-    num_bytes: 151922
-    num_examples: 1000
-  - name: train
-    num_bytes: 3072276
-    num_examples: 20000
-  download_size: 234008884
-  dataset_size: 3376346
-- config_name: PAN-X.vi
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 1565143
-    num_examples: 10000
-  - name: test
-    num_bytes: 1580216
-    num_examples: 10000
-  - name: train
-    num_bytes: 3153227
-    num_examples: 20000
-  download_size: 234008884
-  dataset_size: 6298586
-- config_name: PAN-X.yo
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 13245
-    num_examples: 100
-  - name: test
-    num_bytes: 13533
-    num_examples: 100
-  - name: train
-    num_bytes: 14709
-    num_examples: 100
-  download_size: 234008884
-  dataset_size: 41487
-- config_name: PAN-X.zh
-  features:
-  - name: tokens
-    sequence: string
-  - name: ner_tags
-    sequence:
-      class_label:
-        names:
-          0: O
-          1: B-PER
-          2: I-PER
-          3: B-ORG
-          4: I-ORG
-          5: B-LOC
-          6: I-LOC
-  - name: langs
-    sequence: string
-  splits:
-  - name: validation
-    num_bytes: 4491325
-    num_examples: 10000
-  - name: test
-    num_bytes: 4363172
-    num_examples: 10000
-  - name: train
-    num_bytes: 8832051
-    num_examples: 20000
-  download_size: 234008884
-  dataset_size: 17686548
-- config_name: MLQA.ar.ar
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 8368114
-    num_examples: 5335
-  - name: validation
-    num_bytes: 824108
-    num_examples: 517
-  download_size: 75719050
-  dataset_size: 9192222
-- config_name: MLQA.ar.de
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 2183942
-    num_examples: 1649
-  - name: validation
-    num_bytes: 364837
-    num_examples: 207
-  download_size: 75719050
-  dataset_size: 2548779
-- config_name: MLQA.ar.vi
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 3290629
-    num_examples: 2047
-  - name: validation
-    num_bytes: 288446
-    num_examples: 163
-  download_size: 75719050
-  dataset_size: 3579075
-- config_name: MLQA.ar.zh
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 3229872
-    num_examples: 1912
-  - name: validation
-    num_bytes: 340049
-    num_examples: 188
-  download_size: 75719050
-  dataset_size: 3569921
-- config_name: MLQA.ar.en
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 8225662
-    num_examples: 5335
-  - name: validation
-    num_bytes: 810089
-    num_examples: 517
-  download_size: 75719050
-  dataset_size: 9035751
-- config_name: MLQA.ar.es
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 3041378
-    num_examples: 1978
-  - name: validation
-    num_bytes: 228180
-    num_examples: 161
-  download_size: 75719050
-  dataset_size: 3269558
-- config_name: MLQA.ar.hi
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 3039396
-    num_examples: 1831
-  - name: validation
-    num_bytes: 281770
-    num_examples: 186
-  download_size: 75719050
-  dataset_size: 3321166
-- config_name: MLQA.de.ar
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 1620006
-    num_examples: 1649
-  - name: validation
-    num_bytes: 200174
-    num_examples: 207
-  download_size: 75719050
-  dataset_size: 1820180
-- config_name: MLQA.de.de
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 4366102
-    num_examples: 4517
-  - name: validation
-    num_bytes: 488367
-    num_examples: 512
-  download_size: 75719050
-  dataset_size: 4854469
-- config_name: MLQA.de.vi
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 1688483
-    num_examples: 1675
-  - name: validation
-    num_bytes: 216075
-    num_examples: 182
-  download_size: 75719050
-  dataset_size: 1904558
-- config_name: MLQA.de.zh
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 1679180
-    num_examples: 1621
-  - name: validation
-    num_bytes: 184318
-    num_examples: 190
-  download_size: 75719050
-  dataset_size: 1863498
-- config_name: MLQA.de.en
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 4343144
-    num_examples: 4517
-  - name: validation
-    num_bytes: 485894
-    num_examples: 512
-  download_size: 75719050
-  dataset_size: 4829038
-- config_name: MLQA.de.es
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 1716615
-    num_examples: 1776
-  - name: validation
-    num_bytes: 170582
-    num_examples: 196
-  download_size: 75719050
-  dataset_size: 1887197
-- config_name: MLQA.de.hi
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 1371074
-    num_examples: 1430
-  - name: validation
-    num_bytes: 153871
-    num_examples: 163
-  download_size: 75719050
-  dataset_size: 1524945
-- config_name: MLQA.vi.ar
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 3205185
-    num_examples: 2047
-  - name: validation
-    num_bytes: 230335
-    num_examples: 163
-  download_size: 75719050
-  dataset_size: 3435520
-- config_name: MLQA.vi.de
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 2227033
-    num_examples: 1675
-  - name: validation
-    num_bytes: 277185
-    num_examples: 182
-  download_size: 75719050
-  dataset_size: 2504218
-- config_name: MLQA.vi.vi
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 7922085
-    num_examples: 5495
-  - name: validation
-    num_bytes: 726518
-    num_examples: 511
-  download_size: 75719050
-  dataset_size: 8648603
-- config_name: MLQA.vi.zh
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 2989660
-    num_examples: 1943
-  - name: validation
-    num_bytes: 269389
-    num_examples: 184
-  download_size: 75719050
-  dataset_size: 3259049
-- config_name: MLQA.vi.en
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 7843431
-    num_examples: 5495
-  - name: validation
-    num_bytes: 719273
-    num_examples: 511
-  download_size: 75719050
-  dataset_size: 8562704
-- config_name: MLQA.vi.es
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 2866597
-    num_examples: 2018
-  - name: validation
-    num_bytes: 283461
-    num_examples: 189
-  download_size: 75719050
-  dataset_size: 3150058
-- config_name: MLQA.vi.hi
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 2776664
-    num_examples: 1947
-  - name: validation
-    num_bytes: 255007
-    num_examples: 177
-  download_size: 75719050
-  dataset_size: 3031671
-- config_name: MLQA.zh.ar
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 1731483
-    num_examples: 1912
-  - name: validation
-    num_bytes: 175349
-    num_examples: 188
-  download_size: 75719050
-  dataset_size: 1906832
-- config_name: MLQA.zh.de
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 1390018
-    num_examples: 1621
-  - name: validation
-    num_bytes: 174605
-    num_examples: 190
-  download_size: 75719050
-  dataset_size: 1564623
-- config_name: MLQA.zh.vi
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 1806186
-    num_examples: 1943
-  - name: validation
-    num_bytes: 172934
-    num_examples: 184
-  download_size: 75719050
-  dataset_size: 1979120
-- config_name: MLQA.zh.zh
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 4422350
-    num_examples: 5137
-  - name: validation
-    num_bytes: 443810
-    num_examples: 504
-  download_size: 75719050
-  dataset_size: 4866160
-- config_name: MLQA.zh.en
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 4450985
-    num_examples: 5137
-  - name: validation
-    num_bytes: 446868
-    num_examples: 504
-  download_size: 75719050
-  dataset_size: 4897853
-- config_name: MLQA.zh.es
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 1736283
-    num_examples: 1947
-  - name: validation
-    num_bytes: 138073
-    num_examples: 161
-  download_size: 75719050
-  dataset_size: 1874356
-- config_name: MLQA.zh.hi
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 1578219
-    num_examples: 1767
-  - name: validation
-    num_bytes: 184401
-    num_examples: 189
-  download_size: 75719050
-  dataset_size: 1762620
-- config_name: MLQA.en.ar
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 6739219
-    num_examples: 5335
-  - name: validation
-    num_bytes: 630843
-    num_examples: 517
-  download_size: 75719050
-  dataset_size: 7370062
-- config_name: MLQA.en.de
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 5056722
-    num_examples: 4517
-  - name: validation
-    num_bytes: 594936
-    num_examples: 512
-  download_size: 75719050
-  dataset_size: 5651658
-- config_name: MLQA.en.vi
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 7056698
-    num_examples: 5495
-  - name: validation
-    num_bytes: 640646
-    num_examples: 511
-  download_size: 75719050
-  dataset_size: 7697344
-- config_name: MLQA.en.zh
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 6539307
-    num_examples: 5137
-  - name: validation
-    num_bytes: 608444
-    num_examples: 504
-  download_size: 75719050
-  dataset_size: 7147751
-- config_name: MLQA.en.en
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 14004648
-    num_examples: 11590
-  - name: validation
-    num_bytes: 1329112
-    num_examples: 1148
-  download_size: 75719050
-  dataset_size: 15333760
-- config_name: MLQA.en.es
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 6179249
-    num_examples: 5253
-  - name: validation
-    num_bytes: 555462
-    num_examples: 500
-  download_size: 75719050
-  dataset_size: 6734711
-- config_name: MLQA.en.hi
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 6378866
-    num_examples: 4918
-  - name: validation
-    num_bytes: 623171
-    num_examples: 507
-  download_size: 75719050
-  dataset_size: 7002037
-- config_name: MLQA.es.ar
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 1740282
-    num_examples: 1978
-  - name: validation
-    num_bytes: 148649
-    num_examples: 161
-  download_size: 75719050
-  dataset_size: 1888931
-- config_name: MLQA.es.de
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 1404025
-    num_examples: 1776
-  - name: validation
-    num_bytes: 144186
-    num_examples: 196
-  download_size: 75719050
-  dataset_size: 1548211
-- config_name: MLQA.es.vi
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 1747969
-    num_examples: 2018
-  - name: validation
-    num_bytes: 176841
-    num_examples: 189
-  download_size: 75719050
-  dataset_size: 1924810
-- config_name: MLQA.es.zh
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 1678451
-    num_examples: 1947
-  - name: validation
-    num_bytes: 126646
-    num_examples: 161
-  download_size: 75719050
-  dataset_size: 1805097
-- config_name: MLQA.es.en
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 4362737
-    num_examples: 5253
-  - name: validation
-    num_bytes: 419068
-    num_examples: 500
-  download_size: 75719050
-  dataset_size: 4781805
-- config_name: MLQA.es.es
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 4394333
-    num_examples: 5253
-  - name: validation
-    num_bytes: 422071
-    num_examples: 500
-  download_size: 75719050
-  dataset_size: 4816404
-- config_name: MLQA.es.hi
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 1523523
-    num_examples: 1723
-  - name: validation
-    num_bytes: 181834
-    num_examples: 187
-  download_size: 75719050
-  dataset_size: 1705357
-- config_name: MLQA.hi.ar
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 4445589
-    num_examples: 1831
-  - name: validation
-    num_bytes: 410424
-    num_examples: 186
-  download_size: 75719050
-  dataset_size: 4856013
-- config_name: MLQA.hi.de
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 3022864
-    num_examples: 1430
-  - name: validation
-    num_bytes: 301713
-    num_examples: 163
-  download_size: 75719050
-  dataset_size: 3324577
-- config_name: MLQA.hi.vi
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 4743484
-    num_examples: 1947
-  - name: validation
-    num_bytes: 419106
-    num_examples: 177
-  download_size: 75719050
-  dataset_size: 5162590
-- config_name: MLQA.hi.zh
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 4354875
-    num_examples: 1767
-  - name: validation
-    num_bytes: 424246
-    num_examples: 189
-  download_size: 75719050
-  dataset_size: 4779121
-- config_name: MLQA.hi.en
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 11449261
-    num_examples: 4918
-  - name: validation
-    num_bytes: 1097857
-    num_examples: 507
-  download_size: 75719050
-  dataset_size: 12547118
-- config_name: MLQA.hi.es
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 3862229
-    num_examples: 1723
-  - name: validation
-    num_bytes: 420402
-    num_examples: 187
-  download_size: 75719050
-  dataset_size: 4282631
-- config_name: MLQA.hi.hi
-  features:
-  - name: id
-    dtype: string
-  - name: title
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: test
-    num_bytes: 11810475
-    num_examples: 4918
-  - name: validation
-    num_bytes: 1136784
-    num_examples: 507
-  download_size: 75719050
-  dataset_size: 12947259
-- config_name: XQuAD.ar
-  features:
-  - name: id
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: validation
-    num_bytes: 1722799
-    num_examples: 1190
-  download_size: 1582988
-  dataset_size: 1722799
-- config_name: XQuAD.de
-  features:
-  - name: id
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: validation
-    num_bytes: 1283301
-    num_examples: 1190
-  download_size: 669810
-  dataset_size: 1283301
-- config_name: XQuAD.vi
-  features:
-  - name: id
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: validation
-    num_bytes: 1477239
-    num_examples: 1190
-  download_size: 911401
-  dataset_size: 1477239
-- config_name: XQuAD.zh
-  features:
-  - name: id
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: validation
-    num_bytes: 984241
-    num_examples: 1190
-  download_size: 808652
-  dataset_size: 984241
-- config_name: XQuAD.en
-  features:
-  - name: id
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: validation
-    num_bytes: 1116123
-    num_examples: 1190
-  download_size: 609383
-  dataset_size: 1116123
-- config_name: XQuAD.es
-  features:
-  - name: id
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: validation
-    num_bytes: 1273499
-    num_examples: 1190
-  download_size: 684322
-  dataset_size: 1273499
-- config_name: XQuAD.hi
-  features:
-  - name: id
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: validation
-    num_bytes: 2682975
-    num_examples: 1190
-  download_size: 1680538
-  dataset_size: 2682975
-- config_name: XQuAD.el
-  features:
-  - name: id
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: validation
-    num_bytes: 2206690
-    num_examples: 1190
-  download_size: 1918889
-  dataset_size: 2206690
-- config_name: XQuAD.ru
-  features:
-  - name: id
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: validation
-    num_bytes: 2136990
-    num_examples: 1190
-  download_size: 1896368
-  dataset_size: 2136990
-- config_name: XQuAD.th
-  features:
-  - name: id
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: validation
-    num_bytes: 2854959
-    num_examples: 1190
-  download_size: 1809143
-  dataset_size: 2854959
-- config_name: XQuAD.tr
-  features:
-  - name: id
-    dtype: string
-  - name: context
-    dtype: string
-  - name: question
-    dtype: string
-  - name: answers
-    sequence:
-    - name: answer_start
-      dtype: int32
-    - name: text
-      dtype: string
-  splits:
-  - name: validation
-    num_bytes: 1210763
-    num_examples: 1190
-  download_size: 729506
-  dataset_size: 1210763
-- config_name: bucc18.de
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 248707
-    num_examples: 1038
-  - name: test
-    num_bytes: 2325701
-    num_examples: 9580
-  download_size: 30719200
-  dataset_size: 2574408
-- config_name: bucc18.fr
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 212513
-    num_examples: 929
-  - name: test
-    num_bytes: 2082419
-    num_examples: 9086
-  download_size: 22706544
-  dataset_size: 2294932
-- config_name: bucc18.zh
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 55739
-    num_examples: 257
-  - name: test
-    num_bytes: 415925
-    num_examples: 1899
-  download_size: 7114794
-  dataset_size: 471664
-- config_name: bucc18.ru
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 761347
-    num_examples: 2374
-  - name: test
-    num_bytes: 4641678
-    num_examples: 14435
-  download_size: 41354312
-  dataset_size: 5403025
-- config_name: PAWS-X.de
-  features:
-  - name: sentence1
-    dtype: string
-  - name: sentence2
-    dtype: string
-  - name: label
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 500009
-    num_examples: 2000
-  - name: test
-    num_bytes: 510194
-    num_examples: 2000
-  - name: train
-    num_bytes: 12451883
-    num_examples: 49380
-  download_size: 30282057
-  dataset_size: 13462086
-- config_name: PAWS-X.en
-  features:
-  - name: sentence1
-    dtype: string
-  - name: sentence2
-    dtype: string
-  - name: label
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 478291
-    num_examples: 2000
-  - name: test
-    num_bytes: 480738
-    num_examples: 2000
-  - name: train
-    num_bytes: 11827719
-    num_examples: 49175
-  download_size: 30282057
-  dataset_size: 12786748
-- config_name: PAWS-X.es
-  features:
-  - name: sentence1
-    dtype: string
-  - name: sentence2
-    dtype: string
-  - name: label
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 494069
-    num_examples: 1961
-  - name: test
-    num_bytes: 505047
-    num_examples: 2000
-  - name: train
-    num_bytes: 12462107
-    num_examples: 49401
-  download_size: 30282057
-  dataset_size: 13461223
-- config_name: PAWS-X.fr
-  features:
-  - name: sentence1
-    dtype: string
-  - name: sentence2
-    dtype: string
-  - name: label
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 516111
-    num_examples: 1988
-  - name: test
-    num_bytes: 521031
-    num_examples: 2000
-  - name: train
-    num_bytes: 12948512
-    num_examples: 49399
-  download_size: 30282057
-  dataset_size: 13985654
-- config_name: PAWS-X.ja
-  features:
-  - name: sentence1
-    dtype: string
-  - name: sentence2
-    dtype: string
-  - name: label
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 647774
-    num_examples: 2000
-  - name: test
-    num_bytes: 654640
-    num_examples: 2000
-  - name: train
-    num_bytes: 14695653
-    num_examples: 49401
-  download_size: 30282057
-  dataset_size: 15998067
-- config_name: PAWS-X.ko
-  features:
-  - name: sentence1
-    dtype: string
-  - name: sentence2
-    dtype: string
-  - name: label
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 540787
-    num_examples: 2000
-  - name: test
-    num_bytes: 547978
-    num_examples: 1999
-  - name: train
-    num_bytes: 13542657
-    num_examples: 49164
-  download_size: 30282057
-  dataset_size: 14631422
-- config_name: PAWS-X.zh
-  features:
-  - name: sentence1
-    dtype: string
-  - name: sentence2
-    dtype: string
-  - name: label
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 459120
-    num_examples: 2000
-  - name: test
-    num_bytes: 460638
-    num_examples: 2000
-  - name: train
-    num_bytes: 10469712
-    num_examples: 49401
-  download_size: 30282057
-  dataset_size: 11389470
-- config_name: tatoeba.afr
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 179651
-    num_examples: 1000
-  download_size: 59635
-  dataset_size: 179651
-- config_name: tatoeba.ara
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 192666
-    num_examples: 1000
-  download_size: 72650
-  dataset_size: 192666
-- config_name: tatoeba.ben
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 211719
-    num_examples: 1000
-  download_size: 91703
-  dataset_size: 211719
-- config_name: tatoeba.bul
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 222295
-    num_examples: 1000
-  download_size: 102279
-  dataset_size: 222295
-- config_name: tatoeba.deu
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 225583
-    num_examples: 1000
-  download_size: 105567
-  dataset_size: 225583
-- config_name: tatoeba.cmn
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 188947
-    num_examples: 1000
-  download_size: 68931
-  dataset_size: 188947
-- config_name: tatoeba.ell
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 198977
-    num_examples: 1000
-  download_size: 78961
-  dataset_size: 198977
-- config_name: tatoeba.est
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 179744
-    num_examples: 1000
-  download_size: 59728
-  dataset_size: 179744
-- config_name: tatoeba.eus
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 186084
-    num_examples: 1000
-  download_size: 66068
-  dataset_size: 186084
-- config_name: tatoeba.fin
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 195685
-    num_examples: 1000
-  download_size: 75669
-  dataset_size: 195685
-- config_name: tatoeba.fra
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 200034
-    num_examples: 1000
-  download_size: 80018
-  dataset_size: 200034
-- config_name: tatoeba.heb
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 203516
-    num_examples: 1000
-  download_size: 83500
-  dataset_size: 203516
-- config_name: tatoeba.hin
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 242574
-    num_examples: 1000
-  download_size: 122558
-  dataset_size: 242574
-- config_name: tatoeba.hun
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 188905
-    num_examples: 1000
-  download_size: 68889
-  dataset_size: 188905
-- config_name: tatoeba.ind
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 194860
-    num_examples: 1000
-  download_size: 74844
-  dataset_size: 194860
-- config_name: tatoeba.ita
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 185849
-    num_examples: 1000
-  download_size: 65833
-  dataset_size: 185849
-- config_name: tatoeba.jav
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 38529
-    num_examples: 205
-  download_size: 13913
-  dataset_size: 38529
-- config_name: tatoeba.jpn
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 213099
-    num_examples: 1000
-  download_size: 93083
-  dataset_size: 213099
-- config_name: tatoeba.kat
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 161696
-    num_examples: 746
-  download_size: 72160
-  dataset_size: 161696
-- config_name: tatoeba.kaz
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 116194
-    num_examples: 575
-  download_size: 47178
-  dataset_size: 116194
-- config_name: tatoeba.kor
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 199155
-    num_examples: 1000
-  download_size: 79139
-  dataset_size: 199155
-- config_name: tatoeba.mal
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 177173
-    num_examples: 687
-  download_size: 94717
-  dataset_size: 177173
-- config_name: tatoeba.mar
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 220558
-    num_examples: 1000
-  download_size: 100542
-  dataset_size: 220558
-- config_name: tatoeba.nld
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 193279
-    num_examples: 1000
-  download_size: 73263
-  dataset_size: 193279
-- config_name: tatoeba.pes
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 213735
-    num_examples: 1000
-  download_size: 93719
-  dataset_size: 213735
-- config_name: tatoeba.por
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 195201
-    num_examples: 1000
-  download_size: 75185
-  dataset_size: 195201
-- config_name: tatoeba.rus
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 212488
-    num_examples: 1000
-  download_size: 92472
-  dataset_size: 212488
-- config_name: tatoeba.spa
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 192282
-    num_examples: 1000
-  download_size: 72266
-  dataset_size: 192282
-- config_name: tatoeba.swh
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 67283
-    num_examples: 390
-  download_size: 20467
-  dataset_size: 67283
-- config_name: tatoeba.tam
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 76297
-    num_examples: 307
-  download_size: 39441
-  dataset_size: 76297
-- config_name: tatoeba.tel
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 53239
-    num_examples: 234
-  download_size: 25143
-  dataset_size: 53239
-- config_name: tatoeba.tgl
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 188154
-    num_examples: 1000
-  download_size: 68138
-  dataset_size: 188154
-- config_name: tatoeba.tha
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 128974
-    num_examples: 548
-  download_size: 63198
-  dataset_size: 128974
-- config_name: tatoeba.tur
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 191901
-    num_examples: 1000
-  download_size: 71885
-  dataset_size: 191901
-- config_name: tatoeba.urd
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 208728
-    num_examples: 1000
-  download_size: 88712
-  dataset_size: 208728
-- config_name: tatoeba.vie
-  features:
-  - name: source_sentence
-    dtype: string
-  - name: target_sentence
-    dtype: string
-  - name: source_lang
-    dtype: string
-  - name: target_lang
-    dtype: string
-  splits:
-  - name: validation
-    num_bytes: 211423
-    num_examples: 1000
-  download_size: 91407
-  dataset_size: 211423
-- config_name: udpos.Afrikaans
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 91302
-    num_examples: 194
-  - name: test
-    num_bytes: 174256
-    num_examples: 425
-  - name: train
-    num_bytes: 586382
-    num_examples: 1315
-  download_size: 355216681
-  dataset_size: 851940
-- config_name: udpos.Arabic
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 593662
-    num_examples: 909
-  - name: test
-    num_bytes: 973834
-    num_examples: 1680
-  - name: train
-    num_bytes: 4453694
-    num_examples: 6075
-  download_size: 355216681
-  dataset_size: 6021190
-- config_name: udpos.Basque
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 438683
-    num_examples: 1798
-  - name: test
-    num_bytes: 444656
-    num_examples: 1799
-  - name: train
-    num_bytes: 1327725
-    num_examples: 5396
-  download_size: 355216681
-  dataset_size: 2211064
-- config_name: udpos.Bulgarian
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 347129
-    num_examples: 1115
-  - name: test
-    num_bytes: 339959
-    num_examples: 1116
-  - name: train
-    num_bytes: 2689779
-    num_examples: 8907
-  download_size: 355216681
-  dataset_size: 3376867
-- config_name: udpos.Dutch
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 393604
-    num_examples: 1394
-  - name: test
-    num_bytes: 397916
-    num_examples: 1471
-  - name: train
-    num_bytes: 4518018
-    num_examples: 18051
-  download_size: 355216681
-  dataset_size: 5309538
-- config_name: udpos.English
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 1042052
-    num_examples: 3974
-  - name: test
-    num_bytes: 1421160
-    num_examples: 5440
-  - name: train
-    num_bytes: 6225545
-    num_examples: 21253
-  download_size: 355216681
-  dataset_size: 8688757
-- config_name: udpos.Estonian
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 814183
-    num_examples: 3125
-  - name: test
-    num_bytes: 1065713
-    num_examples: 3760
-  - name: train
-    num_bytes: 6614929
-    num_examples: 25749
-  download_size: 355216681
-  dataset_size: 8494825
-- config_name: udpos.Finnish
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 656658
-    num_examples: 3239
-  - name: test
-    num_bytes: 1025738
-    num_examples: 4422
-  - name: train
-    num_bytes: 5613742
-    num_examples: 27198
-  download_size: 355216681
-  dataset_size: 7296138
-- config_name: udpos.French
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 1294108
-    num_examples: 5979
-  - name: test
-    num_bytes: 1731061
-    num_examples: 9465
-  - name: train
-    num_bytes: 10118993
-    num_examples: 47308
-  download_size: 355216681
-  dataset_size: 13144162
-- config_name: udpos.German
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 6044862
-    num_examples: 19233
-  - name: test
-    num_bytes: 7345899
-    num_examples: 22458
-  - name: train
-    num_bytes: 54773981
-    num_examples: 166849
-  download_size: 355216681
-  dataset_size: 68164742
-- config_name: udpos.Greek
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 1062459
-    num_examples: 2559
-  - name: test
-    num_bytes: 1028677
-    num_examples: 2809
-  - name: train
-    num_bytes: 8932140
-    num_examples: 28152
-  download_size: 355216681
-  dataset_size: 11023276
-- config_name: udpos.Hebrew
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 210025
-    num_examples: 484
-  - name: test
-    num_bytes: 223877
-    num_examples: 491
-  - name: train
-    num_bytes: 2505703
-    num_examples: 5241
-  download_size: 355216681
-  dataset_size: 2939605
-- config_name: udpos.Hindi
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 839714
-    num_examples: 1659
-  - name: test
-    num_bytes: 1400237
-    num_examples: 2684
-  - name: train
-    num_bytes: 6690274
-    num_examples: 13304
-  download_size: 355216681
-  dataset_size: 8930225
-- config_name: udpos.Hungarian
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 215891
-    num_examples: 441
-  - name: test
-    num_bytes: 193740
-    num_examples: 449
-  - name: train
-    num_bytes: 372238
-    num_examples: 910
-  download_size: 355216681
-  dataset_size: 781869
-- config_name: udpos.Indonesian
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 220875
-    num_examples: 559
-  - name: test
-    num_bytes: 557113
-    num_examples: 1557
-  - name: train
-    num_bytes: 1710690
-    num_examples: 4477
-  download_size: 355216681
-  dataset_size: 2488678
-- config_name: udpos.Italian
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 989008
-    num_examples: 2278
-  - name: test
-    num_bytes: 1337881
-    num_examples: 3518
-  - name: train
-    num_bytes: 11299329
-    num_examples: 29685
-  download_size: 355216681
-  dataset_size: 13626218
-- config_name: udpos.Japanese
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 200368
-    num_examples: 511
-  - name: test
-    num_bytes: 928914
