diff --git a/config.json b/config.json index 4d5c0f559e60e60d1ab84844e526239d862623e3..96cabf12e7ece3489be953092bcea765e65ca161 100644 --- a/config.json +++ b/config.json @@ -1,5 +1,5 @@ { - "_name_or_path": "/project/lt200056-opgpth/boss/stanford_alpaca/llama_2_13b_fixed_resized", + "_name_or_path": "/project/lt200056-opgpth/weight_llama_2_finetune_13b_512_th100_16/checkpoint-1300", "architectures": [ "LlamaForCausalLM" ], @@ -21,7 +21,7 @@ "rope_scaling": null, "rope_theta": 10000.0, "tie_word_embeddings": false, - "torch_dtype": "bfloat16", + "torch_dtype": "float32", "transformers_version": "4.33.1", "use_cache": true, "vocab_size": 56554 diff --git a/generation_config.json b/generation_config.json new file mode 100644 index 0000000000000000000000000000000000000000..d91279883b380cc7513492e8dfb095e65f7a58af --- /dev/null +++ b/generation_config.json @@ -0,0 +1,7 @@ +{ + "_from_model_config": true, + "bos_token_id": 1, + "eos_token_id": 2, + "pad_token_id": 0, + "transformers_version": "4.33.1" +} diff --git a/pytorch_model-00001-of-00056.bin b/pytorch_model-00001-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..84cf99384f746f307216a06276a1990456d708db --- /dev/null +++ b/pytorch_model-00001-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9966dbeaf5e41173f8eeb11c3e40eb1864fcd5df1d64810d49d66603beae57d6 +size 1158226858 diff --git a/pytorch_model-00002-of-00056.bin b/pytorch_model-00002-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..2f1084e7a4ef4af2d21fe74f4b4fea8ae4803bfc --- /dev/null +++ b/pytorch_model-00002-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6dc02fb48e8613f00c1d7d9dcf9e4b8952dee471bf794669f2c4682cfda5a5cb +size 985663935 diff --git a/pytorch_model-00003-of-00056.bin b/pytorch_model-00003-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..f36109c14fbfbb61fe6d8d8a67807a5657a5922b --- /dev/null +++ b/pytorch_model-00003-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6d507fa7ee67735b02a7e3c635dea61182a481e80a94c116e531e69bcecf326a +size 985705505 diff --git a/pytorch_model-00004-of-00056.bin b/pytorch_model-00004-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..e5224de3026a01125ee348ebfdc7a04754305f1d --- /dev/null +++ b/pytorch_model-00004-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6644b31bc9e987a622cb1644fadb2a1d3b66d1160f8b0d5b70e7b26d7b23bb20 +size 985705505 diff --git a/pytorch_model-00005-of-00056.bin b/pytorch_model-00005-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..531f9d18299eb11c44db0c0ec0f2d9636b300db7 --- /dev/null +++ b/pytorch_model-00005-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fcd734367a1716bd325e39e8094ae4f7f0879c4fa0716edad76a035016c2f30e +size 954247615 diff --git a/pytorch_model-00006-of-00056.bin b/pytorch_model-00006-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..5babbc5b092a4f2aa8fe83e82ad26d15b3dde6b9 --- /dev/null +++ b/pytorch_model-00006-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8ac8f3f9cf08dd3844363f2a10e16bfe1303ba6d02ff3c15b2318a52c7d43493 +size 880805998 diff --git a/pytorch_model-00007-of-00056.bin b/pytorch_model-00007-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..feb24178d6dd4cb71f0f1421043a882ea32f6303 --- /dev/null +++ b/pytorch_model-00007-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d1d521933c6fa395ea5060fde6ed17dbef175be83be4f545dea29db8b6d2e084 +size 985705505 diff --git a/pytorch_model-00008-of-00056.bin b/pytorch_model-00008-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..c1ebb0529d6f56b3e41f3fdccdcee24ed462c27a --- /dev/null +++ b/pytorch_model-00008-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2bca248afad326647fc4b900c31b0aa4f6aaa278559b276ff56d7c784cf5f4c9 +size 985705505 diff --git a/pytorch_model-00009-of-00056.bin b/pytorch_model-00009-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..a34d9bd60e05dfb07c329e71dbe2059908920545 --- /dev/null +++ b/pytorch_model-00009-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d441f8a0f5d4617138c4729bcf3cd29a1a2a6e4bf6c7a1a78c0934d175fd93f0 +size 954247615 diff --git a/pytorch_model-00010-of-00056.bin b/pytorch_model-00010-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..23f7543bd157da3f2ecd8e66dda3e3547170ee35 --- /dev/null +++ b/pytorch_model-00010-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cad316ba6cc01c842bd7f70718fecebc4dc22dbea7b045818931281cdd612060 +size 880805998 diff --git a/pytorch_model-00011-of-00056.bin b/pytorch_model-00011-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..0f8e69163a1b68c51eef06dfaf0acdf2e622fa4d --- /dev/null +++ b/pytorch_model-00011-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9555c0b63c01917774a3add1f12a59df0b56ca8d362c0335dad7e1c5b862baa3 +size 985705505 diff --git a/pytorch_model-00012-of-00056.bin