-    num_examples: 2372
-  - name: train
-    num_bytes: 2792963
-    num_examples: 7125
-  download_size: 355216681
-  dataset_size: 3922245
-- config_name: udpos.Kazakh
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: test
-    num_bytes: 228936
-    num_examples: 1047
-  - name: train
-    num_bytes: 11450
-    num_examples: 31
-  download_size: 355216681
-  dataset_size: 240386
-- config_name: udpos.Korean
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 782599
-    num_examples: 3016
-  - name: test
-    num_bytes: 1162551
-    num_examples: 4276
-  - name: train
-    num_bytes: 7341303
-    num_examples: 27410
-  download_size: 355216681
-  dataset_size: 9286453
-- config_name: udpos.Chinese
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 594460
-    num_examples: 3038
-  - name: test
-    num_bytes: 1236063
-    num_examples: 5528
-  - name: train
-    num_bytes: 4218915
-    num_examples: 18998
-  download_size: 355216681
-  dataset_size: 6049438
-- config_name: udpos.Marathi
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 8509
-    num_examples: 46
-  - name: test
-    num_bytes: 7883
-    num_examples: 47
-  - name: train
-    num_bytes: 59035
-    num_examples: 373
-  download_size: 355216681
-  dataset_size: 75427
-- config_name: udpos.Persian
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 317065
-    num_examples: 599
-  - name: test
-    num_bytes: 320695
-    num_examples: 600
-  - name: train
-    num_bytes: 2400788
-    num_examples: 4798
-  download_size: 355216681
-  dataset_size: 3038548
-- config_name: udpos.Portuguese
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 712409
-    num_examples: 1770
-  - name: test
-    num_bytes: 1082594
-    num_examples: 2681
-  - name: train
-    num_bytes: 7669580
-    num_examples: 17992
-  download_size: 355216681
-  dataset_size: 9464583
-- config_name: udpos.Russian
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 3457043
-    num_examples: 9960
-  - name: test
-    num_bytes: 4236717
-    num_examples: 11336
-  - name: train
-    num_bytes: 24230182
-    num_examples: 67435
-  download_size: 355216681
-  dataset_size: 31923942
-- config_name: udpos.Spanish
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 1498777
-    num_examples: 3054
-  - name: test
-    num_bytes: 1476512
-    num_examples: 3147
-  - name: train
-    num_bytes: 13858442
-    num_examples: 28492
-  download_size: 355216681
-  dataset_size: 16833731
-- config_name: udpos.Tagalog
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: test
-    num_bytes: 5165
-    num_examples: 55
-  download_size: 355216681
-  dataset_size: 5165
-- config_name: udpos.Tamil
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 40043
-    num_examples: 80
-  - name: test
-    num_bytes: 62378
-    num_examples: 120
-  - name: train
-    num_bytes: 202608
-    num_examples: 400
-  download_size: 355216681
-  dataset_size: 305029
-- config_name: udpos.Telugu
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 18002
-    num_examples: 131
-  - name: test
-    num_bytes: 19587
-    num_examples: 146
-  - name: train
-    num_bytes: 138061
-    num_examples: 1051
-  download_size: 355216681
-  dataset_size: 175650
-- config_name: udpos.Thai
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: test
-    num_bytes: 561348
-    num_examples: 1000
-  download_size: 355216681
-  dataset_size: 561348
-- config_name: udpos.Turkish
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 186467
-    num_examples: 988
-  - name: test
-    num_bytes: 827394
-    num_examples: 4785
-  - name: train
-    num_bytes: 704417
-    num_examples: 3664
-  download_size: 355216681
-  dataset_size: 1718278
-- config_name: udpos.Urdu
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 284273
-    num_examples: 552
-  - name: test
-    num_bytes: 288565
-    num_examples: 535
-  - name: train
-    num_bytes: 2107374
-    num_examples: 4043
-  download_size: 355216681
-  dataset_size: 2680212
-- config_name: udpos.Vietnamese
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: validation
-    num_bytes: 206200
-    num_examples: 800
-  - name: test
-    num_bytes: 214075
-    num_examples: 800
-  - name: train
-    num_bytes: 367347
-    num_examples: 1400
-  download_size: 355216681
-  dataset_size: 787622
-- config_name: udpos.Yoruba
-  features:
-  - name: tokens
-    sequence: string
-  - name: pos_tags
-    sequence:
-      class_label:
-        names:
-          0: ADJ
-          1: ADP
-          2: ADV
-          3: AUX
-          4: CCONJ
-          5: DET
-          6: INTJ
-          7: NOUN
-          8: NUM
-          9: PART
-          10: PRON
-          11: PROPN
-          12: PUNCT
-          13: SCONJ
-          14: SYM
-          15: VERB
-          16: X
-  splits:
-  - name: test
-    num_bytes: 44668
-    num_examples: 100
-  download_size: 355216681
-  dataset_size: 44668
----
-
-# Dataset Card for "xtreme"
-
-## Table of Contents
-- [Dataset Description](#dataset-description)
-  - [Dataset Summary](#dataset-summary)
-  - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards)
-  - [Languages](#languages)
-- [Dataset Structure](#dataset-structure)
-  - [Data Instances](#data-instances)
-  - [Data Fields](#data-fields)
-  - [Data Splits](#data-splits)
-- [Dataset Creation](#dataset-creation)
-  - [Curation Rationale](#curation-rationale)
-  - [Source Data](#source-data)
-  - [Annotations](#annotations)
-  - [Personal and Sensitive Information](#personal-and-sensitive-information)
-- [Considerations for Using the Data](#considerations-for-using-the-data)
-  - [Social Impact of Dataset](#social-impact-of-dataset)
-  - [Discussion of Biases](#discussion-of-biases)
-  - [Other Known Limitations](#other-known-limitations)
-- [Additional Information](#additional-information)
-  - [Dataset Curators](#dataset-curators)
-  - [Licensing Information](#licensing-information)
-  - [Citation Information](#citation-information)
-  - [Contributions](#contributions)
-
-## Dataset Description
-
-- **Homepage:** [https://github.com/google-research/xtreme](https://github.com/google-research/xtreme)
-- **Repository:** [More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
-- **Paper:** [More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
-- **Point of Contact:** [More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
-- **Size of downloaded dataset files:** 15143.21 MB
-- **Size of the generated dataset:** 1027.42 MB
-- **Total amount of disk used:** 16170.64 MB
-
-### Dataset Summary
-
-The Cross-lingual Natural Language Inference (XNLI) corpus is a crowd-sourced collection of 5,000 test and
-2,500 dev pairs for the MultiNLI corpus. The pairs are annotated with textual entailment and translated into
-14 languages: French, Spanish, German, Greek, Bulgarian, Russian, Turkish, Arabic, Vietnamese, Thai, Chinese,
-Hindi, Swahili and Urdu. This results in 112.5k annotated pairs. Each premise can be associated with the
-corresponding hypothesis in the 15 languages, summing up to more than 1.5M combinations. The corpus is made to
-evaluate how to perform inference in any language (including low-resources ones like Swahili or Urdu) when only
-English NLI data is available at training time. One solution is cross-lingual sentence encoding, for which XNLI
-is an evaluation benchmark.
-The Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of
-the cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages
-(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of
-syntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,
-and availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil
-(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the
-Niger-Congo languages Swahili and Yoruba, spoken in Africa.
-
-### Supported Tasks and Leaderboards
-
-[More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
-
-### Languages
-
-[More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
-
-## Dataset Structure
-
-### Data Instances
-
-#### MLQA.ar.ar
-
-- **Size of downloaded dataset files:** 72.21 MB
-- **Size of the generated dataset:** 8.77 MB
-- **Total amount of disk used:** 80.98 MB
-
-An example of 'validation' looks as follows.
-```
-
-```
-
-#### MLQA.ar.de
-
-- **Size of downloaded dataset files:** 72.21 MB
-- **Size of the generated dataset:** 2.43 MB
-- **Total amount of disk used:** 74.64 MB
-
-An example of 'validation' looks as follows.
-```
-
-```
-
-#### MLQA.ar.en
-
-- **Size of downloaded dataset files:** 72.21 MB
-- **Size of the generated dataset:** 8.62 MB
-- **Total amount of disk used:** 80.83 MB
-
-An example of 'validation' looks as follows.
-```
-
-```
-
-#### MLQA.ar.es
-
-- **Size of downloaded dataset files:** 72.21 MB
-- **Size of the generated dataset:** 3.12 MB
-- **Total amount of disk used:** 75.33 MB
-
-An example of 'validation' looks as follows.
-```
-
-```
-
-#### MLQA.ar.hi
-
-- **Size of downloaded dataset files:** 72.21 MB
-- **Size of the generated dataset:** 3.17 MB
-- **Total amount of disk used:** 75.38 MB
-
-An example of 'validation' looks as follows.
-```
-
-```
-
-### Data Fields
-
-The data fields are the same among all splits.
-
-#### MLQA.ar.ar
-- `id`: a `string` feature.
-- `title`: a `string` feature.
-- `context`: a `string` feature.
-- `question`: a `string` feature.
-- `answers`: a dictionary feature containing:
-  - `answer_start`: a `int32` feature.
-  - `text`: a `string` feature.
-
-#### MLQA.ar.de
-- `id`: a `string` feature.
-- `title`: a `string` feature.
-- `context`: a `string` feature.
-- `question`: a `string` feature.
-- `answers`: a dictionary feature containing:
-  - `answer_start`: a `int32` feature.
-  - `text`: a `string` feature.
-
-#### MLQA.ar.en
-- `id`: a `string` feature.
-- `title`: a `string` feature.
-- `context`: a `string` feature.
-- `question`: a `string` feature.
-- `answers`: a dictionary feature containing:
-  - `answer_start`: a `int32` feature.
-  - `text`: a `string` feature.
-
-#### MLQA.ar.es
-- `id`: a `string` feature.
-- `title`: a `string` feature.
-- `context`: a `string` feature.
-- `question`: a `string` feature.
-- `answers`: a dictionary feature containing:
-  - `answer_start`: a `int32` feature.
-  - `text`: a `string` feature.
-
-#### MLQA.ar.hi
-- `id`: a `string` feature.
-- `title`: a `string` feature.
-- `context`: a `string` feature.
-- `question`: a `string` feature.
-- `answers`: a dictionary feature containing:
-  - `answer_start`: a `int32` feature.
-  - `text`: a `string` feature.
-
-### Data Splits
-
-|   name   |validation|test|
-|----------|---------:|---:|
-|MLQA.ar.ar|       517|5335|
-|MLQA.ar.de|       207|1649|
-|MLQA.ar.en|       517|5335|
-|MLQA.ar.es|       161|1978|
-|MLQA.ar.hi|       186|1831|
-
-## Dataset Creation
-
-### Curation Rationale
-
-[More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
-
-### Source Data
-
-#### Initial Data Collection and Normalization
-
-[More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
-
-#### Who are the source language producers?
-
-[More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
-
-### Annotations
-
-#### Annotation process
-
-[More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
-
-#### Who are the annotators?
-
-[More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
-
-### Personal and Sensitive Information
-
-[More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
-
-## Considerations for Using the Data
-
-### Social Impact of Dataset
-
-[More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
-
-### Discussion of Biases
-
-[More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
-
-### Other Known Limitations
-
-[More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
-
-## Additional Information
-
-### Dataset Curators
-
-[More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
-
-### Licensing Information
-
-[More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
-
-### Citation Information
-
-```
-  @InProceedings{conneau2018xnli,
-  author = {Conneau, Alexis
-                 and Rinott, Ruty
-                 and Lample, Guillaume
-                 and Williams, Adina
-                 and Bowman, Samuel R.
-                 and Schwenk, Holger
-                 and Stoyanov, Veselin},
-  title = {XNLI: Evaluating Cross-lingual Sentence Representations},
-  booktitle = {Proceedings of the 2018 Conference on Empirical Methods
-               in Natural Language Processing},
-  year = {2018},
-  publisher = {Association for Computational Linguistics},
-  location = {Brussels, Belgium},
-}
-@article{hu2020xtreme,
-      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},
-      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},
-      journal   = {CoRR},
-      volume    = {abs/2003.11080},
-      year      = {2020},
-      archivePrefix = {arXiv},
-      eprint    = {2003.11080}
-}
-
-```
-
-
-### Contributions
-
-Thanks to [@thomwolf](https://github.com/thomwolf), [@jplu](https://github.com/jplu), [@lewtun](https://github.com/lewtun), [@lvwerra](https://github.com/lvwerra), [@lhoestq](https://github.com/lhoestq), [@patrickvonplaten](https://github.com/patrickvonplaten), [@mariamabarham](https://github.com/mariamabarham) for adding this dataset.
\ No newline at end of file
diff --git a/SQuAD/xtreme-train.parquet b/SQuAD/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..6b0f701ac1c3ce508a32013543594d368ebff11f
--- /dev/null
+++ b/SQuAD/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eb6b4a505ee00c3053461f6409951526bed8623ccd5fff7070ea65886a7735ed
+size 14452916
diff --git a/SQuAD/xtreme-validation.parquet b/SQuAD/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..19a45e557f5251d1790c3cd57bbfd60ac2549f80
--- /dev/null
+++ b/SQuAD/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f91abd2d24170d92957801bb04c28055c43a82ee6950374f8ad7f42024df2cd1
+size 1818538
diff --git a/XNLI/xtreme-test.parquet b/XNLI/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..238ab294a165b2f5bc3263dff686a88438dab71d
--- /dev/null
+++ b/XNLI/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7f6443b4bdef015701d14889bccab168761fdd3c18965e8d3e4b444ca9cb6aab
+size 5908397
diff --git a/XNLI/xtreme-validation.parquet b/XNLI/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..03c8ecda95392c622f19f664f9fd90fab054923c
--- /dev/null
+++ b/XNLI/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a49dc7babe720695225e4cacea6d8fc7b9fba442692a0ffa5e29fdd88954a9c4
+size 2973224
diff --git a/XQuAD.ar/xtreme-validation.parquet b/XQuAD.ar/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..e6c85f6601cad0ee4aecda3a61907cbd1636dd39
--- /dev/null
+++ b/XQuAD.ar/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d7db95d92030936ed95306d8b186c22f568dd8188b1337bc7a72c662d6d3d036
+size 263001
diff --git a/XQuAD.de/xtreme-validation.parquet b/XQuAD.de/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..a52ac1afa3ed3953bcff8f2c5ae43159fe0680b4
--- /dev/null
+++ b/XQuAD.de/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:15333997be85e3203a687fe491a8c62a1f75a7f814b85aa26fb2e891a7b9f28e
+size 241956
diff --git a/XQuAD.el/xtreme-validation.parquet b/XQuAD.el/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..acffa98f4c1cb15554fcc7c295c0f55108526ba7
--- /dev/null
+++ b/XQuAD.el/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a5b6d45c059d7d48055a20e549635d9fc67758996b7ba54236d644a26664297c
+size 324378
diff --git a/XQuAD.en/xtreme-validation.parquet b/XQuAD.en/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..33811efe453ce142a48c232675405a1533776136
--- /dev/null
+++ b/XQuAD.en/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:68e0e2a8e4dd50d457316cb5eff4c5f667f429cb256df9dcfc905b334c9971a1
+size 212371
diff --git a/XQuAD.es/xtreme-validation.parquet b/XQuAD.es/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..2e80d05dedece17275f1849222efb8ff0e0d82ff
--- /dev/null
+++ b/XQuAD.es/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fdfd33fa5efe89eb1fb0bed536e45486a2df7b3c038813fc51ac9604f8da02e
+size 236873
diff --git a/XQuAD.hi/xtreme-validation.parquet b/XQuAD.hi/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..ac86c01ad1b5bc7c6a2c8dca6608f7c88c9b293a
--- /dev/null
+++ b/XQuAD.hi/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:84fabe15921f392d9b9c9ef64bbb5a4c90887f128365cabaf189c2cd0bd66a59
+size 322082
diff --git a/XQuAD.ru/xtreme-validation.parquet b/XQuAD.ru/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..8609b3b84a3694c639d5face4dcad710a19f5783
--- /dev/null
+++ b/XQuAD.ru/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8484d0d7038e79715e73ccf91a6e912bb1598bc4bd9a15967c4d09240dfb719a
+size 321727
diff --git a/XQuAD.th/xtreme-validation.parquet b/XQuAD.th/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..3091cdc37f084761f7c7a62b3e8fc4b2dba554bd
--- /dev/null
+++ b/XQuAD.th/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5d9cefdd6ae926966124f1c58beaf7e9f65b03c3fb29e62d8a11f91e4f4db5d6
+size 337306
diff --git a/XQuAD.tr/xtreme-validation.parquet b/XQuAD.tr/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..3bd521e83719fd25dba062966c97643c2b246697
--- /dev/null
+++ b/XQuAD.tr/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e9e5cc9e4609ee9b8cf1dcf2ab88eb7beb1bd69b4395372c1de523a7ba03c236
+size 228363
diff --git a/XQuAD.vi/xtreme-validation.parquet b/XQuAD.vi/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..f8b472441dc88f43687d5e96b397c623ac4612db
--- /dev/null
+++ b/XQuAD.vi/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:263faa7c96c31da07487aeff950d807f97ba89131c11baf65f53a028d31486a6
+size 237643
diff --git a/XQuAD.zh/xtreme-validation.parquet b/XQuAD.zh/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..0bd967e6f6a61456e192922e85ee9a8623a293af
--- /dev/null
+++ b/XQuAD.zh/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80b9bf716a992de6e88aed3bd931aa4a1bf2ec8ec4241e4c0937d206dc456d44
+size 205767
diff --git a/bucc18.de/xtreme-test.parquet b/bucc18.de/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..9423b254b0b929af0e9972252b080001933d7adf
--- /dev/null
+++ b/bucc18.de/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8cedf4e3fd64a987b5069913f882516c61ae430b1833889628b05ab00eb597ff
+size 1477633
diff --git a/bucc18.de/xtreme-validation.parquet b/bucc18.de/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..e90d4223c57049ecc8f7479aaeb72498bb09606e
--- /dev/null
+++ b/bucc18.de/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b1264078f5a4573214fa34b0fe4e3580c26ee6e44331a199a4e3a962ee9d5a31
+size 158495
diff --git a/bucc18.fr/xtreme-test.parquet b/bucc18.fr/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..db4c6dbc1bb29333f2470692600cfee3d65a429e
--- /dev/null
+++ b/bucc18.fr/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5367711b797bc3e25b132b0b0a40cce85ec7b4ef7e62845af7c5678f969b97b5
+size 1306858
diff --git a/bucc18.fr/xtreme-validation.parquet b/bucc18.fr/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..09e1dbdf05614fd78ede6611659bc9045c980ff3
--- /dev/null
+++ b/bucc18.fr/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d0dde6421ce745fccc49d05c68df49f2f5dbbac51ee72116b305e42e40319564
+size 130236
diff --git a/bucc18.ru/xtreme-test.parquet b/bucc18.ru/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..1ea3d14b1bb2b1a8ceadfbceb0750ce01d31bdfd
--- /dev/null
+++ b/bucc18.ru/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8612b40635ca6865840a1fccd56c42d0236a2f60f2044406f7e08f8a4169e7a4
+size 2646079
diff --git a/bucc18.ru/xtreme-validation.parquet b/bucc18.ru/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..dadca11f0291c47237b775aa74cfbaa4825b452d
--- /dev/null
+++ b/bucc18.ru/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:863fc2539b7ff75596df1e4ef1ee162bcbcf1c06b6dba6e70ad5383f9fdf4bff
+size 428395
diff --git a/bucc18.zh/xtreme-test.parquet b/bucc18.zh/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..5a430c30958a590ddab8269fe041018ad83fede4
--- /dev/null
+++ b/bucc18.zh/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3f14e649a31368bf256837b62e5ca5ec4c1fefd1e6c0be5f0df6875698782d2e
+size 279659
diff --git a/bucc18.zh/xtreme-validation.parquet b/bucc18.zh/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..0a1479ec13b15eb85723042145eea48839c0776d
--- /dev/null
+++ b/bucc18.zh/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1c01fdcc0d46205f04724e3c3bbab57b63ddbfefb4e512997221108fdfafc17f
+size 40717
diff --git a/dataset_infos.json b/dataset_infos.json
deleted file mode 100644
index cc42ff3127dd37987f1e72d1c47438958c503d95..0000000000000000000000000000000000000000
--- a/dataset_infos.json
+++ /dev/null
@@ -1 +0,0 @@
-{"XNLI": {"description": "\nThe Cross-lingual Natural Language Inference (XNLI) corpus is a crowd-sourced collection of 5,000 test and\n2,500 dev pairs for the MultiNLI corpus. The pairs are annotated with textual entailment and translated into\n14 languages: French, Spanish, German, Greek, Bulgarian, Russian, Turkish, Arabic, Vietnamese, Thai, Chinese,\nHindi, Swahili and Urdu. This results in 112.5k annotated pairs. Each premise can be associated with the\ncorresponding hypothesis in the 15 languages, summing up to more than 1.5M combinations. The corpus is made to\nevaluate how to perform inference in any language (including low-resources ones like Swahili or Urdu) when only\nEnglish NLI data is available at training time. One solution is cross-lingual sentence encoding, for which XNLI\nis an evaluation benchmark.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "  @InProceedings{conneau2018xnli,\n  author = {Conneau, Alexis\n                 and Rinott, Ruty\n                 and Lample, Guillaume\n                 and Williams, Adina\n                 and Bowman, Samuel R.\n                 and Schwenk, Holger\n                 and Stoyanov, Veselin},\n  title = {XNLI: Evaluating Cross-lingual Sentence Representations},\n  booktitle = {Proceedings of the 2018 Conference on Empirical Methods\n               in Natural Language Processing},\n  year = {2018},\n  publisher = {Association for Computational Linguistics},\n  location = {Brussels, Belgium},\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://www.nyu.edu/projects/bowman/xnli/", "license": "", "features": {"language": {"dtype": "string", "id": null, "_type": "Value"}, "sentence1": {"dtype": "string", "id": null, "_type": "Value"}, "sentence2": {"dtype": "string", "id": null, "_type": "Value"}, "gold_label": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "XNLI", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 20359500, "num_examples": 75150, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 10049303, "num_examples": 37350, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/XNLI/XNLI-1.0.zip": {"num_bytes": 17865352, "checksum": "4ba1d5e1afdb7161f0f23c66dc787802ccfa8a25a3ddd3b165a35e50df346ab1"}}, "download_size": 17865352, "post_processing_size": null, "dataset_size": 30408803, "size_in_bytes": 48274155}, "tydiqa": {"description": "Gold passage task (GoldP): Given a passage that is guaranteed to contain the\n             answer, predict the single contiguous span of characters that answers the question. This is more similar to\n             existing reading comprehension datasets (as opposed to the information-seeking task outlined above).\n             This task is constructed with two goals in mind: (1) more directly comparing with prior work and (2) providing\n             a simplified way for researchers to use TyDi QA by providing compatibility with existing code for SQuAD 1.1,\n             XQuAD, and MLQA. Toward these goals, the gold passage task differs from the primary task in several ways:\n             only the gold answer passage is provided rather than the entire Wikipedia article;\n             unanswerable questions have been discarded, similar to MLQA and XQuAD;\n             we evaluate with the SQuAD 1.1 metrics like XQuAD; and\n            Thai and Japanese are removed since the lack of whitespace breaks some tools.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{tydiqa,\n  title   = {TyDi QA: A Benchmark for Information-Seeking Question Answering in Typologically Diverse Languages},\n  author  = {Jonathan H. Clark and Eunsol Choi and Michael Collins and Dan Garrette and Tom Kwiatkowski and Vitaly Nikolaev and Jennimaria Palomaki}\n  year    = {2020},\n  journal = {Transactions of the Association for Computational Linguistics}\n  }\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/google-research-datasets/tydiqa", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tydiqa", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 52948607, "num_examples": 49881, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 5006461, "num_examples": 5077, "dataset_name": "xtreme"}}, "download_checksums": {"https://storage.googleapis.com/tydiqa/v1.1/tydiqa-goldp-v1.1-train.json": {"num_bytes": 58004076, "checksum": "cefc8e09ff2548d9b10a678d3a6bbbe5bc036be543f92418819ea676c97be23b"}, "https://storage.googleapis.com/tydiqa/v1.1/tydiqa-goldp-v1.1-dev.json": {"num_bytes": 5617409, "checksum": "b286e0f34bc7f52259359989716f369b160565bd12ad8f3a3e311f9b0dbad1c0"}}, "download_size": 63621485, "post_processing_size": null, "dataset_size": 57955068, "size_in_bytes": 121576553}, "SQuAD": {"description": "Stanford Question Answering Dataset (SQuAD) is a reading comprehension     dataset, consisting of questions posed by crowdworkers on a set of Wikipedia     articles, where the answer to every question is a segment of text, or span,     from the corresponding reading passage, or the question might be unanswerable.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "        @article{2016arXiv160605250R,\n           author = {{Rajpurkar}, Pranav and {Zhang}, Jian and {Lopyrev},\n                     Konstantin and {Liang}, Percy},\n            title = \"{SQuAD: 100,000+ Questions for Machine Comprehension of Text}\",\n          journal = {arXiv e-prints},\n             year = 2016,\n              eid = {arXiv:1606.05250},\n            pages = {arXiv:1606.05250},\n            archivePrefix = {arXiv},\n           eprint = {1606.05250},\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://rajpurkar.github.io/SQuAD-explorer/", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "SQuAD", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 79317110, "num_examples": 87599, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 10472653, "num_examples": 10570, "dataset_name": "xtreme"}}, "download_checksums": {"https://rajpurkar.github.io/SQuAD-explorer/dataset/train-v1.1.json": {"num_bytes": 30288272, "checksum": "3527663986b8295af4f7fcdff1ba1ff3f72d07d61a20f487cb238a6ef92fd955"}, "https://rajpurkar.github.io/SQuAD-explorer/dataset/dev-v1.1.json": {"num_bytes": 4854279, "checksum": "95aa6a52d5d6a735563366753ca50492a658031da74f301ac5238b03966972c9"}}, "download_size": 35142551, "post_processing_size": null, "dataset_size": 89789763, "size_in_bytes": 124932314}, "PAN-X.af": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.af", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 259709, "num_examples": 1000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 257204, "num_examples": 1000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 1321396, "num_examples": 5000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 1838309, "size_in_bytes": 235847193}, "PAN-X.ar": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.ar", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 1808303, "num_examples": 10000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 1811983, "num_examples": 10000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 3634136, "num_examples": 20000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 7254422, "size_in_bytes": 241263306}, "PAN-X.bg": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.bg", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 2310314, "num_examples": 10000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 2306158, "num_examples": 10000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 4600773, "num_examples": 20000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 9217245, "size_in_bytes": 243226129}, "PAN-X.bn": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.bn", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 159088, "num_examples": 1000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 159282, "num_examples": 1000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 1568845, "num_examples": 10000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 1887215, "size_in_bytes": 235896099}, "PAN-X.de": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.de", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 2381565, "num_examples": 10000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 2377639, "num_examples": 10000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 4762352, "num_examples": 20000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 9521556, "size_in_bytes": 243530440}, "PAN-X.el": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.el", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 2533806, "num_examples": 10000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 2547594, "num_examples": 10000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 5063176, "num_examples": 20000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 10144576, "size_in_bytes": 244153460}, "PAN-X.en": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.en", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 1920069, "num_examples": 10000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 1916220, "num_examples": 10000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 3823474, "num_examples": 20000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 7659763, "size_in_bytes": 241668647}, "PAN-X.es": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.es", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 1592525, "num_examples": 10000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 1602291, "num_examples": 10000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 3199161, "num_examples": 20000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 6393977, "size_in_bytes": 240402861}, "PAN-X.et": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.et", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 2030160, "num_examples": 10000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 2021409, "num_examples": 10000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 3023211, "num_examples": 15000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 7074780, "size_in_bytes": 241083664}, "PAN-X.eu": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.eu", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 2296335, "num_examples": 10000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 2249835, "num_examples": 10000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 2292327, "num_examples": 10000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 6838497, "size_in_bytes": 240847381}, "PAN-X.fa": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.fa", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 1782306, "num_examples": 10000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 1770284, "num_examples": 10000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 3529354, "num_examples": 20000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 7081944, "size_in_bytes": 241090828}, "PAN-X.fi": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.fi", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 2131769, "num_examples": 10000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 2130665, "num_examples": 10000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 4273793, "num_examples": 20000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 8536227, "size_in_bytes": 242545111}, "PAN-X.fr": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.fr", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 1664190, "num_examples": 10000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 1675785, "num_examples": 10000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 3335424, "num_examples": 20000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 6675399, "size_in_bytes": 240684283}, "PAN-X.he": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.he", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 2332760, "num_examples": 10000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 2318756, "num_examples": 10000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 4667100, "num_examples": 20000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 9318616, "size_in_bytes": 243327500}, "PAN-X.hi": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.hi", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 190671, "num_examples": 1000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 196190, "num_examples": 1000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 964212, "num_examples": 5000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 1351073, "size_in_bytes": 235359957}, "PAN-X.hu": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.hu", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 2211851, "num_examples": 10000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 2249779, "num_examples": 10000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 4499914, "num_examples": 20000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 8961544, "size_in_bytes": 242970428}, "PAN-X.id": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.id", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 1537979, "num_examples": 10000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 1536879, "num_examples": 10000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 3084007, "num_examples": 20000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 6158865, "size_in_bytes": 240167749}, "PAN-X.it": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.it", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 1908529, "num_examples": 10000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 1928408, "num_examples": 10000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 3874663, "num_examples": 20000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 7711600, "size_in_bytes": 241720484}, "PAN-X.ja": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.ja", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 6323003, "num_examples": 10000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 6448960, "num_examples": 10000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 12670401, "num_examples": 20000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 25442364, "size_in_bytes": 259451248}, "PAN-X.jv": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.jv", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 14600, "num_examples": 100, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 16917, "num_examples": 100, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 16106, "num_examples": 100, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 47623, "size_in_bytes": 234056507}, "PAN-X.ka": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.ka", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 2806901, "num_examples": 10000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 2824641, "num_examples": 10000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 2777362, "num_examples": 10000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 8408904, "size_in_bytes": 242417788}, "PAN-X.kk": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.kk", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 238109, "num_examples": 1000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 236724, "num_examples": 1000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 240276, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 715109, "size_in_bytes": 234723993}, "PAN-X.ko": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.ko", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 2138167, "num_examples": 10000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 2138294, "num_examples": 10000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 4284733, "num_examples": 20000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 8561194, "size_in_bytes": 242570078}, "PAN-X.ml": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.ml", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 290755, "num_examples": 1000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 276926, "num_examples": 1000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 2865204, "num_examples": 10000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 3432885, "size_in_bytes": 237441769}, "PAN-X.mr": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.mr", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 245358, "num_examples": 1000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 255904, "num_examples": 1000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 1248259, "num_examples": 5000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 1749521, "size_in_bytes": 235758405}, "PAN-X.ms": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.ms", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 147515, "num_examples": 1000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 147168, "num_examples": 1000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 2965048, "num_examples": 20000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 3259731, "size_in_bytes": 237268615}, "PAN-X.my": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.my", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 40428, "num_examples": 100, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 37366, "num_examples": 100, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 32735, "num_examples": 100, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 110529, "size_in_bytes": 234119413}, "PAN-X.nl": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.nl", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 2016856, "num_examples": 10000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 2038638, "num_examples": 10000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 4062189, "num_examples": 20000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 8117683, "size_in_bytes": 242126567}, "PAN-X.pt": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.pt", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 1575141, "num_examples": 10000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 1562625, "num_examples": 10000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 3149283, "num_examples": 20000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 6287049, "size_in_bytes": 240295933}, "PAN-X.ru": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.ru", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 2053169, "num_examples": 10000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 2074145, "num_examples": 10000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 4121791, "num_examples": 20000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 8249105, "size_in_bytes": 242257989}, "PAN-X.sw": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.sw", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 136368, "num_examples": 1000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 140231, "num_examples": 1000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 135911, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 412510, "size_in_bytes": 234421394}, "PAN-X.ta": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.ta", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 277625, "num_examples": 1000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 278114, "num_examples": 1000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 4122130, "num_examples": 15000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 4677869, "size_in_bytes": 238686753}, "PAN-X.te": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.te", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 293281, "num_examples": 1000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 296963, "num_examples": 1000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 295410, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 885654, "size_in_bytes": 234894538}, "PAN-X.th": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.th", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 13262737, "num_examples": 10000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 13586928, "num_examples": 10000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 27133029, "num_examples": 20000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 53982694, "size_in_bytes": 287991578}, "PAN-X.tl": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.tl", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 114156, "num_examples": 1000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 117904, "num_examples": 1000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 1168717, "num_examples": 10000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 1400777, "size_in_bytes": 235409661}, "PAN-X.tr": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.tr", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 1915352, "num_examples": 10000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 1911503, "num_examples": 10000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 3779170, "num_examples": 20000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 7606025, "size_in_bytes": 241614909}, "PAN-X.ur": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.ur", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 152148, "num_examples": 1000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 151922, "num_examples": 1000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 3072276, "num_examples": 20000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 3376346, "size_in_bytes": 237385230}, "PAN-X.vi": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.vi", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 1565143, "num_examples": 10000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 1580216, "num_examples": 10000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 3153227, "num_examples": 20000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 6298586, "size_in_bytes": 240307470}, "PAN-X.yo": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.yo", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 13245, "num_examples": 100, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 13533, "num_examples": 100, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 14709, "num_examples": 100, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 41487, "size_in_bytes": 234050371}, "PAN-X.zh": {"description": "The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been\nconstructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset\ncan be loaded with the DaNLP package:\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{pan-x,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},\n  volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}\n  year={2017}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/afshinrahimi/mmner", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "ner_tags": {"feature": {"num_classes": 7, "names": ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}, "langs": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAN-X.zh", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 4491325, "num_examples": 10000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 4363172, "num_examples": 10000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 8832051, "num_examples": 20000, "dataset_name": "xtreme"}}, "download_checksums": {"https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip": {"num_bytes": 234008884, "checksum": "164814b64f749ad000988b5ab45050b5116913b3db466ab173008955ddf649a4"}}, "download_size": 234008884, "post_processing_size": null, "dataset_size": 17686548, "size_in_bytes": 251695432}, "MLQA.ar.ar": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.ar.ar", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 8368114, "num_examples": 5335, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 824108, "num_examples": 517, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 9192222, "size_in_bytes": 84911272}, "MLQA.ar.de": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.ar.de", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 2183942, "num_examples": 1649, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 364837, "num_examples": 207, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 2548779, "size_in_bytes": 78267829}, "MLQA.ar.vi": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.ar.vi", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 3290629, "num_examples": 2047, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 288446, "num_examples": 163, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 3579075, "size_in_bytes": 79298125}, "MLQA.ar.zh": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.ar.zh", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 3229872, "num_examples": 1912, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 340049, "num_examples": 188, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 3569921, "size_in_bytes": 79288971}, "MLQA.ar.en": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.ar.en", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 8225662, "num_examples": 5335, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 810089, "num_examples": 517, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 9035751, "size_in_bytes": 84754801}, "MLQA.ar.es": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.ar.es", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 3041378, "num_examples": 1978, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 228180, "num_examples": 161, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 3269558, "size_in_bytes": 78988608}, "MLQA.ar.hi": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.ar.hi", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 3039396, "num_examples": 1831, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 281770, "num_examples": 186, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 3321166, "size_in_bytes": 79040216}, "MLQA.de.ar": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.de.ar", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 1620006, "num_examples": 1649, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 200174, "num_examples": 207, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 1820180, "size_in_bytes": 77539230}, "MLQA.de.de": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.de.de", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 4366102, "num_examples": 4517, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 488367, "num_examples": 512, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 4854469, "size_in_bytes": 80573519}, "MLQA.de.vi": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.de.vi", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 1688483, "num_examples": 1675, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 216075, "num_examples": 182, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 1904558, "size_in_bytes": 77623608}, "MLQA.de.zh": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.de.zh", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 1679180, "num_examples": 1621, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 184318, "num_examples": 190, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 1863498, "size_in_bytes": 77582548}, "MLQA.de.en": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.de.en", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 4343144, "num_examples": 4517, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 485894, "num_examples": 512, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 4829038, "size_in_bytes": 80548088}, "MLQA.de.es": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.de.es", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 1716615, "num_examples": 1776, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 170582, "num_examples": 196, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 1887197, "size_in_bytes": 77606247}, "MLQA.de.hi": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.de.hi", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 1371074, "num_examples": 1430, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 153871, "num_examples": 163, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 1524945, "size_in_bytes": 77243995}, "MLQA.vi.ar": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.vi.ar", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 3205185, "num_examples": 2047, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 230335, "num_examples": 163, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 3435520, "size_in_bytes": 79154570}, "MLQA.vi.de": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.vi.de", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 2227033, "num_examples": 1675, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 277185, "num_examples": 182, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 2504218, "size_in_bytes": 78223268}, "MLQA.vi.vi": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.vi.vi", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 7922085, "num_examples": 5495, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 726518, "num_examples": 511, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 8648603, "size_in_bytes": 84367653}, "MLQA.vi.zh": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.vi.zh", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 2989660, "num_examples": 1943, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 269389, "num_examples": 184, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 3259049, "size_in_bytes": 78978099}, "MLQA.vi.en": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.vi.en", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 7843431, "num_examples": 5495, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 719273, "num_examples": 511, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 8562704, "size_in_bytes": 84281754}, "MLQA.vi.es": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.vi.es", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 2866597, "num_examples": 2018, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 283461, "num_examples": 189, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 3150058, "size_in_bytes": 78869108}, "MLQA.vi.hi": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.vi.hi", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 2776664, "num_examples": 1947, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 255007, "num_examples": 177, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 3031671, "size_in_bytes": 78750721}, "MLQA.zh.ar": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.zh.ar", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 1731483, "num_examples": 1912, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 175349, "num_examples": 188, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 1906832, "size_in_bytes": 77625882}, "MLQA.zh.de": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.zh.de", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 1390018, "num_examples": 1621, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 174605, "num_examples": 190, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 1564623, "size_in_bytes": 77283673}, "MLQA.zh.vi": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.zh.vi", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 1806186, "num_examples": 1943, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 172934, "num_examples": 184, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 1979120, "size_in_bytes": 77698170}, "MLQA.zh.zh": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.zh.zh", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 4422350, "num_examples": 5137, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 443810, "num_examples": 504, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 4866160, "size_in_bytes": 80585210}, "MLQA.zh.en": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.zh.en", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 4450985, "num_examples": 5137, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 446868, "num_examples": 504, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 4897853, "size_in_bytes": 80616903}, "MLQA.zh.es": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.zh.es", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 1736283, "num_examples": 1947, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 138073, "num_examples": 161, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 1874356, "size_in_bytes": 77593406}, "MLQA.zh.hi": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.zh.hi", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 1578219, "num_examples": 1767, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 184401, "num_examples": 189, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 1762620, "size_in_bytes": 77481670}, "MLQA.en.ar": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.en.ar", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 6739219, "num_examples": 5335, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 630843, "num_examples": 517, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 7370062, "size_in_bytes": 83089112}, "MLQA.en.de": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.en.de", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 5056722, "num_examples": 4517, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 594936, "num_examples": 512, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 5651658, "size_in_bytes": 81370708}, "MLQA.en.vi": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.en.vi", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 7056698, "num_examples": 5495, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 640646, "num_examples": 511, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 7697344, "size_in_bytes": 83416394}, "MLQA.en.zh": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.en.zh", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 6539307, "num_examples": 5137, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 608444, "num_examples": 504, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 7147751, "size_in_bytes": 82866801}, "MLQA.en.en": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.en.en", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 14004648, "num_examples": 11590, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 1329112, "num_examples": 1148, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 15333760, "size_in_bytes": 91052810}, "MLQA.en.es": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.en.es", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 6179249, "num_examples": 5253, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 555462, "num_examples": 500, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 6734711, "size_in_bytes": 82453761}, "MLQA.en.hi": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.en.hi", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 6378866, "num_examples": 4918, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 623171, "num_examples": 507, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 7002037, "size_in_bytes": 82721087}, "MLQA.es.ar": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.es.ar", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 1740282, "num_examples": 1978, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 148649, "num_examples": 161, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 1888931, "size_in_bytes": 77607981}, "MLQA.es.de": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.es.de", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 1404025, "num_examples": 1776, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 144186, "num_examples": 196, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 1548211, "size_in_bytes": 77267261}, "MLQA.es.vi": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.es.vi", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 1747969, "num_examples": 2018, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 176841, "num_examples": 189, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 1924810, "size_in_bytes": 77643860}, "MLQA.es.zh": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.es.zh", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 1678451, "num_examples": 1947, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 126646, "num_examples": 161, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 1805097, "size_in_bytes": 77524147}, "MLQA.es.en": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.es.en", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 4362737, "num_examples": 5253, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 419068, "num_examples": 500, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 4781805, "size_in_bytes": 80500855}, "MLQA.es.es": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.es.es", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 4394333, "num_examples": 5253, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 422071, "num_examples": 500, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 4816404, "size_in_bytes": 80535454}, "MLQA.es.hi": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.es.hi", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 1523523, "num_examples": 1723, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 181834, "num_examples": 187, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 1705357, "size_in_bytes": 77424407}, "MLQA.hi.ar": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.hi.ar", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 4445589, "num_examples": 1831, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 410424, "num_examples": 186, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 4856013, "size_in_bytes": 80575063}, "MLQA.hi.de": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.hi.de", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 3022864, "num_examples": 1430, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 301713, "num_examples": 163, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 3324577, "size_in_bytes": 79043627}, "MLQA.hi.vi": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.hi.vi", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 4743484, "num_examples": 1947, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 419106, "num_examples": 177, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 5162590, "size_in_bytes": 80881640}, "MLQA.hi.zh": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.hi.zh", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 4354875, "num_examples": 1767, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 424246, "num_examples": 189, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 4779121, "size_in_bytes": 80498171}, "MLQA.hi.en": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.hi.en", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 11449261, "num_examples": 4918, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 1097857, "num_examples": 507, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 12547118, "size_in_bytes": 88266168}, "MLQA.hi.es": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.hi.es", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 3862229, "num_examples": 1723, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 420402, "num_examples": 187, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 4282631, "size_in_bytes": 80001681}, "MLQA.hi.hi": {"description": "      MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.\nMLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,\nGerman, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between\n4 different languages on average.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "@article{lewis2019mlqa,\ntitle={MLQA: Evaluating Cross-lingual Extractive Question Answering},\nauthor={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},\njournal={arXiv preprint arXiv:1910.07475},\nyear={2019}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/MLQA", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "MLQA.hi.hi", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 11810475, "num_examples": 4918, "dataset_name": "xtreme"}, "validation": {"name": "validation", "num_bytes": 1136784, "num_examples": 507, "dataset_name": "xtreme"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip": {"num_bytes": 75719050, "checksum": "246e8089933d13007fe80684d5c5c0713d6834cf8b3b4a0ec7c66f0a0d2baac8"}}, "download_size": 75719050, "post_processing_size": null, "dataset_size": 12947259, "size_in_bytes": 88666309}, "XQuAD.ar": {"description": "XQuAD (Cross-lingual Question Answering Dataset) is a benchmark dataset for evaluating cross-lingual question\nanswering performance. The dataset consists of a subset of 240 paragraphs and 1190 question-answer pairs from\nthe development set of SQuAD v1.1 (Rajpurkar et al., 2016) together with their professional translations into\nten languages: Spanish, German, Greek, Russian, Turkish, Arabic, Vietnamese, Thai, Chinese, and Hindi. Consequently,\nthe dataset is entirely parallel across 11 languages.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n  @article{Artetxe:etal:2019,\n      author    = {Mikel Artetxe and Sebastian Ruder and Dani Yogatama},\n      title     = {On the cross-lingual transferability of monolingual representations},\n      journal   = {CoRR},\n      volume    = {abs/1910.11856},\n      year      = {2019},\n      archivePrefix = {arXiv},\n      eprint    = {1910.11856}\n}\n\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/deepmind/xquad", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "XQuAD.ar", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 1722799, "num_examples": 1190, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/deepmind/xquad/raw/master/xquad.ar.json": {"num_bytes": 1582988, "checksum": "abdabd7afed5c635d99cca0f3f0d0c9d9ed0bc77451e963c2e4e0638c29e486d"}}, "download_size": 1582988, "post_processing_size": null, "dataset_size": 1722799, "size_in_bytes": 3305787}, "XQuAD.de": {"description": "XQuAD (Cross-lingual Question Answering Dataset) is a benchmark dataset for evaluating cross-lingual question\nanswering performance. The dataset consists of a subset of 240 paragraphs and 1190 question-answer pairs from\nthe development set of SQuAD v1.1 (Rajpurkar et al., 2016) together with their professional translations into\nten languages: Spanish, German, Greek, Russian, Turkish, Arabic, Vietnamese, Thai, Chinese, and Hindi. Consequently,\nthe dataset is entirely parallel across 11 languages.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n  @article{Artetxe:etal:2019,\n      author    = {Mikel Artetxe and Sebastian Ruder and Dani Yogatama},\n      title     = {On the cross-lingual transferability of monolingual representations},\n      journal   = {CoRR},\n      volume    = {abs/1910.11856},\n      year      = {2019},\n      archivePrefix = {arXiv},\n      eprint    = {1910.11856}\n}\n\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/deepmind/xquad", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "XQuAD.de", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 1283301, "num_examples": 1190, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/deepmind/xquad/raw/master/xquad.de.json": {"num_bytes": 669810, "checksum": "990b5d746746ed65ed4702ea5f35f99ffa4e2f1c390c07d003642acd937916f9"}}, "download_size": 669810, "post_processing_size": null, "dataset_size": 1283301, "size_in_bytes": 1953111}, "XQuAD.vi": {"description": "XQuAD (Cross-lingual Question Answering Dataset) is a benchmark dataset for evaluating cross-lingual question\nanswering performance. The dataset consists of a subset of 240 paragraphs and 1190 question-answer pairs from\nthe development set of SQuAD v1.1 (Rajpurkar et al., 2016) together with their professional translations into\nten languages: Spanish, German, Greek, Russian, Turkish, Arabic, Vietnamese, Thai, Chinese, and Hindi. Consequently,\nthe dataset is entirely parallel across 11 languages.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n  @article{Artetxe:etal:2019,\n      author    = {Mikel Artetxe and Sebastian Ruder and Dani Yogatama},\n      title     = {On the cross-lingual transferability of monolingual representations},\n      journal   = {CoRR},\n      volume    = {abs/1910.11856},\n      year      = {2019},\n      archivePrefix = {arXiv},\n      eprint    = {1910.11856}\n}\n\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/deepmind/xquad", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "XQuAD.vi", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 1477239, "num_examples": 1190, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/deepmind/xquad/raw/master/xquad.vi.json": {"num_bytes": 911401, "checksum": "f619a1eb11fb42d3ab0834259e488a65f585447ef6154437bfb7199d85161a04"}}, "download_size": 911401, "post_processing_size": null, "dataset_size": 1477239, "size_in_bytes": 2388640}, "XQuAD.zh": {"description": "XQuAD (Cross-lingual Question Answering Dataset) is a benchmark dataset for evaluating cross-lingual question\nanswering performance. The dataset consists of a subset of 240 paragraphs and 1190 question-answer pairs from\nthe development set of SQuAD v1.1 (Rajpurkar et al., 2016) together with their professional translations into\nten languages: Spanish, German, Greek, Russian, Turkish, Arabic, Vietnamese, Thai, Chinese, and Hindi. Consequently,\nthe dataset is entirely parallel across 11 languages.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n  @article{Artetxe:etal:2019,\n      author    = {Mikel Artetxe and Sebastian Ruder and Dani Yogatama},\n      title     = {On the cross-lingual transferability of monolingual representations},\n      journal   = {CoRR},\n      volume    = {abs/1910.11856},\n      year      = {2019},\n      archivePrefix = {arXiv},\n      eprint    = {1910.11856}\n}\n\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/deepmind/xquad", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "XQuAD.zh", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 984241, "num_examples": 1190, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/deepmind/xquad/raw/master/xquad.zh.json": {"num_bytes": 808652, "checksum": "691d0b3359bc6b8faa8de931dfdfe21d50a65861ae348e32a0d1a0190b0c8835"}}, "download_size": 808652, "post_processing_size": null, "dataset_size": 984241, "size_in_bytes": 1792893}, "XQuAD.en": {"description": "XQuAD (Cross-lingual Question Answering Dataset) is a benchmark dataset for evaluating cross-lingual question\nanswering performance. The dataset consists of a subset of 240 paragraphs and 1190 question-answer pairs from\nthe development set of SQuAD v1.1 (Rajpurkar et al., 2016) together with their professional translations into\nten languages: Spanish, German, Greek, Russian, Turkish, Arabic, Vietnamese, Thai, Chinese, and Hindi. Consequently,\nthe dataset is entirely parallel across 11 languages.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n  @article{Artetxe:etal:2019,\n      author    = {Mikel Artetxe and Sebastian Ruder and Dani Yogatama},\n      title     = {On the cross-lingual transferability of monolingual representations},\n      journal   = {CoRR},\n      volume    = {abs/1910.11856},\n      year      = {2019},\n      archivePrefix = {arXiv},\n      eprint    = {1910.11856}\n}\n\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/deepmind/xquad", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "XQuAD.en", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 1116123, "num_examples": 1190, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/deepmind/xquad/raw/master/xquad.en.json": {"num_bytes": 609383, "checksum": "e4c57d1c9143aaa1c5d265ba5987a65f4e69528d2a98f29d6e75019b10344f29"}}, "download_size": 609383, "post_processing_size": null, "dataset_size": 1116123, "size_in_bytes": 1725506}, "XQuAD.es": {"description": "XQuAD (Cross-lingual Question Answering Dataset) is a benchmark dataset for evaluating cross-lingual question\nanswering performance. The dataset consists of a subset of 240 paragraphs and 1190 question-answer pairs from\nthe development set of SQuAD v1.1 (Rajpurkar et al., 2016) together with their professional translations into\nten languages: Spanish, German, Greek, Russian, Turkish, Arabic, Vietnamese, Thai, Chinese, and Hindi. Consequently,\nthe dataset is entirely parallel across 11 languages.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n  @article{Artetxe:etal:2019,\n      author    = {Mikel Artetxe and Sebastian Ruder and Dani Yogatama},\n      title     = {On the cross-lingual transferability of monolingual representations},\n      journal   = {CoRR},\n      volume    = {abs/1910.11856},\n      year      = {2019},\n      archivePrefix = {arXiv},\n      eprint    = {1910.11856}\n}\n\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/deepmind/xquad", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "XQuAD.es", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 1273499, "num_examples": 1190, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/deepmind/xquad/raw/master/xquad.es.json": {"num_bytes": 684322, "checksum": "dcbae93ec3a9f4b9e78fd834a171d6f96c1a875e10e15b7530b7e4ef4971e37e"}}, "download_size": 684322, "post_processing_size": null, "dataset_size": 1273499, "size_in_bytes": 1957821}, "XQuAD.hi": {"description": "XQuAD (Cross-lingual Question Answering Dataset) is a benchmark dataset for evaluating cross-lingual question\nanswering performance. The dataset consists of a subset of 240 paragraphs and 1190 question-answer pairs from\nthe development set of SQuAD v1.1 (Rajpurkar et al., 2016) together with their professional translations into\nten languages: Spanish, German, Greek, Russian, Turkish, Arabic, Vietnamese, Thai, Chinese, and Hindi. Consequently,\nthe dataset is entirely parallel across 11 languages.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n  @article{Artetxe:etal:2019,\n      author    = {Mikel Artetxe and Sebastian Ruder and Dani Yogatama},\n      title     = {On the cross-lingual transferability of monolingual representations},\n      journal   = {CoRR},\n      volume    = {abs/1910.11856},\n      year      = {2019},\n      archivePrefix = {arXiv},\n      eprint    = {1910.11856}\n}\n\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/deepmind/xquad", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "XQuAD.hi", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 2682975, "num_examples": 1190, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/deepmind/xquad/raw/master/xquad.hi.json": {"num_bytes": 1680538, "checksum": "df2cce3532b37e9beb8979704b5c9a4bf874358f105395a298b89427b43b9d24"}}, "download_size": 1680538, "post_processing_size": null, "dataset_size": 2682975, "size_in_bytes": 4363513}, "XQuAD.el": {"description": "XQuAD (Cross-lingual Question Answering Dataset) is a benchmark dataset for evaluating cross-lingual question\nanswering performance. The dataset consists of a subset of 240 paragraphs and 1190 question-answer pairs from\nthe development set of SQuAD v1.1 (Rajpurkar et al., 2016) together with their professional translations into\nten languages: Spanish, German, Greek, Russian, Turkish, Arabic, Vietnamese, Thai, Chinese, and Hindi. Consequently,\nthe dataset is entirely parallel across 11 languages.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n  @article{Artetxe:etal:2019,\n      author    = {Mikel Artetxe and Sebastian Ruder and Dani Yogatama},\n      title     = {On the cross-lingual transferability of monolingual representations},\n      journal   = {CoRR},\n      volume    = {abs/1910.11856},\n      year      = {2019},\n      archivePrefix = {arXiv},\n      eprint    = {1910.11856}\n}\n\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/deepmind/xquad", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "XQuAD.el", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 2206690, "num_examples": 1190, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/deepmind/xquad/raw/master/xquad.el.json": {"num_bytes": 1918889, "checksum": "821cf0f88e73fa258fd2f548b19b6ec39f7025059e16f6f9fc8cd797c9c3663e"}}, "download_size": 1918889, "post_processing_size": null, "dataset_size": 2206690, "size_in_bytes": 4125579}, "XQuAD.ru": {"description": "XQuAD (Cross-lingual Question Answering Dataset) is a benchmark dataset for evaluating cross-lingual question\nanswering performance. The dataset consists of a subset of 240 paragraphs and 1190 question-answer pairs from\nthe development set of SQuAD v1.1 (Rajpurkar et al., 2016) together with their professional translations into\nten languages: Spanish, German, Greek, Russian, Turkish, Arabic, Vietnamese, Thai, Chinese, and Hindi. Consequently,\nthe dataset is entirely parallel across 11 languages.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n  @article{Artetxe:etal:2019,\n      author    = {Mikel Artetxe and Sebastian Ruder and Dani Yogatama},\n      title     = {On the cross-lingual transferability of monolingual representations},\n      journal   = {CoRR},\n      volume    = {abs/1910.11856},\n      year      = {2019},\n      archivePrefix = {arXiv},\n      eprint    = {1910.11856}\n}\n\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/deepmind/xquad", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "XQuAD.ru", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 2136990, "num_examples": 1190, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/deepmind/xquad/raw/master/xquad.ru.json": {"num_bytes": 1896368, "checksum": "208d5b1aa154c52b1b5c5eda16281e455e8fd198cdb9af3f469f0d6037d973bf"}}, "download_size": 1896368, "post_processing_size": null, "dataset_size": 2136990, "size_in_bytes": 4033358}, "XQuAD.th": {"description": "XQuAD (Cross-lingual Question Answering Dataset) is a benchmark dataset for evaluating cross-lingual question\nanswering performance. The dataset consists of a subset of 240 paragraphs and 1190 question-answer pairs from\nthe development set of SQuAD v1.1 (Rajpurkar et al., 2016) together with their professional translations into\nten languages: Spanish, German, Greek, Russian, Turkish, Arabic, Vietnamese, Thai, Chinese, and Hindi. Consequently,\nthe dataset is entirely parallel across 11 languages.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n  @article{Artetxe:etal:2019,\n      author    = {Mikel Artetxe and Sebastian Ruder and Dani Yogatama},\n      title     = {On the cross-lingual transferability of monolingual representations},\n      journal   = {CoRR},\n      volume    = {abs/1910.11856},\n      year      = {2019},\n      archivePrefix = {arXiv},\n      eprint    = {1910.11856}\n}\n\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/deepmind/xquad", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "XQuAD.th", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 2854959, "num_examples": 1190, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/deepmind/xquad/raw/master/xquad.th.json": {"num_bytes": 1809143, "checksum": "5cdda11d0e1e075f7872abf4e6ae830388ce7f617964d542308e9ae4257e0f43"}}, "download_size": 1809143, "post_processing_size": null, "dataset_size": 2854959, "size_in_bytes": 4664102}, "XQuAD.tr": {"description": "XQuAD (Cross-lingual Question Answering Dataset) is a benchmark dataset for evaluating cross-lingual question\nanswering performance. The dataset consists of a subset of 240 paragraphs and 1190 question-answer pairs from\nthe development set of SQuAD v1.1 (Rajpurkar et al., 2016) together with their professional translations into\nten languages: Spanish, German, Greek, Russian, Turkish, Arabic, Vietnamese, Thai, Chinese, and Hindi. Consequently,\nthe dataset is entirely parallel across 11 languages.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n  @article{Artetxe:etal:2019,\n      author    = {Mikel Artetxe and Sebastian Ruder and Dani Yogatama},\n      title     = {On the cross-lingual transferability of monolingual representations},\n      journal   = {CoRR},\n      volume    = {abs/1910.11856},\n      year      = {2019},\n      archivePrefix = {arXiv},\n      eprint    = {1910.11856}\n}\n\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/deepmind/xquad", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"answer_start": {"dtype": "int32", "id": null, "_type": "Value"}, "text": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "XQuAD.tr", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 1210763, "num_examples": 1190, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/deepmind/xquad/raw/master/xquad.tr.json": {"num_bytes": 729506, "checksum": "92179a564774b7696100d144c1e10870d0a966b6fccbdd254a65b9d2ab1971cc"}}, "download_size": 729506, "post_processing_size": null, "dataset_size": 1210763, "size_in_bytes": 1940269}, "bucc18.de": {"description": "Building and Using Comparable Corpora\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://comparable.limsi.fr/bucc2018/", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "bucc18.de", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 248707, "num_examples": 1038, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 2325701, "num_examples": 9580, "dataset_name": "xtreme"}}, "download_checksums": {"https://comparable.limsi.fr/bucc2018/bucc2018-de-en.training-gold.tar.bz2": {"num_bytes": 28189548, "checksum": "766e0fdebbd1438fb87c21254828eb13c8b997d8fbab002103dd060dcac50c5c"}, "https://comparable.limsi.fr/bucc2018/bucc2018-de-en.sample-gold.tar.bz2": {"num_bytes": 2529652, "checksum": "8e16ba8b02ef8d648b06adfbd6dfb188f43524e18f97b2b12a14a086caac62f3"}}, "download_size": 30719200, "post_processing_size": null, "dataset_size": 2574408, "size_in_bytes": 33293608}, "bucc18.fr": {"description": "Building and Using Comparable Corpora\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://comparable.limsi.fr/bucc2018/", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "bucc18.fr", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 212513, "num_examples": 929, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 2082419, "num_examples": 9086, "dataset_name": "xtreme"}}, "download_checksums": {"https://comparable.limsi.fr/bucc2018/bucc2018-fr-en.training-gold.tar.bz2": {"num_bytes": 20757713, "checksum": "53c0d0e7dd97dc89593a2db25a26c5f0ccdc7113e8451263bb0c80e4c4c7dc30"}, "https://comparable.limsi.fr/bucc2018/bucc2018-fr-en.sample-gold.tar.bz2": {"num_bytes": 1948831, "checksum": "df9eb3966954e163c9264076f7c9c1eb56d9d8a91855f9d3afbf2c0fdaef0a08"}}, "download_size": 22706544, "post_processing_size": null, "dataset_size": 2294932, "size_in_bytes": 25001476}, "bucc18.zh": {"description": "Building and Using Comparable Corpora\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://comparable.limsi.fr/bucc2018/", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "bucc18.zh", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 55739, "num_examples": 257, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 415925, "num_examples": 1899, "dataset_name": "xtreme"}}, "download_checksums": {"https://comparable.limsi.fr/bucc2018/bucc2018-zh-en.training-gold.tar.bz2": {"num_bytes": 6344925, "checksum": "3facb71798277c8f44dc78c1f8ae2110f254d0e14799f3508eedd54b4236877a"}, "https://comparable.limsi.fr/bucc2018/bucc2018-zh-en.sample-gold.tar.bz2": {"num_bytes": 769869, "checksum": "a3425be5c0320ee131a0927b66c3e29befb3b481ebf1b87257e660e514bc16ac"}}, "download_size": 7114794, "post_processing_size": null, "dataset_size": 471664, "size_in_bytes": 7586458}, "bucc18.ru": {"description": "Building and Using Comparable Corpora\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://comparable.limsi.fr/bucc2018/", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "bucc18.ru", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 761347, "num_examples": 2374, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 4641678, "num_examples": 14435, "dataset_name": "xtreme"}}, "download_checksums": {"https://comparable.limsi.fr/bucc2018/bucc2018-ru-en.training-gold.tar.bz2": {"num_bytes": 37085079, "checksum": "1895df56e936ca3d4f5b12299ceffe0b7ff4806584c40bdaa3ae1d445f25afa5"}, "https://comparable.limsi.fr/bucc2018/bucc2018-ru-en.sample-gold.tar.bz2": {"num_bytes": 4269233, "checksum": "fce3cabc7ee50ddb4b18aa6fb090e2669c8383d2a29fc97eed6ae70fed9a23e5"}}, "download_size": 41354312, "post_processing_size": null, "dataset_size": 5403025, "size_in_bytes": 46757337}, "PAWS-X.de": {"description": "\nThis dataset contains 23,659 human translated PAWS evaluation pairs and 296,406 machine translated training\npairs in six typologically distinct languages: French, Spanish, German, Chinese, Japanese, and Korean. All\ntranslated pairs are sourced from examples in PAWS-Wiki.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "  @InProceedings{pawsx2019emnlp,\n  title = {{PAWS-X: A Cross-lingual Adversarial Dataset for Paraphrase Identification}},\n  author = {Yang, Yinfei and Zhang, Yuan and Tar, Chris and Baldridge, Jason},\n  booktitle = {Proc. of EMNLP},\n  year = {2019}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/google-research-datasets/paws/tree/master/pawsx", "license": "", "features": {"sentence1": {"dtype": "string", "id": null, "_type": "Value"}, "sentence2": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAWS-X.de", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 500009, "num_examples": 2000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 510194, "num_examples": 2000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 12451883, "num_examples": 49380, "dataset_name": "xtreme"}}, "download_checksums": {"https://storage.googleapis.com/paws/pawsx/x-final.tar.gz": {"num_bytes": 30282057, "checksum": "4146db499101d66e68ae4c8ed3cf9dadecd625f44b7d8cf3d4a0fe93afc4fd9f"}}, "download_size": 30282057, "post_processing_size": null, "dataset_size": 13462086, "size_in_bytes": 43744143}, "PAWS-X.en": {"description": "\nThis dataset contains 23,659 human translated PAWS evaluation pairs and 296,406 machine translated training\npairs in six typologically distinct languages: French, Spanish, German, Chinese, Japanese, and Korean. All\ntranslated pairs are sourced from examples in PAWS-Wiki.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "  @InProceedings{pawsx2019emnlp,\n  title = {{PAWS-X: A Cross-lingual Adversarial Dataset for Paraphrase Identification}},\n  author = {Yang, Yinfei and Zhang, Yuan and Tar, Chris and Baldridge, Jason},\n  booktitle = {Proc. of EMNLP},\n  year = {2019}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/google-research-datasets/paws/tree/master/pawsx", "license": "", "features": {"sentence1": {"dtype": "string", "id": null, "_type": "Value"}, "sentence2": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAWS-X.en", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 478291, "num_examples": 2000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 480738, "num_examples": 2000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 11827719, "num_examples": 49175, "dataset_name": "xtreme"}}, "download_checksums": {"https://storage.googleapis.com/paws/pawsx/x-final.tar.gz": {"num_bytes": 30282057, "checksum": "4146db499101d66e68ae4c8ed3cf9dadecd625f44b7d8cf3d4a0fe93afc4fd9f"}}, "download_size": 30282057, "post_processing_size": null, "dataset_size": 12786748, "size_in_bytes": 43068805}, "PAWS-X.es": {"description": "\nThis dataset contains 23,659 human translated PAWS evaluation pairs and 296,406 machine translated training\npairs in six typologically distinct languages: French, Spanish, German, Chinese, Japanese, and Korean. All\ntranslated pairs are sourced from examples in PAWS-Wiki.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "  @InProceedings{pawsx2019emnlp,\n  title = {{PAWS-X: A Cross-lingual Adversarial Dataset for Paraphrase Identification}},\n  author = {Yang, Yinfei and Zhang, Yuan and Tar, Chris and Baldridge, Jason},\n  booktitle = {Proc. of EMNLP},\n  year = {2019}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/google-research-datasets/paws/tree/master/pawsx", "license": "", "features": {"sentence1": {"dtype": "string", "id": null, "_type": "Value"}, "sentence2": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAWS-X.es", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 494069, "num_examples": 1961, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 505047, "num_examples": 2000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 12462107, "num_examples": 49401, "dataset_name": "xtreme"}}, "download_checksums": {"https://storage.googleapis.com/paws/pawsx/x-final.tar.gz": {"num_bytes": 30282057, "checksum": "4146db499101d66e68ae4c8ed3cf9dadecd625f44b7d8cf3d4a0fe93afc4fd9f"}}, "download_size": 30282057, "post_processing_size": null, "dataset_size": 13461223, "size_in_bytes": 43743280}, "PAWS-X.fr": {"description": "\nThis dataset contains 23,659 human translated PAWS evaluation pairs and 296,406 machine translated training\npairs in six typologically distinct languages: French, Spanish, German, Chinese, Japanese, and Korean. All\ntranslated pairs are sourced from examples in PAWS-Wiki.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "  @InProceedings{pawsx2019emnlp,\n  title = {{PAWS-X: A Cross-lingual Adversarial Dataset for Paraphrase Identification}},\n  author = {Yang, Yinfei and Zhang, Yuan and Tar, Chris and Baldridge, Jason},\n  booktitle = {Proc. of EMNLP},\n  year = {2019}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/google-research-datasets/paws/tree/master/pawsx", "license": "", "features": {"sentence1": {"dtype": "string", "id": null, "_type": "Value"}, "sentence2": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAWS-X.fr", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 516111, "num_examples": 1988, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 521031, "num_examples": 2000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 12948512, "num_examples": 49399, "dataset_name": "xtreme"}}, "download_checksums": {"https://storage.googleapis.com/paws/pawsx/x-final.tar.gz": {"num_bytes": 30282057, "checksum": "4146db499101d66e68ae4c8ed3cf9dadecd625f44b7d8cf3d4a0fe93afc4fd9f"}}, "download_size": 30282057, "post_processing_size": null, "dataset_size": 13985654, "size_in_bytes": 44267711}, "PAWS-X.ja": {"description": "\nThis dataset contains 23,659 human translated PAWS evaluation pairs and 296,406 machine translated training\npairs in six typologically distinct languages: French, Spanish, German, Chinese, Japanese, and Korean. All\ntranslated pairs are sourced from examples in PAWS-Wiki.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "  @InProceedings{pawsx2019emnlp,\n  title = {{PAWS-X: A Cross-lingual Adversarial Dataset for Paraphrase Identification}},\n  author = {Yang, Yinfei and Zhang, Yuan and Tar, Chris and Baldridge, Jason},\n  booktitle = {Proc. of EMNLP},\n  year = {2019}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/google-research-datasets/paws/tree/master/pawsx", "license": "", "features": {"sentence1": {"dtype": "string", "id": null, "_type": "Value"}, "sentence2": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAWS-X.ja", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 647774, "num_examples": 2000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 654640, "num_examples": 2000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 14695653, "num_examples": 49401, "dataset_name": "xtreme"}}, "download_checksums": {"https://storage.googleapis.com/paws/pawsx/x-final.tar.gz": {"num_bytes": 30282057, "checksum": "4146db499101d66e68ae4c8ed3cf9dadecd625f44b7d8cf3d4a0fe93afc4fd9f"}}, "download_size": 30282057, "post_processing_size": null, "dataset_size": 15998067, "size_in_bytes": 46280124}, "PAWS-X.ko": {"description": "\nThis dataset contains 23,659 human translated PAWS evaluation pairs and 296,406 machine translated training\npairs in six typologically distinct languages: French, Spanish, German, Chinese, Japanese, and Korean. All\ntranslated pairs are sourced from examples in PAWS-Wiki.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "  @InProceedings{pawsx2019emnlp,\n  title = {{PAWS-X: A Cross-lingual Adversarial Dataset for Paraphrase Identification}},\n  author = {Yang, Yinfei and Zhang, Yuan and Tar, Chris and Baldridge, Jason},\n  booktitle = {Proc. of EMNLP},\n  year = {2019}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/google-research-datasets/paws/tree/master/pawsx", "license": "", "features": {"sentence1": {"dtype": "string", "id": null, "_type": "Value"}, "sentence2": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAWS-X.ko", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 540787, "num_examples": 2000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 547978, "num_examples": 1999, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 13542657, "num_examples": 49164, "dataset_name": "xtreme"}}, "download_checksums": {"https://storage.googleapis.com/paws/pawsx/x-final.tar.gz": {"num_bytes": 30282057, "checksum": "4146db499101d66e68ae4c8ed3cf9dadecd625f44b7d8cf3d4a0fe93afc4fd9f"}}, "download_size": 30282057, "post_processing_size": null, "dataset_size": 14631422, "size_in_bytes": 44913479}, "PAWS-X.zh": {"description": "\nThis dataset contains 23,659 human translated PAWS evaluation pairs and 296,406 machine translated training\npairs in six typologically distinct languages: French, Spanish, German, Chinese, Japanese, and Korean. All\ntranslated pairs are sourced from examples in PAWS-Wiki.