b/pytorch_model-00012-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..1fa7d1b362be2d2f217498822e2ff81895b8c9ba --- /dev/null +++ b/pytorch_model-00012-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2a4b807a279cd29d8a6d51b9da3b227e6e062701fb1738b122b6646640f5cb48 +size 985705505 diff --git a/pytorch_model-00013-of-00056.bin b/pytorch_model-00013-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..08a6391b0e77a3782378b2306ada21f7611d25ba --- /dev/null +++ b/pytorch_model-00013-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0a1ac813a757dd8187a9038f601e65c1f4edcd9d2e1a127f037b797eabff5861 +size 954247615 diff --git a/pytorch_model-00014-of-00056.bin b/pytorch_model-00014-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..8fbb39bbb26430576cdb087ec4f356bb2f348a71 --- /dev/null +++ b/pytorch_model-00014-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ae332d889e73a54698dc3cf0ec3abdffb7a21db5c27f6cd46dbbd86f1b0de756 +size 880805998 diff --git a/pytorch_model-00015-of-00056.bin b/pytorch_model-00015-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..13f7ea9086495a4427cec78ae89eaa6e66da4319 --- /dev/null +++ b/pytorch_model-00015-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b03991a94edde33767d8de19c08164a606f12f109d5eacc9bebc51119b214572 +size 985705505 diff --git a/pytorch_model-00016-of-00056.bin b/pytorch_model-00016-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..c5add347b2037cf9ba83aeef0be14ae8a5e00bef --- /dev/null +++ b/pytorch_model-00016-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c2d8a76eff54a55230fbc72e713a5784c77c8ddced3efd3a04cb7a7f02047aed +size 985705505 diff --git a/pytorch_model-00017-of-00056.bin b/pytorch_model-00017-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..1568011fd9183efe9b9c8a8cd63b07dc8d751a44 --- /dev/null +++ b/pytorch_model-00017-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d4f247b342e3f19752a8306ed4173decbe5b6f64b38588db66d57b44314416e2 +size 954247615 diff --git a/pytorch_model-00018-of-00056.bin b/pytorch_model-00018-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..46bbb0ef1b4635813f239d12f101e83ff8917133 --- /dev/null +++ b/pytorch_model-00018-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ae2f0cc5ff6e791fb3074702e00ec5cfe089b5f22c2fe74d6e9479099231ed54 +size 880806062 diff --git a/pytorch_model-00019-of-00056.bin b/pytorch_model-00019-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..a8d41d862cef8e7848236f6113ac90cf564b15c6 --- /dev/null +++ b/pytorch_model-00019-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3fc147d740845d2d80874afc50c00e90ed650f0de5c514a2aff0f6b3e88ce3a1 +size 985705505 diff --git a/pytorch_model-00020-of-00056.bin b/pytorch_model-00020-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..c9508b951a18cb6f6248c4266aa927c3a3d26eb3 --- /dev/null +++ b/pytorch_model-00020-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:74e82b65bfa5aab4bb8c5d45624b0f913c68cd5259d3da116800dbf28b38276f +size 985705505 diff --git a/pytorch_model-00021-of-00056.bin b/pytorch_model-00021-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..b04418c5ac0a909613904336aec6040681ba7198 --- /dev/null +++ b/pytorch_model-00021-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:625265621157d02c3b31fcad63b20315a9f19a16e23d74a6344119fa2ba40a3f +size 954247615 diff --git a/pytorch_model-00022-of-00056.bin b/pytorch_model-00022-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..3bad848754767825b9c75c40f6594ef27502044c --- /dev/null +++ b/pytorch_model-00022-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:eec166578f23e347bd3f882be3af236f190022ce4275d04abcd5fcca167b00b6 +size 880806062 diff --git a/pytorch_model-00023-of-00056.bin b/pytorch_model-00023-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..29cc0306f08070010833ed28b768858a6ba6bfed --- /dev/null +++ b/pytorch_model-00023-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cf11e3d793a28d146e8829182973c96f22a3030f633e92cc242ace15ac59763e +size 985705505 diff --git a/pytorch_model-00024-of-00056.bin b/pytorch_model-00024-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..b4c4b82142720d0527e68af2d6cc613818e59c6c --- /dev/null +++ b/pytorch_model-00024-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:471908272a41fc176523c32dd57def56d8971f7cdf87803a4147bdc35d2381fb +size 985705505 diff --git a/pytorch_model-00025-of-00056.bin b/pytorch_model-00025-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..b2951c6fff67a69129024f733a667de1863b0a95 --- /dev/null +++ b/pytorch_model-00025-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:05bd724ec2afded99d571c805f6d0b137aeff1f998fcdec4f6eda288240b9d60 +size 954247615 diff --git a/pytorch_model-00026-of-00056.bin b/pytorch_model-00026-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..2cba950b6f3c238f69cf723535266e337e104c86 --- /dev/null +++ b/pytorch_model-00026-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:abdef20e278f1a242274444d2aa0ca4b02ae1b6873e39811c28fa266f4960dc3 +size 880806062 diff --git a/pytorch_model-00027-of-00056.bin b/pytorch_model-00027-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..d457480273d91f82e93431020676c2b6bdf43c3a --- /dev/null +++ b/pytorch_model-00027-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:65d4331eab1010e219287fd3a8a6624bb76698f79ac67245e6b633db5869f507 +size 985705505 diff --git a/pytorch_model-00028-of-00056.bin b/pytorch_model-00028-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..66c46ab39f4032a226055efffe4d92d95b264f8e --- /dev/null +++ b/pytorch_model-00028-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3ae1a734ee7e09e895d0fd5fd06e11762d1569f0efdfd9e8df83f7297a7c3b9b +size 985705505 diff --git a/pytorch_model-00029-of-00056.bin b/pytorch_model-00029-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..6ffde1a85c9ce09e7ed374c9843a8b377aab882f --- /dev/null +++ b/pytorch_model-00029-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:38dbbc9e7148ca21e26f6b88ba21394c2e302f586f364cfdd5c0e47703d7b0e3 +size 954247615 diff --git a/pytorch_model-00030-of-00056.bin b/pytorch_model-00030-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..c81d40e8f95c72ba3110fb71f2a3aca8e177bed1 --- /dev/null +++ b/pytorch_model-00030-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2247511ef82cf7ddfa6e3c792ef5687293f68d519802ac535fdf10f7dbb5618c +size 880806062 diff --git a/pytorch_model-00031-of-00056.bin b/pytorch_model-00031-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..737d24bf2d7ae2a40f39d7504da936e25a59766f --- /dev/null +++ b/pytorch_model-00031-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f83a5ae93cb596e24c1d5e6e73a8f9a3913224c07abbd7efdcf8bef1d8217641 +size 985705505 diff --git a/pytorch_model-00032-of-00056.bin b/pytorch_model-00032-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..dc6998720cf906e21fee21f1bc2e6093f615a52f --- /dev/null +++ b/pytorch_model-00032-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4778845dd72389477a1d1d63f2265115e1e9a98dd839368e44cbf110088c8a87 +size 985705505 diff --git a/pytorch_model-00033-of-00056.bin b/pytorch_model-00033-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..7cbb192fda13f2c75572c0c37fc9007e1763be40 --- /dev/null +++ b/pytorch_model-00033-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1a6c94b988bae82ca7d6b75a0215dde493313893aaf1fa1e93c576c1bb4ce757 +size 954247615 diff --git a/pytorch_model-00034-of-00056.bin b/pytorch_model-00034-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..48b1fbe36924883baa35cac1004b4bfd29ff1597 --- /dev/null +++ b/pytorch_model-00034-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cd3f5db38d1c8863aeaf37272a31a80391d259ea852e07afb455a3e46c6c43c8 +size 880806062 diff --git a/pytorch_model-00035-of-00056.bin b/pytorch_model-00035-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..11172b1526cf424d79d844f402d0201002495194 --- /dev/null +++ b/pytorch_model-00035-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f4ea4d3af6c1bf51fc533741124018dc16f0b9f5e0bf1ef9ff92413a4b519b09 +size 985705505 diff --git a/pytorch_model-00036-of-00056.bin b/pytorch_model-00036-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..41db2c0077e60a5965e5d84ac47400c8a45fcfb2 --- /dev/null +++ b/pytorch_model-00036-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b83ed4e076277a0a51666f27d718c75ee19589a9d527d984fd35295cbe63e1f0 +size 985705505 diff --git a/pytorch_model-00037-of-00056.bin b/pytorch_model-00037-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..0efa32842d16ebbb67e76282b13727f14a473d11 --- /dev/null +++ b/pytorch_model-00037-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6231e345e34bd61db09670e822c2d67cd283bf2a32a26df92869a3609669ecd5 +size 954247615 diff --git a/pytorch_model-00038-of-00056.bin b/pytorch_model-00038-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..46e1ead34478825ec0f71f8320c0d54aa3f1abd1 --- /dev/null +++ b/pytorch_model-00038-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a7eaf7314d14829e4147be5fae603101864d69b4ee1ed3bc17cca6f47e36b980 +size 880806062 diff --git a/pytorch_model-00039-of-00056.bin b/pytorch_model-00039-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..00efa4c89428414cd32dd821ea4f879a4906b93f --- /dev/null +++ b/pytorch_model-00039-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6bf369e3601bad61dcd16ea5e9dbbeeee91cf816c1102bfe95bb30e8f5f48c17 +size 985705505 diff --git a/pytorch_model-00040-of-00056.bin b/pytorch_model-00040-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..413c6e2c66c25886f518d00e90b4f57ccdbb0aea --- /dev/null +++ b/pytorch_model-00040-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:657329de60eb71e2e09b3c238860561d1996519e654ad677c82d23107d15aef7 +size 985705505 diff --git a/pytorch_model-00041-of-00056.bin b/pytorch_model-00041-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..b6d00a5bf9c8c0747013eb0ccc3b1a1b8dade120 --- /dev/null +++ b/pytorch_model-00041-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6c64d9802724aa75730fda1cdde03d9e7a8c8261459a9f61c2406d14d71c9180 +size 954247615 diff --git a/pytorch_model-00042-of-00056.bin b/pytorch_model-00042-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..ce9a77a73516f206902bd6a0d56dd68ad73b7271 --- /dev/null +++ b/pytorch_model-00042-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9cf0089fda0b12e88c34f7d9608b94f36628582df9dce2a93b9cec0ffee8983e +size 880806062 diff --git a/pytorch_model-00043-of-00056.bin b/pytorch_model-00043-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..a462ff2c3ca57dcb1740dae98448cf7a853e4249 --- /dev/null +++ b/pytorch_model-00043-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b36b563a86b71ece39d1e8d63158da29ca362a82ad88a356cca20c5518c53755 +size 985705505 diff --git a/pytorch_model-00044-of-00056.bin b/pytorch_model-00044-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..c86b0fff3e80053c78c282782fbf2954a04ab928 --- /dev/null +++ b/pytorch_model-00044-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b5617da8015fe1a4c87b6444ab5ec8b63bf2ea157522ece3faa0b5527157f39b +size 985705505 diff --git a/pytorch_model-00045-of-00056.bin b/pytorch_model-00045-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..d41a0647c55f003008e3c3dcddbb9e713ac62f9c --- /dev/null +++ b/pytorch_model-00045-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1439e9b3de71793052f66ca51fb4ba8b14a2ffd441c3f2762a63c166aa227774 +size 954247615 diff --git a/pytorch_model-00046-of-00056.bin b/pytorch_model-00046-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..fceb21a38487edc0476a758c9b23e4016bdc0e53 --- /dev/null +++ b/pytorch_model-00046-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:944827f7e3cac58c024aea14df09b29918fb4260022ffb2cd3275022b98c49b6 +size 880806062 diff --git a/pytorch_model-00047-of-00056.bin b/pytorch_model-00047-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..3d1ea752cc1f76b1fe8ab91c2a83e21bc1e12e84 --- /dev/null +++ b/pytorch_model-00047-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:eaea5719c6eb706d36d01b4b4fa506675e7f34e276045cd86731f3db2c09996c +size 985705505 diff --git a/pytorch_model-00048-of-00056.bin b/pytorch_model-00048-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..bc4104d1c5594c6d476b278f23c2e9e9222309df --- /dev/null +++ b/pytorch_model-00048-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:483738a3045f16af8d487d32fdb90e4251976db189102974c8cf10c642d3bcd4 +size 985705505 diff --git a/pytorch_model-00049-of-00056.bin b/pytorch_model-00049-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..8d512e8bf6f995c4b0543db55dceba21bd61b313 --- /dev/null +++ b/pytorch_model-00049-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b9404e8ece24257ccca8180570eabe5f0f244eddea4630eb5991c99447ab826f +size 954247615 diff --git a/pytorch_model-00050-of-00056.bin b/pytorch_model-00050-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..4e3058cc5fe2b93bf7ea8da9ea3324eeb919a44a --- /dev/null +++ b/pytorch_model-00050-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:745624dad1b986631bfaa65d5bed43d760aa3d65989593c59cd12b3aaf097299 +size 880806062 diff --git a/pytorch_model-00051-of-00056.bin b/pytorch_model-00051-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..354fcb004f5c8db63001b74402914bbc88eecd07 --- /dev/null +++ b/pytorch_model-00051-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3cb5310e046b366869ebf496a4fc8341213d9c5738e95fafb7576313ed51e398 +size 985705505 diff --git a/pytorch_model-00052-of-00056.bin b/pytorch_model-00052-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..2f627639408237f53775c7efdf4ddf5efa71484a --- /dev/null +++ b/pytorch_model-00052-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1aa9e34f73f99c80e95c9db91cfbd895309965e1309606c9570114de171047d4 +size 985705505 diff --git a/pytorch_model-00053-of-00056.bin b/pytorch_model-00053-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..15b672ce2ce8a4458b2977fea6d954215e783de3 --- /dev/null +++ b/pytorch_model-00053-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2f7a2ae0972404c862e9950ba5454a7ffcc28483576c56cb79d515b116d7d7ee +size 954247615 diff --git a/pytorch_model-00054-of-00056.bin b/pytorch_model-00054-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..9312bb21b9a5f6c7a66d9d2fc987f7af92b01c63 --- /dev/null +++ b/pytorch_model-00054-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:45bceb66216b53f0d454c002120250bb476027fc9000fcb654906784bf6292bc +size 880806062 diff --git a/pytorch_model-00055-of-00056.bin b/pytorch_model-00055-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..f3663d3162e726e9592cb2bc98985f6ed23a354d --- /dev/null +++ b/pytorch_model-00055-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:542d9c338bd1eddd0803bfe87f903966f1a74739c37a569e1eccde620e123b52 +size 283178781 diff --git a/pytorch_model-00056-of-00056.bin b/pytorch_model-00056-of-00056.bin new file mode 100644 index 0000000000000000000000000000000000000000..de1d187c50ddbf5844b3f9f83f33ea52900f3eae --- /dev/null +++ b/pytorch_model-00056-of-00056.