\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "  @InProceedings{pawsx2019emnlp,\n  title = {{PAWS-X: A Cross-lingual Adversarial Dataset for Paraphrase Identification}},\n  author = {Yang, Yinfei and Zhang, Yuan and Tar, Chris and Baldridge, Jason},\n  booktitle = {Proc. of EMNLP},\n  year = {2019}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/google-research-datasets/paws/tree/master/pawsx", "license": "", "features": {"sentence1": {"dtype": "string", "id": null, "_type": "Value"}, "sentence2": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "PAWS-X.zh", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 459120, "num_examples": 2000, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 460638, "num_examples": 2000, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 10469712, "num_examples": 49401, "dataset_name": "xtreme"}}, "download_checksums": {"https://storage.googleapis.com/paws/pawsx/x-final.tar.gz": {"num_bytes": 30282057, "checksum": "4146db499101d66e68ae4c8ed3cf9dadecd625f44b7d8cf3d4a0fe93afc4fd9f"}}, "download_size": 30282057, "post_processing_size": null, "dataset_size": 11389470, "size_in_bytes": 41671527}, "tatoeba.afr": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.afr", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 179651, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.afr-eng.afr": {"num_bytes": 30586, "checksum": "7bb9e073ad8422d6bfdec7c9ebdcef8ac486e72b237200e447923a8b921a0a56"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.afr-eng.eng": {"num_bytes": 29049, "checksum": "0b700c125efb4030b4cc50d9d765d5884afc24f39296f29b028a1b2a8512034f"}}, "download_size": 59635, "post_processing_size": null, "dataset_size": 179651, "size_in_bytes": 239286}, "tatoeba.ara": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.ara", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 192666, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.ara-eng.ara": {"num_bytes": 43582, "checksum": "e67a0ae072b79cd9e8eb09f166c3bc0b23488d39f5720f2ee0a8350ae17b719f"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.ara-eng.eng": {"num_bytes": 29068, "checksum": "3c11838c963f598a52dcf2f452b666353538257001db5c59c3a5f54a999b336b"}}, "download_size": 72650, "post_processing_size": null, "dataset_size": 192666, "size_in_bytes": 265316}, "tatoeba.ben": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.ben", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 211719, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.ben-eng.ben": {"num_bytes": 65990, "checksum": "ac3385695d6a6c7e5d18e38ad4b8b7d3780f3df23dd0ff3f539071b8269a8613"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.ben-eng.eng": {"num_bytes": 25713, "checksum": "67fbe75fec549d436c3356b6d6f8dd53179b6a908661b5d507d28c7fee83350e"}}, "download_size": 91703, "post_processing_size": null, "dataset_size": 211719, "size_in_bytes": 303422}, "tatoeba.bul": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.bul", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 222295, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.bul-eng.bul": {"num_bytes": 65500, "checksum": "f9fa90cf3599d8c87f7a6ed22f5d648e3ce6687c705656a8c8ea088d891f79d5"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.bul-eng.eng": {"num_bytes": 36779, "checksum": "4b03b3b52ffb7cf4286e0c4453c90910a3021546f160bdf0e4d39d1f45bfbc0b"}}, "download_size": 102279, "post_processing_size": null, "dataset_size": 222295, "size_in_bytes": 324574}, "tatoeba.deu": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.deu", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 225583, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.deu-eng.deu": {"num_bytes": 57121, "checksum": "edfa6f75a42554df388f45891735d5e4214158a99def4b73b5908af4a3054551"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.deu-eng.eng": {"num_bytes": 48446, "checksum": "eb9cc83a42f2c4b22f310d05311207e41abb56c2a084666cac3ee0f84d2d0b84"}}, "download_size": 105567, "post_processing_size": null, "dataset_size": 225583, "size_in_bytes": 331150}, "tatoeba.cmn": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.cmn", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 188947, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.cmn-eng.cmn": {"num_bytes": 33410, "checksum": "965d033966fcd186c89741ad49ab4b0a0b2bbd33e02666635ff3b2be23c1ac1f"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.cmn-eng.eng": {"num_bytes": 35521, "checksum": "29ba36232488f5806aceccac57c59c5e750ddd08edb40eef417e3ada9ff9a239"}}, "download_size": 68931, "post_processing_size": null, "dataset_size": 188947, "size_in_bytes": 257878}, "tatoeba.ell": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.ell", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 198977, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.ell-eng.ell": {"num_bytes": 52202, "checksum": "4fb3d3d30bdafd15100dfad5c4680f8f2ed5ca87ed0a6122e2fe2aa21fee65e8"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.ell-eng.eng": {"num_bytes": 26759, "checksum": "1f630710e718e2f85a4e757d3c7f3d6e78ded0b25c99653b4c552138318d9ffe"}}, "download_size": 78961, "post_processing_size": null, "dataset_size": 198977, "size_in_bytes": 277938}, "tatoeba.est": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.est", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 179744, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.est-eng.est": {"num_bytes": 29996, "checksum": "94856999ef35e5357502d7ecf50419d0108b99270e507d9c57f8b283bd1be9c5"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.est-eng.eng": {"num_bytes": 29732, "checksum": "bb033b10596178452aecf2d97ad25580251375e7d224d8b38dad3d93d69b1e4f"}}, "download_size": 59728, "post_processing_size": null, "dataset_size": 179744, "size_in_bytes": 239472}, "tatoeba.eus": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.eus", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 186084, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.eus-eng.eus": {"num_bytes": 34060, "checksum": "4255fb70a6c268b09fcc59a9b308f0fcaaf45ef45e66fc55bf3c80eac4d8c97b"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.eus-eng.eng": {"num_bytes": 32008, "checksum": "1e80f0174ad544697fd69ddcf21287ca10c5e3cacba2fc42bf1d68c460d14ba2"}}, "download_size": 66068, "post_processing_size": null, "dataset_size": 186084, "size_in_bytes": 252152}, "tatoeba.fin": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.fin", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 195685, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.fin-eng.fin": {"num_bytes": 39857, "checksum": "8db3c734f755d578445947f1182f40faf2a9a0eca37561dd248717c088802d60"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.fin-eng.eng": {"num_bytes": 35812, "checksum": "322e610359f4d24852e673bbe4524d52c26dbf980aca0760e95c66dc21ecd504"}}, "download_size": 75669, "post_processing_size": null, "dataset_size": 195685, "size_in_bytes": 271354}, "tatoeba.fra": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.fra", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 200034, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.fra-eng.fra": {"num_bytes": 43727, "checksum": "644172ff9642fefa9e41c29b7d6f44196518e84350dc44d4992e943c0cca92b6"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.fra-eng.eng": {"num_bytes": 36291, "checksum": "5634220f8a26a9a23b84753a9aec0b0832e6bdaa9da3f83e0bd84c928c3f46e3"}}, "download_size": 80018, "post_processing_size": null, "dataset_size": 200034, "size_in_bytes": 280052}, "tatoeba.heb": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.heb", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 203516, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.heb-eng.heb": {"num_bytes": 47660, "checksum": "4a07ca4b8a6fb7ab499791573a2454730f47acbe209359d7b9372a9f6094a102"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.heb-eng.eng": {"num_bytes": 35840, "checksum": "73c27ed0f76c1d2da199230f05489749e10bb67ab879c6dfee8ca9807d6bd99c"}}, "download_size": 83500, "post_processing_size": null, "dataset_size": 203516, "size_in_bytes": 287016}, "tatoeba.hin": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.hin", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 242574, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.hin-eng.hin": {"num_bytes": 88468, "checksum": "15e4fb0a394be4438319f1d6955d1aea226e2a8c5ad38798b23b76ae43d742ed"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.hin-eng.eng": {"num_bytes": 34090, "checksum": "adfce2269a55dbac69b25c0f4f6eb89e0f9383165485925a2e042e61b9480562"}}, "download_size": 122558, "post_processing_size": null, "dataset_size": 242574, "size_in_bytes": 365132}, "tatoeba.hun": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.hun", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 188905, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.hun-eng.hun": {"num_bytes": 35335, "checksum": "56bd0682be8c1db6568313650b3310d641cc8d0019d12dd7caf201302350eeac"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.hun-eng.eng": {"num_bytes": 33554, "checksum": "5323266e91ddee67ed1ae00d6bbac0cdf3d37749d1b2da3459bf0d424bc71383"}}, "download_size": 68889, "post_processing_size": null, "dataset_size": 188905, "size_in_bytes": 257794}, "tatoeba.ind": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.ind", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 194860, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.ind-eng.ind": {"num_bytes": 39969, "checksum": "4f03cd70cba071f746eacd3ebf6b60fd5a8377ce18b4cc52edec6721f05f352a"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.ind-eng.eng": {"num_bytes": 34875, "checksum": "62a3f5127b60eb3526b8fa3994a68fa1a1f114f3a395307a8808a3517c05ffc5"}}, "download_size": 74844, "post_processing_size": null, "dataset_size": 194860, "size_in_bytes": 269704}, "tatoeba.ita": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.ita", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 185849, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.ita-eng.ita": {"num_bytes": 34237, "checksum": "9b3f369d0ed92273b46dd3b983721636e3d15024ce7125f5103229249c386d26"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.ita-eng.eng": {"num_bytes": 31596, "checksum": "738bf8f981e42d285f4f08bc09238782d285b02050e6e95287aa4e998bb7b24b"}}, "download_size": 65833, "post_processing_size": null, "dataset_size": 185849, "size_in_bytes": 251682}, "tatoeba.jav": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.jav", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 38529, "num_examples": 205, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.jav-eng.jav": {"num_bytes": 7457, "checksum": "ad88399db8f94c2a040aa53e7e862225964fac9308a3beb3d5b38f3eca2f827f"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.jav-eng.eng": {"num_bytes": 6456, "checksum": "172776353690f6c047ea21da969fa6979980d692fff1cfbac17eb25851423760"}}, "download_size": 13913, "post_processing_size": null, "dataset_size": 38529, "size_in_bytes": 52442}, "tatoeba.jpn": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.jpn", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 213099, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.jpn-eng.jpn": {"num_bytes": 53844, "checksum": "56040bd6949170a631039d9f8f4c6440db8761b0065c9686feba55c99a320d46"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.jpn-eng.eng": {"num_bytes": 39239, "checksum": "b42129b34e1bf225ccc25fc00e532a6113af98adbc6605b93021bd8aadeb68b6"}}, "download_size": 93083, "post_processing_size": null, "dataset_size": 213099, "size_in_bytes": 306182}, "tatoeba.kat": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.kat", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 161696, "num_examples": 746, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.kat-eng.kat": {"num_bytes": 50967, "checksum": "6ef69b5efbf355597ed91eb355b33a5f524bdf0875dbeaaccf6375badc20e29b"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.kat-eng.eng": {"num_bytes": 21193, "checksum": "d70a14aa64fd7c6b545f11aea754a632e1cbecb91af27fcf6a98a8449a48a8e7"}}, "download_size": 72160, "post_processing_size": null, "dataset_size": 161696, "size_in_bytes": 233856}, "tatoeba.kaz": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.kaz", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 116194, "num_examples": 575, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.kaz-eng.kaz": {"num_bytes": 29687, "checksum": "f20c682582a80b6aa10f3b933db93bc314449b554ce611e263bc75990b319aef"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.kaz-eng.eng": {"num_bytes": 17491, "checksum": "0ab684e7032c6520540d5785adf00ef206d097221d0dd4dc9bcaabd64068e10d"}}, "download_size": 47178, "post_processing_size": null, "dataset_size": 116194, "size_in_bytes": 163372}, "tatoeba.kor": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.kor", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 199155, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.kor-eng.kor": {"num_bytes": 44054, "checksum": "e550c84184ec35b1a0dab3154284719511a21746e53c40f46eb6ab08179e9188"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.kor-eng.eng": {"num_bytes": 35085, "checksum": "f900cf3c9b72ed5a400e1804702863ff3df00be58eb060902e02285d0e68fab3"}}, "download_size": 79139, "post_processing_size": null, "dataset_size": 199155, "size_in_bytes": 278294}, "tatoeba.mal": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.mal", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 177173, "num_examples": 687, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.mal-eng.mal": {"num_bytes": 72952, "checksum": "1a896f54f85a454fb0123864049c65921ae9dfd0cafda6deef8060f0104d965e"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.mal-eng.eng": {"num_bytes": 21765, "checksum": "4d80cdbb844cd4e33f874e5dc45c1cdda4f80998034448f7eb56b8b6532a6622"}}, "download_size": 94717, "post_processing_size": null, "dataset_size": 177173, "size_in_bytes": 271890}, "tatoeba.mar": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.mar", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 220558, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.mar-eng.mar": {"num_bytes": 72652, "checksum": "b2931584fbe62062beb97cc939e4d208ace5ee56f15808860ab14e130fd3c576"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.mar-eng.eng": {"num_bytes": 27890, "checksum": "709d09b697dca053c814b9d525b72cb47cb011aa860c6598f3e2b1b3dd1280dd"}}, "download_size": 100542, "post_processing_size": null, "dataset_size": 220558, "size_in_bytes": 321100}, "tatoeba.nld": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.nld", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 193279, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.nld-eng.nld": {"num_bytes": 37866, "checksum": "d564d4ce1c621ccaefdbe9f5cb08eacccc7bf2a0b58666303e84ca9c7973bdb7"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.nld-eng.eng": {"num_bytes": 35397, "checksum": "3b8836749df573a53235b85ed6771f31bf2de428f520d2d6a1dd94b61b4ef057"}}, "download_size": 73263, "post_processing_size": null, "dataset_size": 193279, "size_in_bytes": 266542}, "tatoeba.pes": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.pes", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 213735, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.pes-eng.pes": {"num_bytes": 58866, "checksum": "f1553713723491fe5876e1060b18fb4abf0c77be3ba06db2e3307e83aedbbb32"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.pes-eng.eng": {"num_bytes": 34853, "checksum": "b5c6cf8c8d93ff8f2fe26b53f3ee29b62db9c6f7dcddcb086ba48953f4ce926b"}}, "download_size": 93719, "post_processing_size": null, "dataset_size": 213735, "size_in_bytes": 307454}, "tatoeba.por": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.por", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 195201, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.por-eng.por": {"num_bytes": 39331, "checksum": "b0c926a232c9889a87a1a970f9399c5618c2d95baf204321e9da794c0aec16f5"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.por-eng.eng": {"num_bytes": 35854, "checksum": "deb4568cfb7b7cbbc060a7fe97c4639fb4680842f4fcd28df791ffdbb753855a"}}, "download_size": 75185, "post_processing_size": null, "dataset_size": 195201, "size_in_bytes": 270386}, "tatoeba.rus": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.rus", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 212488, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.rus-eng.rus": {"num_bytes": 58822, "checksum": "446ff2cae66053c2277d9735b2c2df6b786cae258385f7ade7bed68d8835c6a0"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.rus-eng.eng": {"num_bytes": 33650, "checksum": "7b26f52d6085b7c4944d6f5f6f5b6e1932085b42112f1444db515ce59e878fb8"}}, "download_size": 92472, "post_processing_size": null, "dataset_size": 212488, "size_in_bytes": 304960}, "tatoeba.spa": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.spa", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 192282, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.spa-eng.spa": {"num_bytes": 37490, "checksum": "f9628cea40481e8251f0999718bd893cff0f261752f5e526b3bc20284e2ca018"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.spa-eng.eng": {"num_bytes": 34776, "checksum": "89e4470f4572040b1ca94b3edad97dcd8bd2f0141f072e12933b8659dadf917d"}}, "download_size": 72266, "post_processing_size": null, "dataset_size": 192282, "size_in_bytes": 264548}, "tatoeba.swh": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.swh", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 67283, "num_examples": 390, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.swh-eng.swh": {"num_bytes": 9645, "checksum": "1c672915446c336cc378676e6dbf91eb54d27bbfd0c61563d349265bc6374753"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.swh-eng.eng": {"num_bytes": 10822, "checksum": "e8539647caff9e329776ae863b6224d432923c6e4e9256b9df92ca58ff282eac"}}, "download_size": 20467, "post_processing_size": null, "dataset_size": 67283, "size_in_bytes": 87750}, "tatoeba.tam": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.tam", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 76297, "num_examples": 307, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.tam-eng.tam": {"num_bytes": 30553, "checksum": "bde87fb1ddedccf6c7a2b70ffdd19a959a573c113c2e7a041c4b623fb2170bde"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.tam-eng.eng": {"num_bytes": 8888, "checksum": "e7c5106acdd100214a161970b2a5c31e7386e5b6a963e3d3afdf30412c90ac53"}}, "download_size": 39441, "post_processing_size": null, "dataset_size": 76297, "size_in_bytes": 115738}, "tatoeba.tel": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.tel", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 53239, "num_examples": 234, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.tel-eng.tel": {"num_bytes": 18337, "checksum": "7e1a1bcd106cce650a09e2f042f1354b55b29bea2bcfa86554dfa0ad12ce8976"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.tel-eng.eng": {"num_bytes": 6806, "checksum": "1efc2ef57d9b1ecebfc4baa45e86fd793e38473304e9c043aebabc3a1b29a294"}}, "download_size": 25143, "post_processing_size": null, "dataset_size": 53239, "size_in_bytes": 78382}, "tatoeba.tgl": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.tgl", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 188154, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.tgl-eng.tgl": {"num_bytes": 36506, "checksum": "f99165dc05190b99f6574fe24db884ff85d111612a25e7a37323f001aafc2a6e"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.tgl-eng.eng": {"num_bytes": 31632, "checksum": "e7c9beda3f3072a968a34a7226a66d1ebf1dcb33cf002805dc752f80a7c620ae"}}, "download_size": 68138, "post_processing_size": null, "dataset_size": 188154, "size_in_bytes": 256292}, "tatoeba.tha": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.tha", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 128974, "num_examples": 548, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.tha-eng.tha": {"num_bytes": 44759, "checksum": "65c7b3c01a56a1ac8971e72e0ea8e74a027718dc84044d8802c0ab36395a3156"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.tha-eng.eng": {"num_bytes": 18439, "checksum": "2881c82d2c5fa59cf0a68bc9e012f5e0b0a716f7357cbecf77c247efc2fd7294"}}, "download_size": 63198, "post_processing_size": null, "dataset_size": 128974, "size_in_bytes": 192172}, "tatoeba.tur": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.tur", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 191901, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.tur-eng.tur": {"num_bytes": 37607, "checksum": "1ffa0acc006018b3105abda41a4d4ca42f3c122964a49b71793546367b079a86"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.tur-eng.eng": {"num_bytes": 34278, "checksum": "a768df7dd3d1344f872a458b32c3a65e24f8381826ccb16ba6677426176c8121"}}, "download_size": 71885, "post_processing_size": null, "dataset_size": 191901, "size_in_bytes": 263786}, "tatoeba.urd": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.urd", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 208728, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.urd-eng.urd": {"num_bytes": 56819, "checksum": "2efc22dc61885a9454aeeee68c8b841c7f9138d53ba644a82308bd210140450b"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.urd-eng.eng": {"num_bytes": 31893, "checksum": "dab35fda3f73b3fd86b6b9f9f9f6242430961aa5d1ac247adbc646867df79cec"}}, "download_size": 88712, "post_processing_size": null, "dataset_size": 208728, "size_in_bytes": 297440}, "tatoeba.vie": {"description": "his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.\nFor each languages, we have selected 1000 English sentences and their translations, if available. Please check\nthis paper for a description of the languages, their families and scripts as well as baseline results.\nPlease note that the English sentences are not identical for all language pairs. This means that the results are\nnot directly comparable across languages. In particular, the sentences tend to have less variety for several\nlow-resource languages, e.g. \"Tom needed water\", \"Tom needs water\", \"Tom is getting water\", ...\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "          @article{tatoeba,\n  title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},\n  author={Mikel, Artetxe and Holger, Schwenk,},\n  journal={arXiv:1812.10464v2},\n  year={2018}\n}\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md", "license": "", "features": {"source_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "target_sentence": {"dtype": "string", "id": null, "_type": "Value"}, "source_lang": {"dtype": "string", "id": null, "_type": "Value"}, "target_lang": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "tatoeba.vie", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 211423, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.vie-eng.vie": {"num_bytes": 52721, "checksum": "6dbb02d778b0bfc8678cd85f87db76de55dd7e409a26fe32ad42d50e0f1fff77"}, "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/tatoeba.vie-eng.eng": {"num_bytes": 38686, "checksum": "a1f60bd8ae6c42224a4c050d2aa1ff4242d14827d64d7831e96ecf2b2c367f5f"}}, "download_size": 91407, "post_processing_size": null, "dataset_size": 211423, "size_in_bytes": 302830}, "udpos.Afrikaans": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Afrikaans", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 91302, "num_examples": 194, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 174256, "num_examples": 425, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 586382, "num_examples": 1315, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 851940, "size_in_bytes": 356068621}, "udpos.Arabic": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Arabic", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 593662, "num_examples": 909, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 973834, "num_examples": 1680, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 4453694, "num_examples": 6075, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 6021190, "size_in_bytes": 361237871}, "udpos.Basque": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Basque", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 438683, "num_examples": 1798, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 444656, "num_examples": 1799, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 1327725, "num_examples": 5396, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 2211064, "size_in_bytes": 357427745}, "udpos.Bulgarian": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Bulgarian", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 347129, "num_examples": 1115, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 339959, "num_examples": 1116, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 2689779, "num_examples": 8907, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 3376867, "size_in_bytes": 358593548}, "udpos.Dutch": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Dutch", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 393604, "num_examples": 1394, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 397916, "num_examples": 1471, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 4518018, "num_examples": 18051, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 5309538, "size_in_bytes": 360526219}, "udpos.English": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.English", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 1042052, "num_examples": 3974, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 1421160, "num_examples": 5440, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 6225545, "num_examples": 21253, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 8688757, "size_in_bytes": 363905438}, "udpos.Estonian": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Estonian", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 814183, "num_examples": 3125, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 1065713, "num_examples": 3760, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 6614929, "num_examples": 25749, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 8494825, "size_in_bytes": 363711506}, "udpos.Finnish": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Finnish", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 656658, "num_examples": 3239, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 1025738, "num_examples": 4422, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 5613742, "num_examples": 27198, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 7296138, "size_in_bytes": 362512819}, "udpos.French": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.French", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 1294108, "num_examples": 5979, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 1731061, "num_examples": 9465, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 10118993, "num_examples": 47308, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 13144162, "size_in_bytes": 368360843}, "udpos.German": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.German", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 6044862, "num_examples": 19233, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 7345899, "num_examples": 22458, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 54773981, "num_examples": 166849, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 68164742, "size_in_bytes": 423381423}, "udpos.Greek": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Greek", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 1062459, "num_examples": 2559, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 1028677, "num_examples": 2809, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 8932140, "num_examples": 28152, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 11023276, "size_in_bytes": 366239957}, "udpos.Hebrew": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Hebrew", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 210025, "num_examples": 484, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 223877, "num_examples": 491, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 2505703, "num_examples": 5241, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 2939605, "size_in_bytes": 358156286}, "udpos.Hindi": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Hindi", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 839714, "num_examples": 1659, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 1400237, "num_examples": 2684, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 6690274, "num_examples": 13304, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 8930225, "size_in_bytes": 364146906}, "udpos.Hungarian": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Hungarian", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 215891, "num_examples": 441, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 193740, "num_examples": 449, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 372238, "num_examples": 910, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 781869, "size_in_bytes": 355998550}, "udpos.Indonesian": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Indonesian", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 220875, "num_examples": 559, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 557113, "num_examples": 1557, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 1710690, "num_examples": 4477, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 2488678, "size_in_bytes": 357705359}, "udpos.Italian": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Italian", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 989008, "num_examples": 2278, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 1337881, "num_examples": 3518, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 11299329, "num_examples": 29685, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 13626218, "size_in_bytes": 368842899}, "udpos.Japanese": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Japanese", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 200368, "num_examples": 511, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 928914, "num_examples": 2372, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 2792963, "num_examples": 7125, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 3922245, "size_in_bytes": 359138926}, "udpos.Kazakh": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Kazakh", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 228936, "num_examples": 1047, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 11450, "num_examples": 31, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 240386, "size_in_bytes": 355457067}, "udpos.Korean": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Korean", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 782599, "num_examples": 3016, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 1162551, "num_examples": 4276, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 7341303, "num_examples": 27410, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 9286453, "size_in_bytes": 364503134}, "udpos.Chinese": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Chinese", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 594460, "num_examples": 3038, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 1236063, "num_examples": 5528, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 4218915, "num_examples": 18998, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 6049438, "size_in_bytes": 361266119}, "udpos.Marathi": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Marathi", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 8509, "num_examples": 46, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 7883, "num_examples": 47, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 59035, "num_examples": 373, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 75427, "size_in_bytes": 355292108}, "udpos.Persian": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Persian", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 317065, "num_examples": 599, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 320695, "num_examples": 600, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 2400788, "num_examples": 4798, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 3038548, "size_in_bytes": 358255229}, "udpos.Portuguese": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Portuguese", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 712409, "num_examples": 1770, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 1082594, "num_examples": 2681, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 7669580, "num_examples": 17992, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 9464583, "size_in_bytes": 364681264}, "udpos.Russian": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Russian", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 3457043, "num_examples": 9960, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 4236717, "num_examples": 11336, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 24230182, "num_examples": 67435, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 31923942, "size_in_bytes": 387140623}, "udpos.Spanish": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Spanish", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 1498777, "num_examples": 3054, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 1476512, "num_examples": 3147, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 13858442, "num_examples": 28492, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 16833731, "size_in_bytes": 372050412}, "udpos.Tagalog": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Tagalog", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 5165, "num_examples": 55, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 5165, "size_in_bytes": 355221846}, "udpos.Tamil": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Tamil", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 40043, "num_examples": 80, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 62378, "num_examples": 120, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 202608, "num_examples": 400, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 305029, "size_in_bytes": 355521710}, "udpos.Telugu": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Telugu", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 18002, "num_examples": 131, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 19587, "num_examples": 146, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 138061, "num_examples": 1051, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 175650, "size_in_bytes": 355392331}, "udpos.Thai": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Thai", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 561348, "num_examples": 1000, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 561348, "size_in_bytes": 355778029}, "udpos.Turkish": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Turkish", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 186467, "num_examples": 988, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 827394, "num_examples": 4785, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 704417, "num_examples": 3664, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 1718278, "size_in_bytes": 356934959}, "udpos.Urdu": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Urdu", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 284273, "num_examples": 552, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 288565, "num_examples": 535, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 2107374, "num_examples": 4043, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 2680212, "size_in_bytes": 357896893}, "udpos.Vietnamese": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Vietnamese", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 206200, "num_examples": 800, "dataset_name": "xtreme"}, "test": {"name": "test", "num_bytes": 214075, "num_examples": 800, "dataset_name": "xtreme"}, "train": {"name": "train", "num_bytes": 367347, "num_examples": 1400, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 787622, "size_in_bytes": 356004303}, "udpos.Yoruba": {"description": "Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological\nfeatures, and syntactic dependencies) across different human languages. UD is an open community effort with over 200\ncontributors producing more than 100 treebanks in over 70 languages. If you\u2019re new to UD, you should start by reading\nthe first part of the Short Introduction and then browsing the annotation guidelines.\n\nThe Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of\nthe cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages\n(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of\nsyntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,\nand availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil\n(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the\nNiger-Congo languages Swahili and Yoruba, spoken in Africa.\n", "citation": "\n@article{hu2020xtreme,\n      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},\n      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},\n      journal   = {CoRR},\n      volume    = {abs/2003.11080},\n      year      = {2020},\n      archivePrefix = {arXiv},\n      eprint    = {2003.11080}\n}\n", "homepage": "https://github.com/google-research/xtreme\thttps://universaldependencies.org/", "license": "", "features": {"tokens": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "pos_tags": {"feature": {"num_classes": 17, "names": ["ADJ", "ADP", "ADV", "AUX", "CCONJ", "DET", "INTJ", "NOUN", "NUM", "PART", "PRON", "PROPN", "PUNCT", "SCONJ", "SYM", "VERB", "X"], "id": null, "_type": "ClassLabel"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "xtreme", "config_name": "udpos.Yoruba", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 44668, "num_examples": 100, "dataset_name": "xtreme"}}, "download_checksums": {"https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz": {"num_bytes": 355216681, "checksum": "5ff973e44345a5f69b94cc1427158e14e851c967d58773cc2ac5a1d3adaca409"}}, "download_size": 355216681, "post_processing_size": null, "dataset_size": 44668, "size_in_bytes": 355261349}}
\ No newline at end of file
diff --git a/tatoeba.afr/xtreme-validation.parquet b/tatoeba.afr/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..93f4bff4324e4da9b10f855fecaf4621516e75de
--- /dev/null
+++ b/tatoeba.afr/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:47266fe4501fb49253684bdbab0c6048d8d36994a01f939b057fb626846a5190
+size 47578
diff --git a/tatoeba.ara/xtreme-validation.parquet b/tatoeba.ara/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..a149aeb76975dbb19572cd8b8e1627db132a63ff
--- /dev/null
+++ b/tatoeba.ara/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1e48642643204d2f7232e58424f600f065e7c8dbef8b9fc446cb838fe6ae0606
+size 51130
diff --git a/tatoeba.ben/xtreme-validation.parquet b/tatoeba.ben/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..a75a1f1d52c61f20d15a7ccc1118499a16b904c3
--- /dev/null
+++ b/tatoeba.ben/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ede9f61fb4c961be2cb0cb0464cc79df9d7e0b053abcc0f7e0eb100a9e482c21
+size 51264
diff --git a/tatoeba.bul/xtreme-validation.parquet b/tatoeba.bul/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..c96a7149f50dce2778c084fff66ffe2061d073e7
--- /dev/null
+++ b/tatoeba.bul/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dd53a52c8be4c034be4c1c5113122f0333943d6f9af505e8f2d3750c0b2f343a
+size 62356
diff --git a/tatoeba.cmn/xtreme-validation.parquet b/tatoeba.cmn/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..216cb789dd7bcf57f4bb2cade9482ef863c0af3e
--- /dev/null
+++ b/tatoeba.cmn/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6ee083188c33c8c65729b92816ca242774c3855156999a0b4299dd0b5cf1a1f4
+size 58183
diff --git a/tatoeba.deu/xtreme-validation.parquet b/tatoeba.deu/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..40f82fa9a6d42c32d2807692a110dafa1ca7f527
--- /dev/null
+++ b/tatoeba.deu/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:238e8c5354221f8b61a71ec41e62a309fb1db1bb0471d06726aca5d265d5d7bf
+size 78968
diff --git a/tatoeba.ell/xtreme-validation.parquet b/tatoeba.ell/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..ed17948445a3086eb149a77ab02f73dcbc54d802
--- /dev/null
+++ b/tatoeba.ell/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c8445b3da403e86c7cb9fef2113a8d36c6bf3045861dbd996ff912c6dc2297eb
+size 52153
diff --git a/tatoeba.est/xtreme-validation.parquet b/tatoeba.est/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..04744ce86effbc68d7ae381f8712b4db2b011c12
--- /dev/null
+++ b/tatoeba.est/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7282353451bb7b2d07cdcb85f2a71d9830395b448598370b116b226c217462a4
+size 49870
diff --git a/tatoeba.eus/xtreme-validation.parquet b/tatoeba.eus/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..d81b184b5d5b3793187db9274f2c76fe92c94f6f
--- /dev/null
+++ b/tatoeba.eus/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c308faae52d2dabb06618adb6fd03bde25def06ccdd8d505e2118c25a1bde7dd
+size 54173
diff --git a/tatoeba.fin/xtreme-validation.parquet b/tatoeba.fin/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..37825eed90e00b04db99387fd1274c59251bf890
--- /dev/null
+++ b/tatoeba.fin/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:838930b7be4e91a9d767647a6401b456f4f474c1abcba50717f526d2ff9710f2
+size 60482
diff --git a/tatoeba.fra/xtreme-validation.parquet b/tatoeba.fra/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..0de614e276be1c00e55147b862ee5304285132c3
--- /dev/null
+++ b/tatoeba.fra/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fcfae3abb23351a2863038c8c27e3e727d1d08ca445b4c6f7857945b9beb53a7
+size 60827
diff --git a/tatoeba.heb/xtreme-validation.parquet b/tatoeba.heb/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..7da0aa5d211b9d8b94242299f875fa64406983c2
--- /dev/null
+++ b/tatoeba.heb/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f64f5fed22f25685a72b1b769fef447a11b2cf649356ebbd3d2e94566c0d6b09
+size 57208
diff --git a/tatoeba.hin/xtreme-validation.parquet b/tatoeba.hin/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..6973e2cafa9ae52748d92ade3b38d8af6c0f448e
--- /dev/null
+++ b/tatoeba.hin/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2fc6f4599db2197342c0ffb69c032a857f179ce7993e91e7a361ea16c3e83e2b
+size 68718
diff --git a/tatoeba.hun/xtreme-validation.parquet b/tatoeba.hun/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..2e6814897646ee8ed113631252e0b2036a379852
--- /dev/null
+++ b/tatoeba.hun/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ab7d9c24423c029b26ed245601e1398d6bc89233ad9f138dc63c949ce10387be
+size 57998
diff --git a/tatoeba.ind/xtreme-validation.parquet b/tatoeba.ind/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..7b01bf3be5ee55216f3e88767adc0932c200ca02
--- /dev/null
+++ b/tatoeba.ind/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:86179c0f74dc060abcbd57808c9f7167cb1d0c99f7bdce7c81538e2e9d3e1d58
+size 56949
diff --git a/tatoeba.ita/xtreme-validation.parquet b/tatoeba.ita/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..f9bb33196cdc3d6b7c35197c5aa6fa047a3222fc
--- /dev/null
+++ b/tatoeba.ita/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0fc64166987b0c4e906a2a828f6d4a71e5f44e9e808a54af318f31f7dc2d1003
+size 52324
diff --git a/tatoeba.jav/xtreme-validation.parquet b/tatoeba.jav/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..ea3a337f439bc76cab6445463898814a4829eb75
--- /dev/null
+++ b/tatoeba.jav/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:78b88a97dd07491beaf5a1d2466b56a35a665710d8c16f9e128630329178edb3
+size 15110
diff --git a/tatoeba.jpn/xtreme-validation.parquet b/tatoeba.jpn/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..d894f130891c548eff86b887c4a6fcb214c3d682
--- /dev/null
+++ b/tatoeba.jpn/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f6625688b45ed9716f07f8e94e19b350ad261cb7aa5e673b81c5754bc7e9b4f3
+size 66522
diff --git a/tatoeba.kat/xtreme-validation.parquet b/tatoeba.kat/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..897c4edee378e84c0964b92e5fca854ea156880a
--- /dev/null
+++ b/tatoeba.kat/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1fec98e8fbac192f6c8f459e85e37deecef6fadcaebb0efb9ca4c6d8327163c3
+size 41661
diff --git a/tatoeba.kaz/xtreme-validation.parquet b/tatoeba.kaz/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..f007747a9e5e256df74b504a5017551f2719cc42
--- /dev/null
+++ b/tatoeba.kaz/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:01496f5e5fc7a4b9c11a898f4e7408f7d7dc0823532128f9e951300523ffcbec
+size 35595
diff --git a/tatoeba.kor/xtreme-validation.parquet b/tatoeba.kor/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..8a868833820d9a3b6e6d78e846bbee63d55dc079
--- /dev/null
+++ b/tatoeba.kor/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ac4b5e11cf35e5d81123067c88f7e7a3726256c1e6c0cae5047f67828dd18ced
+size 61112
diff --git a/tatoeba.mal/xtreme-validation.parquet b/tatoeba.mal/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..cf8a51829b62ec9d36387322ebff14220bb50c54
--- /dev/null
+++ b/tatoeba.mal/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:533ab072a9d85bfff1ce51235a273f5db16eeb8f8bb73acd02503dc5d7d64204
+size 50979
diff --git a/tatoeba.mar/xtreme-validation.parquet b/tatoeba.mar/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..3803a377c0fd892b1b954b049525c71080b67a4f
--- /dev/null
+++ b/tatoeba.mar/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cebf533f3d9dc36567eb04f2a047efe48c586663da844609522b18c7e8273f7a
+size 56477
diff --git a/tatoeba.nld/xtreme-validation.parquet b/tatoeba.nld/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..7ea1be6025a82ed941dfcf235b67b0b76e94a877
--- /dev/null
+++ b/tatoeba.nld/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:15fb3bc43fb815060ec7381a46c4e358fff208b2c6e8a143e270146b87ddd346
+size 59676
diff --git a/tatoeba.pes/xtreme-validation.parquet b/tatoeba.pes/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..d4bbed5d06304e00cefdca3155d23753687f8722
--- /dev/null
+++ b/tatoeba.pes/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b91220ccef2f69d824b83970427af8c6764a381635faa4bb2cb7da6223b49e36
+size 64544
diff --git a/tatoeba.por/xtreme-validation.parquet b/tatoeba.por/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..e83e621b4139a9ab91588d7fc7a35791b66e9aaa
--- /dev/null
+++ b/tatoeba.por/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fcb078750ebf6286c7fd11bfddd7ddb0e1881398f43820f8971f90d91961e7f1
+size 58152
diff --git a/tatoeba.rus/xtreme-validation.parquet b/tatoeba.rus/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..5a9fc463f69c875032cf166a1d493e09bdbf7e17
--- /dev/null
+++ b/tatoeba.rus/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9d787eb39f66428e3df6c4ee76b2c032f5c8cbe38eb17b7565649ca0ad3c6824
+size 61503
diff --git a/tatoeba.spa/xtreme-validation.parquet b/tatoeba.spa/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..6ab5444d4fdaf34b76870e4214af3e02ccac160e
--- /dev/null
+++ b/tatoeba.spa/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:63ae59fddef9bb1a5f1527d72f728314da17ecae6ec958f13b9287bdb68892af
+size 56957
diff --git a/tatoeba.swh/xtreme-validation.parquet b/tatoeba.swh/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..37bbd586c4ddb80194b83f32485c1d2c3c3f5af6
--- /dev/null
+++ b/tatoeba.swh/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4357bd8ab328266e4a4f7dae07780ed4d0c8461a958526ddffc52e589ccf6f08
+size 19264
diff --git a/tatoeba.tam/xtreme-validation.parquet b/tatoeba.tam/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..d807a019e50f44934bd1b83a4a55c27453581416
--- /dev/null
+++ b/tatoeba.tam/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ebaab5fa202491508269df9661a3629e19bb21ad72177747c27ecf4e890f3a3
+size 23550
diff --git a/tatoeba.tel/xtreme-validation.parquet b/tatoeba.tel/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..eb45fd45e51d0d982af9cd452695da1adcfe67c5
--- /dev/null
+++ b/tatoeba.tel/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4dc21f2f17531b10a25dd1162980ba819a09154219dd28d423dcb509f4617cf6
+size 18162
diff --git a/tatoeba.tgl/xtreme-validation.parquet b/tatoeba.tgl/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..7c56a6944c7baa16224da2feca278686cc1b8f1f
--- /dev/null
+++ b/tatoeba.tgl/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:755faef2dd6ed58c4cc17c7fad7947e72ebadbaa37decfc0a07f08b3691e280c
+size 53601
diff --git a/tatoeba.tha/xtreme-validation.parquet b/tatoeba.tha/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..a7bd6dc8639b685da736ceae2c90af52ad2df0fb
--- /dev/null
+++ b/tatoeba.tha/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6e3515b27a8bd5c213d067587fb745952485dd43ca33551a58cd5d3d85cfa853
+size 39561
diff --git a/tatoeba.tur/xtreme-validation.parquet b/tatoeba.tur/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..5d334ae04af8b5ae85b086aaa797b428ebd774de
--- /dev/null
+++ b/tatoeba.tur/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:677bbeb2074958128607d84cbf4786952095acb619de0a2af23284113fb64c02
+size 54039
diff --git a/tatoeba.urd/xtreme-validation.parquet b/tatoeba.urd/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..a77a108c7e6038b5b1490ef2f8c6e9a6907a3329
--- /dev/null
+++ b/tatoeba.urd/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0270f6024db097da8b8179b7e784af85de5d046a1cdfb72a41ca0410f2815d76
+size 60301
diff --git a/tatoeba.vie/xtreme-validation.parquet b/tatoeba.vie/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..14ed09b300bd06151cee26b529dff67471ddc1df
--- /dev/null
+++ b/tatoeba.vie/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64be55a78e9b14111e5febb38ed979921e30e3095335afee1d85e8a817bdba9a
+size 66648
diff --git a/tydiqa/xtreme-train.parquet b/tydiqa/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..f1bcedcda60fdb5d307d8219150a04f2369e3670
--- /dev/null
+++ b/tydiqa/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2e9e33c8520b1c2fbe73ce50b10a9d2b116c3dfffeb79db3da72711309914bfa
+size 26917473
diff --git a/tydiqa/xtreme-validation.parquet b/tydiqa/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..600107e3e5f08dcbf8fe0e0fe9c7ce60dfe8193d
--- /dev/null
+++ b/tydiqa/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a264ec7a1f0d299e447c474171ecccc9bd8c9625ae502feac6b82843830a3386
+size 2484103
diff --git a/udpos.Afrikaans/xtreme-test.parquet b/udpos.Afrikaans/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..34d7083a3eb292d0dfffeab8847b506bd490afa3
--- /dev/null
+++ b/udpos.Afrikaans/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0bae84207c641d1ecba44a007c343973950427c33db351176b3ea8a58f6f4436
+size 44350
diff --git a/udpos.Afrikaans/xtreme-train.parquet b/udpos.Afrikaans/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..65ecf1127bcd5312a2156a5d59fc174da820bb48
--- /dev/null
+++ b/udpos.Afrikaans/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b574d7bb858f7a9b20ee1a99c7b947d65bc4faed7bdfa23b59a19cf75c736e99
+size 125724
diff --git a/udpos.Afrikaans/xtreme-validation.parquet b/udpos.Afrikaans/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..e3243947796e00bb9f42111e42e5849708a34e68
--- /dev/null
+++ b/udpos.Afrikaans/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3860d576cee7670c3935ee7f10286d2fa5a09b4315abe561709eae38dae0521a
+size 23645
diff --git a/udpos.Arabic/xtreme-test.parquet b/udpos.Arabic/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..6e6cd6c9606dcea8c9b24d032ce6fac1f07639c7
--- /dev/null
+++ b/udpos.Arabic/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3128bfe591482a8707c71a383a17ac2efd1555928e344dcbf9a415e4525047a5
+size 214885
diff --git a/udpos.Arabic/xtreme-train.parquet b/udpos.Arabic/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..ea9ab129abc4f570d75508833581720f8f811670
--- /dev/null
+++ b/udpos.Arabic/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:59c5c1685557103eaa2440931a7ceebc00ca98974a343e61d91923f677a22900
+size 849560
diff --git a/udpos.Arabic/xtreme-validation.parquet b/udpos.Arabic/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..3aed502107fb31b5def49585749393353548f03c
--- /dev/null
+++ b/udpos.Arabic/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c4962efb453590c7a4b6f89fb4ff85a84d5111f48ddac8180c88e1f469e67708
+size 121527
diff --git a/udpos.Basque/xtreme-test.parquet b/udpos.Basque/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..f085cf9f93f47d9c638525a7afca461559cf45ab
--- /dev/null
+++ b/udpos.Basque/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:95b755504748d1edb4a54cc03d14f17d1d92e0c63deb5552f08e8fac3c8be56b
+size 141862
diff --git a/udpos.Basque/xtreme-train.parquet b/udpos.Basque/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..58d56d20da47c3708409d29fd2754947735a833d
--- /dev/null
+++ b/udpos.Basque/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d999a9711216f38e6a09c3ca3914b3348e1168718ea2c0795f9d9b9cc2a1564e
+size 421095
diff --git a/udpos.Basque/xtreme-validation.parquet b/udpos.Basque/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..8831c504d0480aae64ab291419652b642eb8d234
--- /dev/null
+++ b/udpos.Basque/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f870392ec098b6b755db7a93ef5aea1cb634639176b0a75fb27934de96c963be
+size 139996
diff --git a/udpos.Bulgarian/xtreme-test.parquet b/udpos.Bulgarian/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..c68288635315b770d2cff81a50f3775dcd289cbb
--- /dev/null
+++ b/udpos.Bulgarian/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bbcfaf834daf803c283320a8541600f7049da6e7b2606d95efb20c30aef0f83a
+size 102132
diff --git a/udpos.Bulgarian/xtreme-train.parquet b/udpos.Bulgarian/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..fd47fb668076b0fe16f8a77e0fd04b7feb9401e7
--- /dev/null
+++ b/udpos.Bulgarian/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b207fbd4813b10f96140f22db48c0ba097cea39e848e9699bbdfc41a9bc8ff2c
+size 720091
diff --git a/udpos.Bulgarian/xtreme-validation.parquet b/udpos.Bulgarian/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..4e5641ec7b51bfe94aa4a59c4b75df40bb0d911d
--- /dev/null
+++ b/udpos.Bulgarian/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7c2672109ffb2b40edb6dab794359d8c37910de435962a3d0d317b724a01b2ee
+size 103786
diff --git a/udpos.Chinese/xtreme-test.parquet b/udpos.Chinese/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..3a1d75f769b6e112b52c88ba0f0034c9aad0fac0
--- /dev/null
+++ b/udpos.Chinese/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2899a0f9e25d43ac9f225da0e16362a481a3732f37d71a0cb62d8b81c671a16d
+size 298101
diff --git a/udpos.Chinese/xtreme-train.parquet b/udpos.Chinese/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..2c790b1a6e25582b65dc0f00f8ab2cfabfb724a7
--- /dev/null
+++ b/udpos.Chinese/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7d2075c6b17259303242b801f6def1eecbe84b16539c996dac73f0c7a5373a45
+size 1014936
diff --git a/udpos.Chinese/xtreme-validation.parquet b/udpos.Chinese/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..41d4427938bb5e8b74f015e21b8d6bd023a59dd8
--- /dev/null
+++ b/udpos.Chinese/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b4c0af8dc00dbfa3e5eb0eb7f66203412dbd1b01dfbe9808547343f2ee8bcb9d
+size 158341
diff --git a/udpos.Dutch/xtreme-test.parquet b/udpos.Dutch/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..86adc5677515ee2f1feceddc3534506dfc1c2272
--- /dev/null
+++ b/udpos.Dutch/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9b2834a35314463a30ace12db7630fefa40da9366b7d6d367aa0a37213aebb03
+size 104881
diff --git a/udpos.Dutch/xtreme-train.parquet b/udpos.Dutch/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..947d98a74a1e5c3497fa6bda97c2b3a09f593cb1
--- /dev/null
+++ b/udpos.Dutch/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d24608fce0c4a79c869763c9a18cf71681ef650aaca001f4997de29340b3530a
+size 1197998
diff --git a/udpos.Dutch/xtreme-validation.parquet b/udpos.Dutch/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..c1cf47e0be9e4f17db4e1440e1c0cb4f4319f97e
--- /dev/null
+++ b/udpos.Dutch/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:29c6d33497a56be21a93dc39968a916b635df24ee0b2d55dbac4d05ab44e376c
+size 107806
diff --git a/udpos.English/xtreme-test.parquet b/udpos.English/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..f9cd3dcaa5f9a0deef1403dbd68b920cef730856
--- /dev/null
+++ b/udpos.English/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:90984b60bbb1ae57f4772a6802fc46404bdb965eb665079a58a0a6b9d9adc707
+size 366801
diff --git a/udpos.English/xtreme-train.parquet b/udpos.English/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..8382e75b91753355a5ec31ac616f1f4d0e2ab56b
--- /dev/null
+++ b/udpos.English/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1e40750f23c1a5605bf4e581a81f6732e66e28fbe72d75ba2b687f6699ece5f6
+size 1484690
diff --git a/udpos.English/xtreme-validation.parquet b/udpos.English/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..475302663573793341d1ff7d158a074cd2e82466
--- /dev/null
+++ b/udpos.English/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d2ce805245a163c9d2dc77decdd5a0c63f3f1e089e58bf72c38cd8340e1ffcd0
+size 264639
diff --git a/udpos.Estonian/xtreme-test.parquet b/udpos.Estonian/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..c79132f563c7fc27df11bc7268df0be2cffdf1cc
--- /dev/null
+++ b/udpos.Estonian/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:146a6cfb9db55007be7e108175537d4ec8e020cb484ffe926d6d35dc088a5394
+size 324033
diff --git a/udpos.Estonian/xtreme-train.parquet b/udpos.Estonian/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..d928a0bd13c759480bdf5a23b2424a768fc9e366
--- /dev/null
+++ b/udpos.Estonian/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:299bec67bd7e21e0bbdd5763c9e8fbad5b1bcd50fde214af83666444c31ab772
+size 2029387
diff --git a/udpos.Estonian/xtreme-validation.parquet b/udpos.Estonian/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..3dc25c1a439fc32ad7eeabbee19d8802f88a092f
--- /dev/null
+++ b/udpos.Estonian/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:58b6711bba14c65ca81d875fea41a2db17754dd1ae51d4b43da71261bc926fc9
+size 265272
diff --git a/udpos.Finnish/xtreme-test.parquet b/udpos.Finnish/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..ae030e8ec3de1b5b758308675a00790ffa957801
--- /dev/null
+++ b/udpos.Finnish/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1c8a835d730d736a93571a4bb62c3dd8040eb943209453bd20c7a57ffab71ae0
+size 360640
diff --git a/udpos.Finnish/xtreme-train.parquet b/udpos.Finnish/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..3b5bbe5d83eb30184c6a87f1dece89a3c5c68668
--- /dev/null
+++ b/udpos.Finnish/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:33ab6e146aa7d4cd88e7ebe2c2c13d836b822444831d69b21bb231d92c287034
+size 1905583
diff --git a/udpos.Finnish/xtreme-validation.parquet b/udpos.Finnish/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..753e18041b55fba7c821c57bed644ab9b6fc6184
--- /dev/null
+++ b/udpos.Finnish/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:343248bd6e338686cfc4f8c099341b9ae9a5d99ff844cb8389bb2f09a9b71ebb
+size 236529
diff --git a/udpos.French/xtreme-test.parquet b/udpos.French/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..7c2a7bc060540203668f4aae9d4caddbb32918c5
--- /dev/null
+++ b/udpos.French/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e3b1153a708126af5f761b272401d510c088ac975407b25c743db6ca0484ad06