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5013ca49f74d41ec5e431f3eefb64809d4f19d353b0a6f6ad99c4c0208e298e8 +size 1158226858 diff --git a/pytorch_model.bin.index.json b/pytorch_model.bin.index.json new file mode 100644 index 0000000000000000000000000000000000000000..a850ba50761682e500b4e42b78006c11c82dd432 --- /dev/null +++ b/pytorch_model.bin.index.json @@ -0,0 +1,370 @@ +{ + "metadata": { + "total_size": 53069189120 + }, + "weight_map": { + "lm_head.weight": "pytorch_model-00056-of-00056.bin", + "model.embed_tokens.weight": "pytorch_model-00001-of-00056.bin", + "model.layers.0.input_layernorm.weight": "pytorch_model-00003-of-00056.bin", + "model.layers.0.mlp.down_proj.weight": "pytorch_model-00003-of-00056.bin", + "model.layers.0.mlp.gate_proj.weight": "pytorch_model-00002-of-00056.bin", + "model.layers.0.mlp.up_proj.weight": "pytorch_model-00002-of-00056.bin", + "model.layers.0.post_attention_layernorm.weight": "pytorch_model-00003-of-00056.bin", + "model.layers.0.self_attn.k_proj.weight": "pytorch_model-00002-of-00056.bin", + "model.layers.0.self_attn.o_proj.weight": "pytorch_model-00002-of-00056.bin", + "model.layers.0.self_attn.q_proj.weight": "pytorch_model-00002-of-00056.bin", + "model.layers.0.self_attn.v_proj.weight": "pytorch_model-00002-of-00056.bin", + "model.layers.1.input_layernorm.weight": "pytorch_model-00004-of-00056.bin", + "model.layers.1.mlp.down_proj.weight": "pytorch_model-00004-of-00056.bin", + "model.layers.1.mlp.gate_proj.weight": "pytorch_model-00003-of-00056.bin", + "model.layers.1.mlp.up_proj.weight": "pytorch_model-00004-of-00056.bin", + "model.layers.1.post_attention_layernorm.weight": "pytorch_model-00004-of-00056.bin", + "model.layers.1.self_attn.k_proj.weight": "pytorch_model-00003-of-00056.bin", + "model.layers.1.self_attn.o_proj.weight": "pytorch_model-00003-of-00056.bin", + "model.layers.1.self_attn.q_proj.weight": "pytorch_model-00003-of-00056.bin", + "model.layers.1.self_attn.v_proj.weight": "pytorch_model-00003-of-00056.bin", + "model.layers.10.input_layernorm.weight": "pytorch_model-00016-of-00056.bin", + "model.layers.10.mlp.down_proj.weight": "pytorch_model-00016-of-00056.bin", + "model.layers.10.mlp.gate_proj.weight": "pytorch_model-00015-of-00056.bin", + "model.layers.10.mlp.up_proj.weight": "pytorch_model-00016-of-00056.bin", + "model.layers.10.post_attention_layernorm.weight": "pytorch_model-00016-of-00056.bin", + "model.layers.10.self_attn.k_proj.weight": "pytorch_model-00015-of-00056.bin", + "model.layers.10.self_attn.o_proj.weight": "pytorch_model-00015-of-00056.bin", + "model.layers.10.self_attn.q_proj.weight": "pytorch_model-00015-of-00056.bin", + "model.layers.10.self_attn.v_proj.weight": "pytorch_model-00015-of-00056.bin", + "model.layers.11.input_layernorm.weight": "pytorch_model-00017-of-00056.bin", + "model.layers.11.mlp.down_proj.weight": "pytorch_model-00017-of-00056.bin", + "model.layers.11.mlp.gate_proj.weight": "pytorch_model-00017-of-00056.bin", + "model.layers.11.mlp.up_proj.weight": "pytorch_model-00017-of-00056.bin", + "model.layers.11.post_attention_layernorm.weight": "pytorch_model-00017-of-00056.bin", + "model.layers.11.self_attn.k_proj.weight": "pytorch_model-00016-of-00056.bin", + "model.layers.11.self_attn.o_proj.weight": "pytorch_model-00016-of-00056.bin", + "model.layers.11.self_attn.q_proj.weight": "pytorch_model-00016-of-00056.bin", + "model.layers.11.self_attn.v_proj.weight": "pytorch_model-00016-of-00056.bin", + "model.layers.12.input_layernorm.weight": "pytorch_model-00019-of-00056.bin", + "model.layers.12.mlp.down_proj.weight": "pytorch_model-00019-of-00056.bin", + "model.layers.12.mlp.gate_proj.weight": "pytorch_model-00018-of-00056.bin", + "model.layers.12.mlp.up_proj.weight": "pytorch_model-00018-of-00056.bin", + "model.layers.12.post_attention_layernorm.weight": "pytorch_model-00019-of-00056.bin", + "model.layers.12.self_attn.k_proj.weight": "pytorch_model-00018-of-00056.bin", + "model.layers.12.self_attn.o_proj.weight": "pytorch_model-00018-of-00056.bin", + "model.layers.12.self_attn.q_proj.weight": "pytorch_model-00017-of-00056.bin", + "model.layers.12.self_attn.v_proj.weight": "pytorch_model-00018-of-00056.bin", + "model.layers.13.input_layernorm.weight": "pytorch_model-00020-of-00056.bin", + "