+size 428077
diff --git a/udpos.French/xtreme-train.parquet b/udpos.French/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..f42a9369487cc0044446026cb296ac1770ec7022
--- /dev/null
+++ b/udpos.French/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f0d17c629371ef91cd498df7b8bc477774817ee1b8e3629cbe0981b53dc92583
+size 2610541
diff --git a/udpos.French/xtreme-validation.parquet b/udpos.French/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..11ac180f1a29db0485434ae6a44779c439e2a7c8
--- /dev/null
+++ b/udpos.French/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1069f0f3c4207250d9ce63b0f65c3c7b8593e38d47b17aa4d32773de3ddcb127
+size 339273
diff --git a/udpos.German/xtreme-test.parquet b/udpos.German/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..e42563a267a6e3eaaedec76e295ba6c091d3e285
--- /dev/null
+++ b/udpos.German/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6966759121f29185027b6161bb21121fe7ad2343364c9364249fc2a3a71d9e6c
+size 1995017
diff --git a/udpos.German/xtreme-train.parquet b/udpos.German/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..febafa3be20c5e27bee91aabac87f096eb82cc4d
--- /dev/null
+++ b/udpos.German/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3148e29d0ccd65e1a8836b1a93294796f25c46560ffb3cab72151ccd53a00471
+size 14960885
diff --git a/udpos.German/xtreme-validation.parquet b/udpos.German/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..7ba0f77b140dd9358d5cb76c9c69ac3fb3a6617a
--- /dev/null
+++ b/udpos.German/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1a670d69c59695e36d2f95aa9832ccd2cbcfdcaef2328bd749a1486f261ebafc
+size 1664712
diff --git a/udpos.Greek/xtreme-test.parquet b/udpos.Greek/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..aed4b6530229a77d32ef8ee44f5bcbff12f09fd0
--- /dev/null
+++ b/udpos.Greek/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ce0fbc296a503ab1376db6b16c68cf54bb4036b8156ae7f51c9b9518bfab29f9
+size 276071
diff --git a/udpos.Greek/xtreme-train.parquet b/udpos.Greek/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..79df4825968a43177c4df9ad44b6cb736c0626e6
--- /dev/null
+++ b/udpos.Greek/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e4f4482fc47b37043844a2de8587035b244ed4bfa625d33f1d7274ddbc43dbd5
+size 2209644
diff --git a/udpos.Greek/xtreme-validation.parquet b/udpos.Greek/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..fec1098efa7ca180cfc09460820e5aa6e8eb194e
--- /dev/null
+++ b/udpos.Greek/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:729f5f8e4a62bb247866d938beaef8d32c5a98dd13cef649fe3260805a0b32f7
+size 277137
diff --git a/udpos.Hebrew/xtreme-test.parquet b/udpos.Hebrew/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..77dce66baa1367b7865eb5a5ae6ec7d389b04130
--- /dev/null
+++ b/udpos.Hebrew/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:048727c6928b386a955765d6064d450aca0303e750e6a2ef3dde9e94dab9cd2f
+size 50797
diff --git a/udpos.Hebrew/xtreme-train.parquet b/udpos.Hebrew/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..6542ebed77262238ec1f9bc96f0f3b4e5552c0c4
--- /dev/null
+++ b/udpos.Hebrew/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ba1fcefe675b98579598134efb859b65552e8777f9a9ee75f2dd62a3948c9ebe
+size 525210
diff --git a/udpos.Hebrew/xtreme-validation.parquet b/udpos.Hebrew/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..59e97b3df0664de3bb7f3d2eb2ed857821a90e9d
--- /dev/null
+++ b/udpos.Hebrew/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a583974fabfe08e3fbfa007602899d9569b62b8b912caa742be47b13acdef5fd
+size 48647
diff --git a/udpos.Hindi/xtreme-test.parquet b/udpos.Hindi/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..c479877e7435832582b8ff9071e71cee92070a29
--- /dev/null
+++ b/udpos.Hindi/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:85a85b526342e44d69800a0ad321cf362b272b7ccb6737df850b2eb04c154fc5
+size 243958
diff --git a/udpos.Hindi/xtreme-train.parquet b/udpos.Hindi/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..54810177c75373f8962ac0c7a018a548401a1098
--- /dev/null
+++ b/udpos.Hindi/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b7ab0cdff302d26041571a8d4b9535fc74439c1a74796f2c7e5e6c34732494e2
+size 1082915
diff --git a/udpos.Hindi/xtreme-validation.parquet b/udpos.Hindi/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..9b085de0a56dace224216d7ab1f8b4a2970b4a51
--- /dev/null
+++ b/udpos.Hindi/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:03232dfd4c8767abfa2bb68f110854fd036e95363f7d2217c127347be5468ecb
+size 141192
diff --git a/udpos.Hungarian/xtreme-test.parquet b/udpos.Hungarian/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..0bf7bfd493f94feafe04cdf2c6cc3860aa372627
--- /dev/null
+++ b/udpos.Hungarian/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f41095f25ffdff04364242992f7eb01b5549e146fdff4270374a49604f8a9ee2
+size 65581
diff --git a/udpos.Hungarian/xtreme-train.parquet b/udpos.Hungarian/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..926f6699d27a22cf5aad779478be1f952c37ea9d
--- /dev/null
+++ b/udpos.Hungarian/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dd749a3066657ad98dfa0299e69a01df50b23d24013db52e78135ab3daee8f63
+size 117553
diff --git a/udpos.Hungarian/xtreme-validation.parquet b/udpos.Hungarian/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..babe700add195a981b566d783181c830dfb3dc8e
--- /dev/null
+++ b/udpos.Hungarian/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c7b1f08c8f31a8216ec67796ca332e61347a45a944473dcdaba9a8d39358bd88
+size 68691
diff --git a/udpos.Indonesian/xtreme-test.parquet b/udpos.Indonesian/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..9a3ccd416ed0acec4e896caeb9fd562f59232707
--- /dev/null
+++ b/udpos.Indonesian/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:83a27ab419836547b973101a1992307cca07b735156d162002440d96457c1d7b
+size 147689
diff --git a/udpos.Indonesian/xtreme-train.parquet b/udpos.Indonesian/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..a8fe690746845b2c3a921f1ef802bb926cb41a2f
--- /dev/null
+++ b/udpos.Indonesian/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:520ac317178b5df8b874b1b2686525777eaf79f54b26e50010e72cbe345f5b9c
+size 470108
diff --git a/udpos.Indonesian/xtreme-validation.parquet b/udpos.Indonesian/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..58d7d9f3940a752eb5926482335c3c0468584a8a
--- /dev/null
+++ b/udpos.Indonesian/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0f10a64aeb2504d5137bfb8e4408183d2d36a8584f22638c4c0714a24470b229
+size 66311
diff --git a/udpos.Italian/xtreme-test.parquet b/udpos.Italian/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..7482a0aedaabcbe7995e3a939716de79d9c88746
--- /dev/null
+++ b/udpos.Italian/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:35d3c2c30da5a6976d1a2858765d6d8807e81d92bcbcf67c5199462f877016a1
+size 343486
diff --git a/udpos.Italian/xtreme-train.parquet b/udpos.Italian/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..b6286a4383cb23a400392cb05b160fa329019c2f
--- /dev/null
+++ b/udpos.Italian/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:903fd7cec14a0a57eeba69d1d2a06f45c489c46d71000554bec9b677e30d55c5
+size 2679535
diff --git a/udpos.Italian/xtreme-validation.parquet b/udpos.Italian/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..ef573a71bd39d4d859d9084256cbaa6dca6ea6cd
--- /dev/null
+++ b/udpos.Italian/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0bec2db1c9dccbf60732312a561256f370ccee67ff09940fa7724ffbc982964b
+size 232760
diff --git a/udpos.Japanese/xtreme-test.parquet b/udpos.Japanese/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..55e9afafb740fa814fc2498de0cc9d18dcd1aee2
--- /dev/null
+++ b/udpos.Japanese/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a2cd7aab669d7c9232023e9c8c004143f28f32eaf28cebd459f8105e98b9ee11
+size 226208
diff --git a/udpos.Japanese/xtreme-train.parquet b/udpos.Japanese/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..b7a83357e97d9eb692c51baf6902e9effd63cd61
--- /dev/null
+++ b/udpos.Japanese/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e0577dc2052121a1cb27178bc7cdff428ab66c883ba6d970dd6627aef8bbb59c
+size 729297
diff --git a/udpos.Japanese/xtreme-validation.parquet b/udpos.Japanese/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..04d45c92112655468641f2bfd9cac31ac66bc986
--- /dev/null
+++ b/udpos.Japanese/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8f7fb35edb83738cf6851bcc57fe53857e0f92a151e3d5307a6edf4605dfffe9
+size 56612
diff --git a/udpos.Kazakh/xtreme-test.parquet b/udpos.Kazakh/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..c76514c9f971d154546abceac07e763956f4fb6a
--- /dev/null
+++ b/udpos.Kazakh/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e0f69d6cbcf2609924461fbd70470638070cdef113af6b3ad1659cbf61ad2b4c
+size 70229
diff --git a/udpos.Kazakh/xtreme-train.parquet b/udpos.Kazakh/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..920e425ec1345c7fbb5379424666de21d884dc1a
--- /dev/null
+++ b/udpos.Kazakh/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b6451a05988d0157b8b5a478d4b3c197db7bf904e1a843e72ac28e076a84454c
+size 6021
diff --git a/udpos.Korean/xtreme-test.parquet b/udpos.Korean/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..acc34a63c2fa061ad14adef6ef1c4f8e25489a62
--- /dev/null
+++ b/udpos.Korean/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f57cc7974afadb2fd4b6e459452e0c8cba103f77ef3ca409189fa77fd68c7e03
+size 407048
diff --git a/udpos.Korean/xtreme-train.parquet b/udpos.Korean/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..3dc91e7127bbd762b219aa01ccf75db2dcfb165c
--- /dev/null
+++ b/udpos.Korean/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2a3d479fa36c99853806942d66abc3660374b8689bec48ccd0a09e6ef59c8223
+size 2428861
diff --git a/udpos.Korean/xtreme-validation.parquet b/udpos.Korean/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..8f29a9391434b4c22c80fb0d0607f0438a857dcd
--- /dev/null
+++ b/udpos.Korean/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:28f0a1b274751fda6d3ada5b6d214cee1171798b3c09bd46a22511a35f68bec1
+size 278727
diff --git a/udpos.Marathi/xtreme-test.parquet b/udpos.Marathi/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..bd7b489221558f6a3136b4b9e1054a7439bec24b
--- /dev/null
+++ b/udpos.Marathi/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ad1cdb31e4210f4e8b58c5ea38676ab34b9d837864725f1f3b5b86485ca44380
+size 4122
diff --git a/udpos.Marathi/xtreme-train.parquet b/udpos.Marathi/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..b48d8f57e2a064293493831b1f7ca4c1807f342d
--- /dev/null
+++ b/udpos.Marathi/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9b177b1f4e2c5e431a64a65e0f50c0124fc8110a1b34c163d0f0d62125fc88b1
+size 13553
diff --git a/udpos.Marathi/xtreme-validation.parquet b/udpos.Marathi/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..730b9c70e6492620b287b575515f7a9003d424f5
--- /dev/null
+++ b/udpos.Marathi/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ef230c83fb4376b7d2f623b84b7bf8a2bec9d0ba8c7ac0068fcba9c978793c1
+size 4401
diff --git a/udpos.Persian/xtreme-test.parquet b/udpos.Persian/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..c737d3f015ec7f737cd1813e8a4d25f7afb0b201
--- /dev/null
+++ b/udpos.Persian/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b781b691dff55f01212da0ecadabee716f3884d464f37a9876b75636b9abf694
+size 67261
diff --git a/udpos.Persian/xtreme-train.parquet b/udpos.Persian/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..2b136e422cdb1c3e53dfc9717f4b8fb99aa86384
--- /dev/null
+++ b/udpos.Persian/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:78e52973578f68acedde2fa56c56f8e757901fd1fe07cc734590368d421c9610
+size 472303
diff --git a/udpos.Persian/xtreme-validation.parquet b/udpos.Persian/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..af62648953f6ff0e72c6c8bcf8fdca054443bf5c
--- /dev/null
+++ b/udpos.Persian/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:47f712f3570fe5159a8ad766c95f3fa06eef02b8720e3c9bc52cfc76a291667c
+size 67243
diff --git a/udpos.Portuguese/xtreme-test.parquet b/udpos.Portuguese/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..8dadd0822e4348c9e8adb86a2591f7570587b7e6
--- /dev/null
+++ b/udpos.Portuguese/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:12fcbb544aeca27d0d1a3318501bc53a79e546a833f587938d4119cc948a1c89
+size 292791
diff --git a/udpos.Portuguese/xtreme-train.parquet b/udpos.Portuguese/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..bad00b44c2f53ca10257588b718ef1d23a7a68b0
--- /dev/null
+++ b/udpos.Portuguese/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:07b096218adc105100186fa958c51968d2edd0a2e78a890beda0022e6e006220
+size 2018067
diff --git a/udpos.Portuguese/xtreme-validation.parquet b/udpos.Portuguese/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..32a134efc294c7667dfc2c6b1c4f5b2e494aba1e
--- /dev/null
+++ b/udpos.Portuguese/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0a98ffb6c3771407f3f8ad2d84a7969853d21766cb1b8250564e616c4281f8db
+size 194517
diff --git a/udpos.Russian/xtreme-test.parquet b/udpos.Russian/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..bc525c922297a0ef12ad7991718001e7c8e8ab41
--- /dev/null
+++ b/udpos.Russian/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eb4b294dbf7d186bbbf70b3968876ea0b989804512ab2fc193aec447528b2ec9
+size 1198017
diff --git a/udpos.Russian/xtreme-train.parquet b/udpos.Russian/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..2cbed204ee06fb32a03faef78c35962408e10404
--- /dev/null
+++ b/udpos.Russian/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05f04cc1b169487350ab7faedccded9e6a43ff0b3b21d4a9742aaeb22c76b164
+size 6651747
diff --git a/udpos.Russian/xtreme-validation.parquet b/udpos.Russian/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..a4b02347458337ba1c3cf733939a42a7f7425813
--- /dev/null
+++ b/udpos.Russian/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e96dc35baea4ce6149c831225e0afdf6db356db6309014a9b944ab73ed4c065a
+size 967647
diff --git a/udpos.Spanish/xtreme-test.parquet b/udpos.Spanish/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..84e67ded3ed39378759609587007bc1fbd7e03bc
--- /dev/null
+++ b/udpos.Spanish/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2b01beb18c5b4044d99645292a8b2c6d4439cc552267430583a22024294fa772
+size 381108
diff --git a/udpos.Spanish/xtreme-train.parquet b/udpos.Spanish/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..07fcd74bdea81a8d52dccbc1be9b34ff8faccf57
--- /dev/null
+++ b/udpos.Spanish/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0bbedaafc85537f18e8e0f8c2ab25fd308bf7ffa9249cedaa7892bba7fb13204
+size 3571662
diff --git a/udpos.Spanish/xtreme-validation.parquet b/udpos.Spanish/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..152649e6142b093c60e9b5e52b053b7536966694
--- /dev/null
+++ b/udpos.Spanish/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7617b50835ea6ba0d58b7e3f86af78d781861394b3a677bd33a745b97fc9e19f
+size 394670
diff --git a/udpos.Tagalog/xtreme-test.parquet b/udpos.Tagalog/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..3b0bf285c252ed24c41d175b457889d5860843f8
--- /dev/null
+++ b/udpos.Tagalog/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:edaf9dfe80eac6f343265ed1d66a3f65c6d1f9584f1e82efc57419f734c01959
+size 3326
diff --git a/udpos.Tamil/xtreme-test.parquet b/udpos.Tamil/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..3e91698e31ad1423efec16dfc428afdd6363c56d
--- /dev/null
+++ b/udpos.Tamil/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4b2aecdc38b8a89a90f98ec789e9d5edb94420273f5020ca8f45f29cbf4d98db
+size 16424
diff --git a/udpos.Tamil/xtreme-train.parquet b/udpos.Tamil/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..a464675b22822518d51e4139088a1084c67cb9a9
--- /dev/null
+++ b/udpos.Tamil/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dd5d34bafc905fa97de0f51ab9b18fed614792bbee84f94a580710458f188902
+size 45011
diff --git a/udpos.Tamil/xtreme-validation.parquet b/udpos.Tamil/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..da2b7929e83f0831114120103cf2b766103c40c5
--- /dev/null
+++ b/udpos.Tamil/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:adddc777f3af2d96015b7c2e0c018281aedfec60614afff78a5b8b507f3125ae
+size 12272
diff --git a/udpos.Telugu/xtreme-test.parquet b/udpos.Telugu/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..e870d78c5010471e885c0f0c369e09dd75d8b8a1
--- /dev/null
+++ b/udpos.Telugu/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64ede169d2ef7972489c5cb88c773f963d5796dff82af2aed9701161e9b73169
+size 7351
diff --git a/udpos.Telugu/xtreme-train.parquet b/udpos.Telugu/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..5d0d5236b74a440f6fb1fb0e36e46d9219195acf
--- /dev/null
+++ b/udpos.Telugu/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:025681a1ea6e769a00cb567e3d523e37449f8d4825017111c8a25a3d68e6c908
+size 31689
diff --git a/udpos.Telugu/xtreme-validation.parquet b/udpos.Telugu/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..ef0f40e215a60b466ffb12ac7f9d6079fbd910e0
--- /dev/null
+++ b/udpos.Telugu/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7a607c00fbafc3344739d78ceeb11325b257cb9275e407205db02aeaa71987e0
+size 6936
diff --git a/udpos.Thai/xtreme-test.parquet b/udpos.Thai/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..ce46ae278fca14595b49f2dab5493559e890ee28
--- /dev/null
+++ b/udpos.Thai/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4abf890b18601ee318f5d534ecf3f4aa7c72cd8f469eed63c682d96eade8ee0f
+size 92906
diff --git a/udpos.Turkish/xtreme-test.parquet b/udpos.Turkish/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..53eea2df48b4df18fca0b93a50e17bbe57710765
--- /dev/null
+++ b/udpos.Turkish/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:501816410c2aed06306d9240398f88ed898a5a172e60378a6e821621034691ce
+size 267979
diff --git a/udpos.Turkish/xtreme-train.parquet b/udpos.Turkish/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..e4ade60b506a26bd79e2177d15cc46e5f74f1438
--- /dev/null
+++ b/udpos.Turkish/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f24f9ab0a5057f95c728aafbefdfa23d9bc4944ecb24519f27eb63eed4cd083
+size 247294
diff --git a/udpos.Turkish/xtreme-validation.parquet b/udpos.Turkish/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..ef3a476673ac68a2a25d13f9d587307344f7a151
--- /dev/null
+++ b/udpos.Turkish/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1e06fe95c604368135cbcc928a50ee51b5a6629beab1d4cc99ad1f4f87b63bb1
+size 65763
diff --git a/udpos.Urdu/xtreme-test.parquet b/udpos.Urdu/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..95a146e6081a2ccc7474ca819662c9f8db854ae5
--- /dev/null
+++ b/udpos.Urdu/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5c0c573e67d815a066ac584978f8750e9eb98f35ddc875cbf6866f4538bb5bd6
+size 56663
diff --git a/udpos.Urdu/xtreme-train.parquet b/udpos.Urdu/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..32fed0c381760460d9954e7040cf06f4b911b8c2
--- /dev/null
+++ b/udpos.Urdu/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:df0b25bf752126e0d6ed1189fd9c3bb0f1738c4630100ee840469b76ab785fec
+size 386831
diff --git a/udpos.Urdu/xtreme-validation.parquet b/udpos.Urdu/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..9c88c95ec88f9ed0bae2e29012c540d4918eb47f
--- /dev/null
+++ b/udpos.Urdu/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:241d74e21e16091bda29bb60e5bd5edd2eb5e80e847c09fad5a4fcdd859e58f7
+size 55995
diff --git a/udpos.Vietnamese/xtreme-test.parquet b/udpos.Vietnamese/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..73015ceae7d9780b44749b5af501db1fc59c3464
--- /dev/null
+++ b/udpos.Vietnamese/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2f043ca92ec9f76b921bbce7eb686e3997ea5e51c6a8d8cc057ec3c40c3d2727
+size 50354
diff --git a/udpos.Vietnamese/xtreme-train.parquet b/udpos.Vietnamese/xtreme-train.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..f20082e0f783c7428ba49909018c1f355c6ba039
--- /dev/null
+++ b/udpos.Vietnamese/xtreme-train.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80be571de25711602f221a3d1e8346195784ead509a7bbce9acf6f78498d7618
+size 83558
diff --git a/udpos.Vietnamese/xtreme-validation.parquet b/udpos.Vietnamese/xtreme-validation.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..340e6ed59fde8820bc0d5c45c1e9cd638e2a288f
--- /dev/null
+++ b/udpos.Vietnamese/xtreme-validation.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4d65d1669197f408d00f8a1728ba8abd6e185df0631306ec689105dd8ce63325
+size 47258
diff --git a/udpos.Yoruba/xtreme-test.parquet b/udpos.Yoruba/xtreme-test.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..ca8aae4bc398f9d7fd6c042a4cf2fc49f6545d48
--- /dev/null
+++ b/udpos.Yoruba/xtreme-test.parquet
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9179f8e516caacc45b2b9f10ede0b96ebf4d739550cfd3d45dc31ff011004fc5
+size 10132
diff --git a/xtreme.py b/xtreme.py
deleted file mode 100644
index 6a16fff932ccbca2c0c6cd95c9f074cbb21c218d..0000000000000000000000000000000000000000
--- a/xtreme.py
+++ /dev/null
@@ -1,940 +0,0 @@
-"""TODO(xtreme): Add a description here."""
-
-
-import csv
-import json
-import os
-import textwrap
-
-import datasets
-
-
-# TODO(xtreme): BibTeX citation
-_CITATION = """\
-@article{hu2020xtreme,
-      author    = {Junjie Hu and Sebastian Ruder and Aditya Siddhant and Graham Neubig and Orhan Firat and Melvin Johnson},
-      title     = {XTREME: A Massively Multilingual Multi-task Benchmark for Evaluating Cross-lingual Generalization},
-      journal   = {CoRR},
-      volume    = {abs/2003.11080},
-      year      = {2020},
-      archivePrefix = {arXiv},
-      eprint    = {2003.11080}
-}
-"""
-
-# TODO(xtrem):
-_DESCRIPTION = """\
-The Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark is a benchmark for the evaluation of
-the cross-lingual generalization ability of pre-trained multilingual models. It covers 40 typologically diverse languages
-(spanning 12 language families) and includes nine tasks that collectively require reasoning about different levels of
-syntax and semantics. The languages in XTREME are selected to maximize language diversity, coverage in existing tasks,
-and availability of training data. Among these are many under-studied languages, such as the Dravidian languages Tamil
-(spoken in southern India, Sri Lanka, and Singapore), Telugu and Malayalam (spoken mainly in southern India), and the
-Niger-Congo languages Swahili and Yoruba, spoken in Africa.
-"""
-_MLQA_LANG = ["ar", "de", "vi", "zh", "en", "es", "hi"]
-_XQUAD_LANG = ["ar", "de", "vi", "zh", "en", "es", "hi", "el", "ru", "th", "tr"]
-_PAWSX_LANG = ["de", "en", "es", "fr", "ja", "ko", "zh"]
-_BUCC_LANG = ["de", "fr", "zh", "ru"]
-_TATOEBA_LANG = [
-    "afr",
-    "ara",
-    "ben",
-    "bul",
-    "deu",
-    "cmn",
-    "ell",
-    "est",
-    "eus",
-    "fin",
-    "fra",
-    "heb",
-    "hin",
-    "hun",
-    "ind",
-    "ita",
-    "jav",
-    "jpn",
-    "kat",
-    "kaz",
-    "kor",
-    "mal",
-    "mar",
-    "nld",
-    "pes",
-    "por",
-    "rus",
-    "spa",
-    "swh",
-    "tam",
-    "tel",
-    "tgl",
-    "tha",
-    "tur",
-    "urd",
-    "vie",
-]
-
-_UD_POS_LANG = [
-    "Afrikaans",
-    "Arabic",
-    "Basque",
-    "Bulgarian",
-    "Dutch",
-    "English",
-    "Estonian",
-    "Finnish",
-    "French",
-    "German",
-    "Greek",
-    "Hebrew",
-    "Hindi",
-    "Hungarian",
-    "Indonesian",
-    "Italian",
-    "Japanese",
-    "Kazakh",
-    "Korean",
-    "Chinese",
-    "Marathi",
-    "Persian",
-    "Portuguese",
-    "Russian",
-    "Spanish",
-    "Tagalog",
-    "Tamil",
-    "Telugu",
-    "Thai",
-    "Turkish",
-    "Urdu",
-    "Vietnamese",
-    "Yoruba",
-]
-_PAN_X_LANG = [
-    "af",
-    "ar",
-    "bg",
-    "bn",
-    "de",
-    "el",
-    "en",
-    "es",
-    "et",
-    "eu",
-    "fa",
-    "fi",
-    "fr",
-    "he",
-    "hi",
-    "hu",
-    "id",
-    "it",
-    "ja",
-    "jv",
-    "ka",
-    "kk",
-    "ko",
-    "ml",
-    "mr",
-    "ms",
-    "my",
-    "nl",
-    "pt",
-    "ru",
-    "sw",
-    "ta",
-    "te",
-    "th",
-    "tl",
-    "tr",
-    "ur",
-    "vi",
-    "yo",
-    "zh",
-]
-
-_NAMES = ["XNLI", "tydiqa", "SQuAD"]
-for lang in _PAN_X_LANG:
-    _NAMES.append(f"PAN-X.{lang}")
-for lang1 in _MLQA_LANG:
-    for lang2 in _MLQA_LANG:
-        _NAMES.append(f"MLQA.{lang1}.{lang2}")
-for lang in _XQUAD_LANG:
-    _NAMES.append(f"XQuAD.{lang}")
-for lang in _BUCC_LANG:
-    _NAMES.append(f"bucc18.{lang}")
-for lang in _PAWSX_LANG:
-    _NAMES.append(f"PAWS-X.{lang}")
-for lang in _TATOEBA_LANG:
-    _NAMES.append(f"tatoeba.{lang}")
-for lang in _UD_POS_LANG:
-    _NAMES.append(f"udpos.{lang}")
-
-_DESCRIPTIONS = {
-    "tydiqa": textwrap.dedent(
-        """Gold passage task (GoldP): Given a passage that is guaranteed to contain the
-             answer, predict the single contiguous span of characters that answers the question. This is more similar to
-             existing reading comprehension datasets (as opposed to the information-seeking task outlined above).
-             This task is constructed with two goals in mind: (1) more directly comparing with prior work and (2) providing
-             a simplified way for researchers to use TyDi QA by providing compatibility with existing code for SQuAD 1.1,
-             XQuAD, and MLQA. Toward these goals, the gold passage task differs from the primary task in several ways:
-             only the gold answer passage is provided rather than the entire Wikipedia article;
-             unanswerable questions have been discarded, similar to MLQA and XQuAD;
-             we evaluate with the SQuAD 1.1 metrics like XQuAD; and
-            Thai and Japanese are removed since the lack of whitespace breaks some tools.
-             """
-    ),
-    "XNLI": textwrap.dedent(
-        """
-          The Cross-lingual Natural Language Inference (XNLI) corpus is a crowd-sourced collection of 5,000 test and
-          2,500 dev pairs for the MultiNLI corpus. The pairs are annotated with textual entailment and translated into
-          14 languages: French, Spanish, German, Greek, Bulgarian, Russian, Turkish, Arabic, Vietnamese, Thai, Chinese,
-          Hindi, Swahili and Urdu. This results in 112.5k annotated pairs. Each premise can be associated with the
-          corresponding hypothesis in the 15 languages, summing up to more than 1.5M combinations. The corpus is made to
-          evaluate how to perform inference in any language (including low-resources ones like Swahili or Urdu) when only
-          English NLI data is available at training time. One solution is cross-lingual sentence encoding, for which XNLI
-          is an evaluation benchmark."""
-    ),
-    "PAWS-X": textwrap.dedent(
-        """
-          This dataset contains 23,659 human translated PAWS evaluation pairs and 296,406 machine translated training
-          pairs in six typologically distinct languages: French, Spanish, German, Chinese, Japanese, and Korean. All
-          translated pairs are sourced from examples in PAWS-Wiki."""
-    ),
-    "XQuAD": textwrap.dedent(
-        """\
-          XQuAD (Cross-lingual Question Answering Dataset) is a benchmark dataset for evaluating cross-lingual question
-          answering performance. The dataset consists of a subset of 240 paragraphs and 1190 question-answer pairs from
-          the development set of SQuAD v1.1 (Rajpurkar et al., 2016) together with their professional translations into
-          ten languages: Spanish, German, Greek, Russian, Turkish, Arabic, Vietnamese, Thai, Chinese, and Hindi. Consequently,
-          the dataset is entirely parallel across 11 languages."""
-    ),
-    "MLQA": textwrap.dedent(
-        """\
-          MLQA (MultiLingual Question Answering) is a benchmark dataset for evaluating cross-lingual question answering performance.
-    MLQA consists of over 5K extractive QA instances (12K in English) in SQuAD format in seven languages - English, Arabic,
-    German, Spanish, Hindi, Vietnamese and Simplified Chinese. MLQA is highly parallel, with QA instances parallel between
-    4 different languages on average."""
-    ),
-    "tatoeba": textwrap.dedent(
-        """\
-          his data is extracted from the Tatoeba corpus, dated Saturday 2018/11/17.
-          For each languages, we have selected 1000 English sentences and their translations, if available. Please check
-          this paper for a description of the languages, their families and scripts as well as baseline results.
-          Please note that the English sentences are not identical for all language pairs. This means that the results are
-          not directly comparable across languages. In particular, the sentences tend to have less variety for several
-          low-resource languages, e.g. "Tom needed water", "Tom needs water", "Tom is getting water", ...
-                    """
-    ),
-    "bucc18": textwrap.dedent(
-        """Building and Using Comparable Corpora
-          """
-    ),
-    "udpos": textwrap.dedent(
-        """\
-    Universal Dependencies (UD) is a framework for consistent annotation of grammar (parts of speech, morphological
-    features, and syntactic dependencies) across different human languages. UD is an open community effort with over 200
-    contributors producing more than 100 treebanks in over 70 languages. If you’re new to UD, you should start by reading
-    the first part of the Short Introduction and then browsing the annotation guidelines.
-    """
-    ),
-    "SQuAD": textwrap.dedent(
-        """\
-    Stanford Question Answering Dataset (SQuAD) is a reading comprehension \
-    dataset, consisting of questions posed by crowdworkers on a set of Wikipedia \
-    articles, where the answer to every question is a segment of text, or span, \
-    from the corresponding reading passage, or the question might be unanswerable."""