model.layers.13.mlp.down_proj.weight": "pytorch_model-00020-of-00056.bin", + "model.layers.13.mlp.gate_proj.weight": "pytorch_model-00019-of-00056.bin", + "model.layers.13.mlp.up_proj.weight": "pytorch_model-00020-of-00056.bin", + "model.layers.13.post_attention_layernorm.weight": "pytorch_model-00020-of-00056.bin", + "model.layers.13.self_attn.k_proj.weight": "pytorch_model-00019-of-00056.bin", + "model.layers.13.self_attn.o_proj.weight": "pytorch_model-00019-of-00056.bin", + "model.layers.13.self_attn.q_proj.weight": "pytorch_model-00019-of-00056.bin", + "model.layers.13.self_attn.v_proj.weight": "pytorch_model-00019-of-00056.bin", + "model.layers.14.input_layernorm.weight": "pytorch_model-00021-of-00056.bin", + "model.layers.14.mlp.down_proj.weight": "pytorch_model-00021-of-00056.bin", + "model.layers.14.mlp.gate_proj.weight": "pytorch_model-00021-of-00056.bin", + "model.layers.14.mlp.up_proj.weight": "pytorch_model-00021-of-00056.bin", + "model.layers.14.post_attention_layernorm.weight": "pytorch_model-00021-of-00056.bin", + "model.layers.14.self_attn.k_proj.weight": "pytorch_model-00020-of-00056.bin", + "model.layers.14.self_attn.o_proj.weight": "pytorch_model-00020-of-00056.bin", + "model.layers.14.self_attn.q_proj.weight": "pytorch_model-00020-of-00056.bin", + "model.layers.14.self_attn.v_proj.weight": "pytorch_model-00020-of-00056.bin", + "model.layers.15.input_layernorm.weight": "pytorch_model-00023-of-00056.bin", + "model.layers.15.mlp.down_proj.weight": "pytorch_model-00023-of-00056.bin", + "model.layers.15.mlp.gate_proj.weight": "pytorch_model-00022-of-00056.bin", + "model.layers.15.mlp.up_proj.weight": "pytorch_model-00022-of-00056.bin", + "model.layers.15.post_attention_layernorm.weight": "pytorch_model-00023-of-00056.bin", + "model.layers.15.self_attn.k_proj.weight": "pytorch_model-00022-of-00056.bin", + "model.layers.15.self_attn.o_proj.weight": "pytorch_model-00022-of-00056.bin", + "model.layers.15.self_attn.q_proj.weight": "pytorch_model-00021-of-00056.bin", + "model.layers.15.self_attn.v_proj.weight": "pytorch_model-00022-of-00056.bin", + "model.layers.16.input_layernorm.weight": "pytorch_model-00024-of-00056.bin", + "model.layers.16.mlp.down_proj.weight": "pytorch_model-00024-of-00056.bin", + "model.layers.16.mlp.gate_proj.weight": "pytorch_model-00023-of-00056.bin", + "model.layers.16.mlp.up_proj.weight": "pytorch_model-00024-of-00056.bin", + "model.layers.16.post_attention_layernorm.weight": "pytorch_model-00024-of-00056.bin", + "model.layers.16.self_attn.k_proj.weight": "pytorch_model-00023-of-00056.bin", + "model.layers.16.self_attn.o_proj.weight": "pytorch_model-00023-of-00056.bin", + "model.layers.16.self_attn.q_proj.weight": "pytorch_model-00023-of-00056.bin", + "model.layers.16.self_attn.v_proj.weight": "pytorch_model-00023-of-00056.bin", + "model.layers.17.input_layernorm.weight": "pytorch_model-00025-of-00056.bin", + "model.layers.17.mlp.down_proj.weight": "pytorch_model-00025-of-00056.bin", + "model.layers.17.mlp.gate_proj.weight": "pytorch_model-00025-of-00056.bin", + "model.layers.17.mlp.up_proj.weight": "pytorch_model-00025-of-00056.bin", + "model.layers.17.post_attention_layernorm.weight": "pytorch_model-00025-of-00056.bin", + "model.layers.17.self_attn.k_proj.weight": "pytorch_model-00024-of-00056.bin", + "model.layers.17.self_attn.o_proj.weight": "pytorch_model-00024-of-00056.bin", + "model.layers.17.self_attn.q_proj.weight": "pytorch_model-00024-of-00056.bin", + "model.layers.17.self_attn.v_proj.weight": "pytorch_model-00024-of-00056.bin", + "model.layers.18.input_layernorm.weight": "pytorch_model-00027-of-00056.bin", + "model.layers.18.mlp.down_proj.weight": "pytorch_model-00027-of-00056.bin", + "model.layers.18.mlp.gate_proj.weight": "pytorch_model-00026-of-00056.bin", + "model.layers.18.mlp.up_proj.weight": "pytorch_model-00026-of-00056.bin", + "model.layers.18.post_attention_layernorm.weight": "pytorch_model-00027-of-00056.bin", + "model.layers.18.self_attn.k_proj.weight": "pytorch_model-00026-of-00056.bin", + "model.layers.18.self_attn.o_proj.weight": "pytorch_model-00026-of-00056.bin", + "model.layers.18.self_attn.q_proj.weight": "pytorch_model-00025-of-00056.bin", + "model.layers.18.self_attn.v_proj.weight": "pytorch_model-00026-of-00056.bin", + "model.layers.19.input_layernorm.weight": "pytorch_model-00028-of-00056.bin", + "model.layers.19.mlp.down_proj.weight": "pytorch_model-00028-of-00056.bin", + "model.layers.19.mlp.gate_proj.weight": "pytorch_model-00027-of-00056.bin", + "model.layers.19.mlp.up_proj.weight": "pytorch_model-00028-of-00056.bin", + "model.layers.19.post_attention_layernorm.weight": "pytorch_model-00028-of-00056.bin", + "model.layers.19.self_attn.k_proj.weight": "pytorch_model-00027-of-00056.bin", + "model.layers.19.self_attn.o_proj.weight": "pytorch_model-00027-of-00056.bin