-    ),
-    "PAN-X": textwrap.dedent(
-        """\
-    The WikiANN dataset (Pan et al. 2017) is a dataset with NER annotations for PER, ORG and LOC. It has been
-    constructed using the linked entities in Wikipedia pages for 282 different languages including Danish. The dataset
-    can be loaded with the DaNLP package:"""
-    ),
-}
-_CITATIONS = {
-    "tydiqa": textwrap.dedent(
-        (
-            """\
-            @article{tydiqa,
-              title   = {TyDi QA: A Benchmark for Information-Seeking Question Answering in Typologically Diverse Languages},
-              author  = {Jonathan H. Clark and Eunsol Choi and Michael Collins and Dan Garrette and Tom Kwiatkowski and Vitaly Nikolaev and Jennimaria Palomaki}
-              year    = {2020},
-              journal = {Transactions of the Association for Computational Linguistics}
-              }"""
-        )
-    ),
-    "XNLI": textwrap.dedent(
-        """\
-          @InProceedings{conneau2018xnli,
-          author = {Conneau, Alexis
-                         and Rinott, Ruty
-                         and Lample, Guillaume
-                         and Williams, Adina
-                         and Bowman, Samuel R.
-                         and Schwenk, Holger
-                         and Stoyanov, Veselin},
-          title = {XNLI: Evaluating Cross-lingual Sentence Representations},
-          booktitle = {Proceedings of the 2018 Conference on Empirical Methods
-                       in Natural Language Processing},
-          year = {2018},
-          publisher = {Association for Computational Linguistics},
-          location = {Brussels, Belgium},
-        }"""
-    ),
-    "XQuAD": textwrap.dedent(
-        """
-          @article{Artetxe:etal:2019,
-              author    = {Mikel Artetxe and Sebastian Ruder and Dani Yogatama},
-              title     = {On the cross-lingual transferability of monolingual representations},
-              journal   = {CoRR},
-              volume    = {abs/1910.11856},
-              year      = {2019},
-              archivePrefix = {arXiv},
-              eprint    = {1910.11856}
-        }
-        """
-    ),
-    "MLQA": textwrap.dedent(
-        """\
-          @article{lewis2019mlqa,
-          title={MLQA: Evaluating Cross-lingual Extractive Question Answering},
-          author={Lewis, Patrick and Oguz, Barlas and Rinott, Ruty and Riedel, Sebastian and Schwenk, Holger},
-          journal={arXiv preprint arXiv:1910.07475},
-          year={2019}"""
-    ),
-    "PAWS-X": textwrap.dedent(
-        """\
-          @InProceedings{pawsx2019emnlp,
-          title = {{PAWS-X: A Cross-lingual Adversarial Dataset for Paraphrase Identification}},
-          author = {Yang, Yinfei and Zhang, Yuan and Tar, Chris and Baldridge, Jason},
-          booktitle = {Proc. of EMNLP},
-          year = {2019}
-        }"""
-    ),
-    "tatoeba": textwrap.dedent(
-        """\
-                    @article{tatoeba,
-            title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},
-            author={Mikel, Artetxe and Holger, Schwenk,},
-            journal={arXiv:1812.10464v2},
-            year={2018}
-          }"""
-    ),
-    "bucc18": textwrap.dedent(""""""),
-    "udpos": textwrap.dedent(""""""),
-    "SQuAD": textwrap.dedent(
-        """\
-        @article{2016arXiv160605250R,
-           author = {{Rajpurkar}, Pranav and {Zhang}, Jian and {Lopyrev},
-                     Konstantin and {Liang}, Percy},
-            title = "{SQuAD: 100,000+ Questions for Machine Comprehension of Text}",
-          journal = {arXiv e-prints},
-             year = 2016,
-              eid = {arXiv:1606.05250},
-            pages = {arXiv:1606.05250},
-            archivePrefix = {arXiv},
-           eprint = {1606.05250},
-}"""
-    ),
-    "PAN-X": textwrap.dedent(
-        """\
-                    @article{pan-x,
-            title={Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond},
-            author={Xiaoman, Pan and Boliang, Zhang and Jonathan, May and Joel, Nothman and Kevin, Knight and Heng, Ji},
-            volume={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers}
-            year={2017}
-          }"""
-    ),
-}
-
-_TEXT_FEATURES = {
-    "XNLI": {
-        "language": "language",
-        "sentence1": "sentence1",
-        "sentence2": "sentence2",
-    },
-    "tydiqa": {
-        "id": "id",
-        "title": "title",
-        "context": "context",
-        "question": "question",
-        "answers": "answers",
-    },
-    "XQuAD": {
-        "id": "id",
-        "context": "context",
-        "question": "question",
-        "answers": "answers",
-    },
-    "MLQA": {
-        "id": "id",
-        "title": "title",
-        "context": "context",
-        "question": "question",
-        "answers": "answers",
-    },
-    "tatoeba": {
-        "source_sentence": "",
-        "target_sentence": "",
-        "source_lang": "",
-        "target_lang": "",
-    },
-    "bucc18": {
-        "source_sentence": "",
-        "target_sentence": "",
-        "source_lang": "",
-        "target_lang": "",
-    },
-    "PAWS-X": {"sentence1": "sentence1", "sentence2": "sentence2"},
-    "udpos": {"tokens": "", "pos_tags": ""},
-    "SQuAD": {
-        "id": "id",
-        "title": "title",
-        "context": "context",
-        "question": "question",
-        "answers": "answers",
-    },
-    "PAN-X": {"tokens": "", "ner_tags": "", "lang": ""},
-}
-_DATA_URLS = {
-    "tydiqa": "https://storage.googleapis.com/tydiqa/",
-    "XNLI": "https://dl.fbaipublicfiles.com/XNLI/XNLI-1.0.zip",
-    "XQuAD": "https://github.com/deepmind/xquad/raw/master/",
-    "MLQA": "https://dl.fbaipublicfiles.com/MLQA/MLQA_V1.zip",
-    "PAWS-X": "https://storage.googleapis.com/paws/pawsx/x-final.tar.gz",
-    "bucc18": "https://comparable.limsi.fr/bucc2018/",
-    "tatoeba": "https://github.com/facebookresearch/LASER/raw/main/data/tatoeba/v1/",
-    "udpos": "https://lindat.mff.cuni.cz/repository/xmlui/bitstream/handle/11234/1-3105/ud-treebanks-v2.5.tgz",
-    "SQuAD": "https://rajpurkar.github.io/SQuAD-explorer/dataset/",
-    "PAN-X": "https://s3.amazonaws.com/datasets.huggingface.co/wikiann/1.1.0/panx_dataset.zip",
-}
-
-_URLS = {
-    "tydiqa": "https://github.com/google-research-datasets/tydiqa",
-    "XQuAD": "https://github.com/deepmind/xquad",
-    "XNLI": "https://www.nyu.edu/projects/bowman/xnli/",
-    "MLQA": "https://github.com/facebookresearch/MLQA",
-    "PAWS-X": "https://github.com/google-research-datasets/paws/tree/master/pawsx",
-    "bucc18": "https://comparable.limsi.fr/bucc2018/",
-    "tatoeba": "https://github.com/facebookresearch/LASER/blob/main/data/tatoeba/v1/README.md",
-    "udpos": "https://universaldependencies.org/",
-    "SQuAD": "https://rajpurkar.github.io/SQuAD-explorer/",
-    "PAN-X": "https://github.com/afshinrahimi/mmner",
-}
-
-
-class XtremeConfig(datasets.BuilderConfig):
-    """BuilderConfig for Break"""
-
-    def __init__(self, data_url, citation, url, text_features, **kwargs):
-        """
-        Args:
-            text_features: `dict[string, string]`, map from the name of the feature
-        dict for each text field to the name of the column in the tsv file
-            label_column:
-            label_classes
-            **kwargs: keyword arguments forwarded to super.
-        """
-        super(XtremeConfig, self).__init__(version=datasets.Version("1.0.0", ""), **kwargs)
-        self.text_features = text_features
-        self.data_url = data_url
-        self.citation = citation
-        self.url = url
-
-
-class Xtreme(datasets.GeneratorBasedBuilder):
-    """TODO(xtreme): Short description of my dataset."""
-
-    # TODO(xtreme): Set up version.
-    VERSION = datasets.Version("0.1.0")
-    BUILDER_CONFIGS = [
-        XtremeConfig(
-            name=name,
-            description=_DESCRIPTIONS[name.split(".")[0]],
-            citation=_CITATIONS[name.split(".")[0]],
-            text_features=_TEXT_FEATURES[name.split(".")[0]],
-            data_url=_DATA_URLS[name.split(".")[0]],
-            url=_URLS[name.split(".")[0]],
-        )
-        for name in _NAMES
-    ]
-
-    def _info(self):
-        features = {text_feature: datasets.Value("string") for text_feature in self.config.text_features.keys()}
-        if "answers" in features.keys():
-            features["answers"] = datasets.features.Sequence(
-                {
-                    "answer_start": datasets.Value("int32"),
-                    "text": datasets.Value("string"),
-                }
-            )
-        if self.config.name.startswith("PAWS-X"):
-            features = PawsxParser.features
-        elif self.config.name == "XNLI":
-            features["gold_label"] = datasets.Value("string")
-        elif self.config.name.startswith("udpos"):
-            features = UdposParser.features
-        elif self.config.name.startswith("PAN-X"):
-            features = PanxParser.features
-        return datasets.DatasetInfo(
-            # This is the description that will appear on the datasets page.
-            description=self.config.description + "\n" + _DESCRIPTION,
-            # datasets.features.FeatureConnectors
-            features=datasets.Features(
-                features
-                # These are the features of your dataset like images, labels ...
-            ),
-            # If there's a common (input, target) tuple from the features,
-            # specify them here. They'll be used if as_supervised=True in
-            # builder.as_dataset.
-            supervised_keys=None,
-            # Homepage of the dataset for documentation
-            homepage="https://github.com/google-research/xtreme" + "\t" + self.config.url,
-            citation=self.config.citation + "\n" + _CITATION,
-        )
-
-    def _split_generators(self, dl_manager):
-        """Returns SplitGenerators."""
-        if self.config.name == "tydiqa":
-            train_url = "v1.1/tydiqa-goldp-v1.1-train.json"
-            dev_url = "v1.1/tydiqa-goldp-v1.1-dev.json"
-            urls_to_download = {
-                "train": self.config.data_url + train_url,
-                "dev": self.config.data_url + dev_url,
-            }
-            dl_dir = dl_manager.download_and_extract(urls_to_download)
-            return [
-                datasets.SplitGenerator(
-                    name=datasets.Split.TRAIN,
-                    # These kwargs will be passed to _generate_examples
-                    gen_kwargs={"filepath": dl_dir["train"]},
-                ),
-                datasets.SplitGenerator(
-                    name=datasets.Split.VALIDATION,
-                    # These kwargs will be passed to _generate_examples
-                    gen_kwargs={"filepath": dl_dir["dev"]},
-                ),
-            ]
-        if self.config.name == "XNLI":
-            dl_dir = dl_manager.download_and_extract(self.config.data_url)
-            data_dir = os.path.join(dl_dir, "XNLI-1.0")
-            return [
-                datasets.SplitGenerator(
-                    name=datasets.Split.TEST,
-                    gen_kwargs={"filepath": os.path.join(data_dir, "xnli.test.tsv")},
-                ),
-                datasets.SplitGenerator(
-                    name=datasets.Split.VALIDATION,
-                    gen_kwargs={"filepath": os.path.join(data_dir, "xnli.dev.tsv")},
-                ),
-            ]
-
-        if self.config.name.startswith("MLQA"):
-            mlqa_downloaded_files = dl_manager.download_and_extract(self.config.data_url)
-            l1 = self.config.name.split(".")[1]
-            l2 = self.config.name.split(".")[2]
-            return [
-                datasets.SplitGenerator(
-                    name=datasets.Split.TEST,
-                    # These kwargs will be passed to _generate_examples
-                    gen_kwargs={
-                        "filepath": os.path.join(
-                            os.path.join(mlqa_downloaded_files, "MLQA_V1/test"),
-                            f"test-context-{l1}-question-{l2}.json",
-                        )
-                    },
-                ),
-                datasets.SplitGenerator(
-                    name=datasets.Split.VALIDATION,
-                    # These kwargs will be passed to _generate_examples
-                    gen_kwargs={
-                        "filepath": os.path.join(
-                            os.path.join(mlqa_downloaded_files, "MLQA_V1/dev"),
-                            f"dev-context-{l1}-question-{l2}.json",
-                        )
-                    },
-                ),
-            ]
-
-        if self.config.name.startswith("XQuAD"):
-            lang = self.config.name.split(".")[1]
-            xquad_downloaded_file = dl_manager.download_and_extract(self.config.data_url + f"xquad.{lang}.json")
-            return [
-                datasets.SplitGenerator(
-                    name=datasets.Split.VALIDATION,
-                    # These kwargs will be passed to _generate_examples
-                    gen_kwargs={"filepath": xquad_downloaded_file},
-                ),
-            ]
-        if self.config.name.startswith("PAWS-X"):
-            return PawsxParser.split_generators(dl_manager=dl_manager, config=self.config)
-        elif self.config.name.startswith("tatoeba"):
-            lang = self.config.name.split(".")[1]
-
-            tatoeba_source_data = dl_manager.download_and_extract(self.config.data_url + f"tatoeba.{lang}-eng.{lang}")
-            tatoeba_eng_data = dl_manager.download_and_extract(self.config.data_url + f"tatoeba.{lang}-eng.eng")
-            return [
-                datasets.SplitGenerator(
-                    name=datasets.Split.VALIDATION,
-                    # These kwargs will be passed to _generate_examples
-                    gen_kwargs={"filepath": (tatoeba_source_data, tatoeba_eng_data)},
-                ),
-            ]
-        if self.config.name.startswith("bucc18"):
-            lang = self.config.name.split(".")[1]
-            bucc18_dl_test_archive = dl_manager.download(
-                self.config.data_url + f"bucc2018-{lang}-en.training-gold.tar.bz2"
-            )
-            bucc18_dl_dev_archive = dl_manager.download(
-                self.config.data_url + f"bucc2018-{lang}-en.sample-gold.tar.bz2"
-            )
-            return [
-                datasets.SplitGenerator(
-                    name=datasets.Split.VALIDATION,
-                    gen_kwargs={"filepath": dl_manager.iter_archive(bucc18_dl_dev_archive)},
-                ),
-                datasets.SplitGenerator(
-                    name=datasets.Split.TEST,
-                    gen_kwargs={"filepath": dl_manager.iter_archive(bucc18_dl_test_archive)},
-                ),
-            ]
-        if self.config.name.startswith("udpos"):
-            return UdposParser.split_generators(dl_manager=dl_manager, config=self.config)
-
-        if self.config.name == "SQuAD":
-
-            urls_to_download = {
-                "train": self.config.data_url + "train-v1.1.json",
-                "dev": self.config.data_url + "dev-v1.1.json",
-            }
-            downloaded_files = dl_manager.download_and_extract(urls_to_download)
-
-            return [
-                datasets.SplitGenerator(
-                    name=datasets.Split.TRAIN,
-                    gen_kwargs={"filepath": downloaded_files["train"]},
-                ),
-                datasets.SplitGenerator(
-                    name=datasets.Split.VALIDATION,
-                    gen_kwargs={"filepath": downloaded_files["dev"]},
-                ),
-            ]
-
-        if self.config.name.startswith("PAN-X"):
-            return PanxParser.split_generators(dl_manager=dl_manager, config=self.config)
-
-    def _generate_examples(self, filepath=None, **kwargs):
-        """Yields examples."""
-        # TODO(xtreme): Yields (key, example) tuples from the dataset
-
-        if self.config.name == "tydiqa" or self.config.name.startswith("MLQA") or self.config.name == "SQuAD":
-            with open(filepath, encoding="utf-8") as f:
-                data = json.load(f)
-                for article in data["data"]:
-                    title = article.get("title", "").strip()
-                    for paragraph in article["paragraphs"]:
-                        context = paragraph["context"].strip()
-                        for qa in paragraph["qas"]:
-                            question = qa["question"].strip()
-                            id_ = qa["id"]
-
-                            answer_starts = [answer["answer_start"] for answer in qa["answers"]]
-                            answers = [answer["text"].strip() for answer in qa["answers"]]
-
-                            # Features currently used are "context", "question", and "answers".
-                            # Others are extracted here for the ease of future expansions.
-                            yield id_, {
-                                "title": title,
-                                "context": context,
-                                "question": question,
-                                "id": id_,
-                                "answers": {
-                                    "answer_start": answer_starts,
-                                    "text": answers,
-                                },
-                            }
-        if self.config.name == "XNLI":
-            with open(filepath, encoding="utf-8") as f:
-                data = csv.DictReader(f, delimiter="\t")
-                for id_, row in enumerate(data):
-                    yield id_, {
-                        "sentence1": row["sentence1"],
-                        "sentence2": row["sentence2"],
-                        "language": row["language"],
-                        "gold_label": row["gold_label"],
-                    }
-        if self.config.name.startswith("PAWS-X"):
-            yield from PawsxParser.generate_examples(config=self.config, filepath=filepath, **kwargs)
-        if self.config.name.startswith("XQuAD"):
-            with open(filepath, encoding="utf-8") as f:
-                xquad = json.load(f)
-                for article in xquad["data"]:
-                    for paragraph in article["paragraphs"]:
-                        context = paragraph["context"].strip()
-                        for qa in paragraph["qas"]:
-                            question = qa["question"].strip()
-                            id_ = qa["id"]
-
-                            answer_starts = [answer["answer_start"] for answer in qa["answers"]]
-                            answers = [answer["text"].strip() for answer in qa["answers"]]
-
-                            # Features currently used are "context", "question", and "answers".
-                            # Others are extracted here for the ease of future expansions.
-                            yield id_, {
-                                "context": context,
-                                "question": question,
-                                "id": id_,
-                                "answers": {
-                                    "answer_start": answer_starts,
-                                    "text": answers,
-                                },
-                            }
-        if self.config.name.startswith("bucc18"):
-            lang = self.config.name.split(".")[1]
-            data_dir = f"bucc2018/{lang}-en"
-            for path, file in filepath:
-                if path.startswith(data_dir):
-                    csv_content = [line.decode("utf-8") for line in file]
-                    if path.endswith("en"):
-                        target_sentences = dict(list(csv.reader(csv_content, delimiter="\t", quotechar=None)))
-                    elif path.endswith("gold"):
-                        source_target_ids = list(csv.reader(csv_content, delimiter="\t", quotechar=None))
-                    else:
-                        source_sentences = dict(list(csv.reader(csv_content, delimiter="\t", quotechar=None)))
-
-            for id_, (source_id, target_id) in enumerate(source_target_ids):
-                yield id_, {
-                    "source_sentence": source_sentences[source_id],
-                    "target_sentence": target_sentences[target_id],
-                    "source_lang": source_id,
-                    "target_lang": target_id,
-                }
-        if self.config.name.startswith("tatoeba"):
-            source_file = filepath[0]
-            target_file = filepath[1]
-            source_sentences = []
-            target_sentences = []
-            with open(source_file, encoding="utf-8") as f1:
-                for row in f1:
-                    source_sentences.append(row)
-            with open(target_file, encoding="utf-8") as f2:
-                for row in f2:
-                    target_sentences.append(row)
-            for i in range(len(source_sentences)):
-                yield i, {
-                    "source_sentence": source_sentences[i],
-                    "target_sentence": target_sentences[i],
-                    "source_lang": source_file.split(".")[-1],
-                    "target_lang": "eng",
-                }
-        if self.config.name.startswith("udpos"):
-            yield from UdposParser.generate_examples(config=self.config, filepath=filepath, **kwargs)
-        if self.config.name.startswith("PAN-X"):
-            yield from PanxParser.generate_examples(filepath=filepath, **kwargs)
-
-
-class PanxParser:
-
-    features = datasets.Features(
-        {
-            "tokens": datasets.Sequence(datasets.Value("string")),
-            "ner_tags": datasets.Sequence(
-                datasets.features.ClassLabel(
-                    names=[
-                        "O",
-                        "B-PER",
-                        "I-PER",
-                        "B-ORG",
-                        "I-ORG",
-                        "B-LOC",
-                        "I-LOC",
-                    ]
-                )
-            ),
-            "langs": datasets.Sequence(datasets.Value("string")),
-        }
-    )
-
-    @staticmethod
-    def split_generators(dl_manager=None, config=None):
-        data_dir = dl_manager.download_and_extract(config.data_url)
-        lang = config.name.split(".")[1]
-        archive = os.path.join(data_dir, lang + ".tar.gz")
-        split_filenames = {
-            datasets.Split.TRAIN: "train",
-            datasets.Split.VALIDATION: "dev",
-            datasets.Split.TEST: "test",
-        }
-        return [
-            datasets.SplitGenerator(
-                name=split,
-                gen_kwargs={
-                    "filepath": dl_manager.iter_archive(archive),
-                    "filename": split_filenames[split],
-                },
-            )
-            for split in split_filenames
-        ]
-
-    @staticmethod
-    def generate_examples(filepath=None, filename=None):
-        idx = 1
-        for path, file in filepath:
-            if path.endswith(filename):
-                tokens = []
-                ner_tags = []
-                langs = []
-                for line in file:
-                    line = line.decode("utf-8")
-                    if line == "" or line == "\n":
-                        if tokens:
-                            yield idx, {
-                                "tokens": tokens,
-                                "ner_tags": ner_tags,
-                                "langs": langs,
-                            }
-                            idx += 1
-                            tokens = []
-                            ner_tags = []
-                            langs = []
-                    else:
-                        # pan-x data is tab separated
-                        splits = line.split("\t")
-                        # strip out en: prefix
-                        langs.append(splits[0][:2])
-                        tokens.append(splits[0][3:])
-                        if len(splits) > 1:
-                            ner_tags.append(splits[-1].replace("\n", ""))
-                        else:
-                            # examples have no label in test set
-                            ner_tags.append("O")
-                if tokens:
-                    yield idx, {
-                        "tokens": tokens,
-                        "ner_tags": ner_tags,
-                        "langs": langs,
-                    }
-
-
-class PawsxParser:
-
-    features = datasets.Features(
-        {
-            "sentence1": datasets.Value("string"),
-            "sentence2": datasets.Value("string"),
-            "label": datasets.Value("string"),
-        }
-    )
-
-    @staticmethod
-    def split_generators(dl_manager=None, config=None):
-        lang = config.name.split(".")[1]
-        archive = dl_manager.download(config.data_url)
-        split_filenames = {
-            datasets.Split.TRAIN: "translated_train.tsv" if lang != "en" else "train.tsv",
-            datasets.Split.VALIDATION: "dev_2k.tsv",
-            datasets.Split.TEST: "test_2k.tsv",
-        }
-        return [
-            datasets.SplitGenerator(
-                name=split,
-                gen_kwargs={"filepath": dl_manager.iter_archive(archive), "filename": split_filenames[split]},
-            )
-            for split in split_filenames
-        ]
-
-    @staticmethod
-    def generate_examples(config=None, filepath=None, filename=None):
-        lang = config.name.split(".")[1]
-        for path, file in filepath:
-            if f"/{lang}/" in path and path.endswith(filename):
-                lines = (line.decode("utf-8") for line in file)
-                data = csv.reader(lines, delimiter="\t")
-                next(data)  # skip header
-                for id_, row in enumerate(data):
-                    if len(row) == 4:
-                        yield id_, {
-                            "sentence1": row[1],
-                            "sentence2": row[2],
-                            "label": row[3],
-                        }
-
-
-class UdposParser:
-
-    features = datasets.Features(
-        {
-            "tokens": datasets.Sequence(datasets.Value("string")),
-            "pos_tags": datasets.Sequence(
-                datasets.features.ClassLabel(
-                    names=[
-                        "ADJ",
-                        "ADP",
-                        "ADV",
-                        "AUX",
-                        "CCONJ",
-                        "DET",
-                        "INTJ",
-                        "NOUN",
-                        "NUM",
-                        "PART",
-                        "PRON",
-                        "PROPN",
-                        "PUNCT",
-                        "SCONJ",
-                        "SYM",
-                        "VERB",
-                        "X",
-                    ]
-                )
-            ),
-        }
-    )
-
-    @staticmethod
-    def split_generators(dl_manager=None, config=None):
-        archive = dl_manager.download(config.data_url)
-        split_names = {datasets.Split.TRAIN: "train", datasets.Split.VALIDATION: "dev", datasets.Split.TEST: "test"}
-        split_generators = {
-            split: datasets.SplitGenerator(
-                name=split,
-                gen_kwargs={
-                    "filepath": dl_manager.iter_archive(archive),
-                    "split": split_names[split],
-                },
-            )
-            for split in split_names
-        }
-        lang = config.name.split(".")[1]
-        if lang in ["Tagalog", "Thai", "Yoruba"]:
-            return [split_generators["test"]]
-        elif lang == "Kazakh":
-            return [split_generators["train"], split_generators["test"]]
-        else:
-            return [split_generators["train"], split_generators["validation"], split_generators["test"]]
-
-    @staticmethod
-    def generate_examples(config=None, filepath=None, split=None):
-        lang = config.name.split(".")[1]
-        idx = 0
-        for path, file in filepath:
-            if f"_{lang}" in path and split in path and path.endswith(".conllu"):
-                # For lang other than [see below], we exclude Arabic-NYUAD which does not contains any words, only _
-                if lang in ["Kazakh", "Tagalog", "Thai", "Yoruba"] or "NYUAD" not in path:
-                    lines = (line.decode("utf-8") for line in file)
-                    data = csv.reader(lines, delimiter="\t", quoting=csv.QUOTE_NONE)
-                    tokens = []
-                    pos_tags = []
-                    for id_row, row in enumerate(data):
-                        if len(row) >= 10 and row[1] != "_" and row[3] != "_":
-                            tokens.append(row[1])
-                            pos_tags.append(row[3])
-                        if len(row) == 0 and len(tokens) > 0:
-                            yield idx, {
-                                "tokens": tokens,
-                                "pos_tags": pos_tags,
-                            }
-                            idx += 1
-                            tokens = []
-                            pos_tags = []