", + "model.layers.19.self_attn.q_proj.weight": "pytorch_model-00027-of-00056.bin", + "model.layers.19.self_attn.v_proj.weight": "pytorch_model-00027-of-00056.bin", + "model.layers.2.input_layernorm.weight": "pytorch_model-00005-of-00056.bin", + "model.layers.2.mlp.down_proj.weight": "pytorch_model-00005-of-00056.bin", + "model.layers.2.mlp.gate_proj.weight": "pytorch_model-00005-of-00056.bin", + "model.layers.2.mlp.up_proj.weight": "pytorch_model-00005-of-00056.bin", + "model.layers.2.post_attention_layernorm.weight": "pytorch_model-00005-of-00056.bin", + "model.layers.2.self_attn.k_proj.weight": "pytorch_model-00004-of-00056.bin", + "model.layers.2.self_attn.o_proj.weight": "pytorch_model-00004-of-00056.bin", + "model.layers.2.self_attn.q_proj.weight": "pytorch_model-00004-of-00056.bin", + "model.layers.2.self_attn.v_proj.weight": "pytorch_model-00004-of-00056.bin", + "model.layers.20.input_layernorm.weight": "pytorch_model-00029-of-00056.bin", + "model.layers.20.mlp.down_proj.weight": "pytorch_model-00029-of-00056.bin", + "model.layers.20.mlp.gate_proj.weight": "pytorch_model-00029-of-00056.bin", + "model.layers.20.mlp.up_proj.weight": "pytorch_model-00029-of-00056.bin", + "model.layers.20.post_attention_layernorm.weight": "pytorch_model-00029-of-00056.bin", + "model.layers.20.self_attn.k_proj.weight": "pytorch_model-00028-of-00056.bin", + "model.layers.20.self_attn.o_proj.weight": "pytorch_model-00028-of-00056.bin", + "model.layers.20.self_attn.q_proj.weight": "pytorch_model-00028-of-00056.bin", + "model.layers.20.self_attn.v_proj.weight": "pytorch_model-00028-of-00056.bin", + "model.layers.21.input_layernorm.weight": "pytorch_model-00031-of-00056.bin", + "model.layers.21.mlp.down_proj.weight": "pytorch_model-00031-of-00056.bin", + "model.layers.21.mlp.gate_proj.weight": "pytorch_model-00030-of-00056.bin", + "model.layers.21.mlp.up_proj.weight": "pytorch_model-00030-of-00056.bin", + "model.layers.21.post_attention_layernorm.weight": "pytorch_model-00031-of-00056.bin", + "model.layers.21.self_attn.k_proj.weight": "pytorch_model-00030-of-00056.bin", + "model.layers.21.self_attn.o_proj.weight": "pytorch_model-00030-of-00056.bin", + "model.layers.21.self_attn.q_proj.weight": "pytorch_model-00029-of-00056.bin", + "model.layers.21.self_attn.v_proj.weight": "pytorch_model-00030-of-00056.bin", + "model.layers.22.input_layernorm.weight": "pytorch_model-00032-of-00056.bin", + "model.layers.22.mlp.down_proj.weight": "pytorch_model-00032-of-00056.bin", + "model.layers.22.mlp.gate_proj.weight": "pytorch_model-00031-of-00056.bin", + "model.layers.22.mlp.up_proj.weight": "pytorch_model-00032-of-00056.bin", + "model.layers.22.post_attention_layernorm.weight": "pytorch_model-00032-of-00056.bin", + "model.layers.22.self_attn.k_proj.weight": "pytorch_model-00031-of-00056.bin", + "model.layers.22.self_attn.o_proj.weight": "pytorch_model-00031-of-00056.bin", + "model.layers.22.self_attn.q_proj.weight": "pytorch_model-00031-of-00056.bin", + "model.layers.22.self_attn.v_proj.weight": "pytorch_model-00031-of-00056.bin", + "model.layers.23.input_layernorm.weight": "pytorch_model-00033-of-00056.bin", + "model.layers.23.mlp.down_proj.weight": "pytorch_model-00033-of-00056.bin", + "model.layers.23.mlp.gate_proj.weight": "pytorch_model-00033-of-00056.bin", + "model.layers.23.mlp.up_proj.weight": "pytorch_model-00033-of-00056.bin", + "model.layers.23.post_attention_layernorm.weight": "pytorch_model-00033-of-00056.bin", + "model.layers.23.self_attn.k_proj.weight": "pytorch_model-00032-of-00056.bin", + "model.layers.23.self_attn.o_proj.weight": "pytorch_model-00032-of-00056.bin", + "model.layers.23.self_attn.q_proj.weight": "pytorch_model-00032-of-00056.bin", + "model.layers.23.self_attn.v_proj.weight": "pytorch_model-00032-of-00056.bin", + "model.layers.24.input_layernorm.weight": "pytorch_model-00035-of-00056.bin", + "model.layers.24.mlp.down_proj.weight": "pytorch_model-00035-of-00056.bin", + "model.layers.24.mlp.gate_proj.weight": "pytorch_model-00034-of-00056.bin", + "model.layers.24.mlp.up_proj.weight": "pytorch_model-00034-of-00056.bin", + "model.layers.24.post_attention_layernorm.weight": "pytorch_model-00035-of-00056.bin", + "model.layers.24.self_attn.k_proj.weight": "pytorch_model-00034-of-00056.bin", + "model.layers.24.self_attn.o_proj.weight": "pytorch_model-00034-of-00056.bin", + "model.layers.24.self_attn.q_proj.weight": "pytorch_model-00033-of-00056.bin", + "model.layers.24.self_attn.v_proj.weight": "pytorch_model-00034-of-00056.bin", + "model.layers.25.input_layernorm.weight": "pytorch_model-00036-of-00056.bin", + "model.layers.25.mlp.down_proj.weight": "pytorch_model-00036-of-00056.bin", + "model.layers.25.mlp.gate_proj.weight": "pytorch_model-00035-of-00056.bin", + "model.layers.25.mlp.up_proj.weight": "pytorch_model-00036-of-00056.bin", + "model.layers.25.post_attention_layernorm.weight": "pytorch_model-00036-of-00056.bin", + "model.layers.25.self_attn.k_proj.weight": "pytorch_model-00035-of-00056.bin", + "model.layers.25.self_attn.o_proj.weight": "pytorch_model-00035-of-00056.bin", + "model.layers.25.self_attn.q_proj.weight": "pytorch_model-00035-of-00056.bin", + "model.layers.25.self_attn.v_proj.weight": "pytorch_model-00035-of-00056.bin", + "model.layers.26.input_layernorm.weight": "pytorch_model-00037-of-00056.bin", + "model.layers.26.mlp.down_proj.weight": "pytorch_model-00037-of-00056.bin", + "model.layers.26.mlp.gate_proj.weight": "pytorch_model-00037-of-00056.bin", + "model.layers.26.mlp.up_proj.weight": "pytorch_model-00037-of-00056.bin", + "model.layers.26.post_attention_layernorm.weight": "pytorch_model-00037-of-00056.bin", + "model.layers.26.self_attn.k_proj.weight": "pytorch_model-00036-of-00056.bin", + "model.layers.26.self_attn.o_proj.weight": "pytorch_model-00036-of-00056.bin", + "model.layers.26.self_attn.q_proj.weight": "pytorch_model-00036-of-00056.bin", + "model.layers.26.self_attn.v_proj.weight": "pytorch_model-00036-of-00056.bin", + "model.layers.27.input_layernorm.weight": "pytorch_model-00039-of-00056.bin", + "model.layers.27.mlp.down_proj.weight": "pytorch_model-00039-of-00056.bin", + "model.layers.27.mlp.gate_proj.weight": "pytorch_model-00038-of-00056.bin", + "model.layers.27.mlp.up_proj.weight": "pytorch_model-00038-of-00056.bin", + "model.layers.27.post_attention_layernorm.weight": "pytorch_model-00039-of-00056.bin", + "model.layers.27.self_attn.k_proj.weight": "pytorch_model-00038-of-00056.bin", + "model.layers.27.self_attn.o_proj.weight": "pytorch_model-00038-of-00056.bin", + "model.layers.27.self_attn.q_proj.weight": "pytorch_model-00037-of-00056.bin", + "model.layers.27.self_attn.v_proj.weight": "pytorch_model-00038-of-00056.bin", + "model.layers.28.input_layernorm.weight": "pytorch_model-00040-of-00056.bin", + "model.layers.28.mlp.down_proj.weight": "pytorch_model-00040-of-00056.bin", + "model.layers.28.mlp.gate_proj.weight": "pytorch_model-00039-of-00056.bin", + "model.layers.28.mlp.up_proj.weight": "pytorch_model-00040-of-00056.bin", + "model.layers.28.post_attention_layernorm.weight": "pytorch_model-00040-of-00056.bin", + "model.layers.28.self_attn.k_proj.weight": "pytorch_model-00039-of-00056.bin", + "model.layers.28.self_attn.o_proj.weight": "pytorch_model-00039-of-00056.bin", + "model.layers.28.self_attn.q_proj.weight": "pytorch_model-00039-of-00056.bin", + "model.layers.28.self_attn.v_proj.weight": "pytorch_model-00039-of-00056.bin", + "model.layers.29.input_layernorm.weight": "pytorch_model-00041-of-00056.bin", + "model.layers.29.mlp.down_proj.weight": "pytorch_model-00041-of-00056.bin", + "model.layers.29.mlp.gate_proj.weight": "pytorch_model-00041-of-00056.bin", + "model.layers.29.mlp.up_proj.weight": "pytorch_model-00041-of-00056.bin", + "model.layers.29.post_attention_layernorm.weight": "pytorch_model-00041-of-00056.bin", + "model.layers.29.self_attn.k_proj.weight": "pytorch_model-00040-of-00056.bin", + "model.layers.29.self_attn.o_proj.weight": "pytorch_model-00040-of-00056.bin", + "model.layers.29.self_attn.q_proj.weight": "pytorch_model-00040-of-00056.bin", + "model.layers.29.self_attn.v_proj.weight": "pytorch_model-00040-of-00056.bin", + "model.layers.3.input_layernorm.weight": "pytorch_model-00007-of-00056.bin", + "model.layers.3.mlp.down_proj.weight": "pytorch_model-00007-of-00056.bin", + "model.layers.3.mlp.gate_proj.weight": "pytorch_model-00006-of-00056.bin", + "model.layers.3.mlp.up_proj.weight": "pytorch_model-00006-of-00056.bin", + "model.layers.3.post_attention_layernorm.weight": "pytorch_model-00007-of-00056.bin", + "model.layers.3.self_attn.k_proj.weight": "pytorch_model-00006-of-00056.bin", + "model.layers.3.self_attn.o_proj.weight": "pytorch_model-00006-of-00056.bin", + "model.layers.3.self_attn.q_proj.weight": "pytorch_model-00005-of-00056.bin", + "model.layers.3.self_attn.v_proj.weight": "pytorch_model-00006-of-00056.bin", + "model.layers.30.input_layernorm.weight": "pytorch_model-00043-of-00056.bin", + "model.layers.30.mlp.down_proj.weight": "pytorch_model-00043-of-00056.bin", + "model.layers.30.mlp.gate_proj.weight": "pytorch_model-00042-of-00056.bin", + "model.layers.30.mlp.up_proj.weight": "pytorch_model-00042-of-00056.bin", + "model.layers.30.post_attention_layernorm.weight": "pytorch_model-00043-of-00056.bin", + "model.layers.30.self_attn.k_proj.weight": "pytorch_model-00042-of-00056.bin", + "model.layers.30.self_attn.o_proj.weight": "pytorch_model-00042-of-00056.bin", + "model.layers.30.self_attn.q_proj.weight": "pytorch_model-00041-of-00056.bin", + "model.layers.30.self_attn.v_proj.weight": "pytorch_model-00042-of-00056.bin", + "model.layers.31.input_layernorm.weight": "pytorch_model-00044-of-00056.bin", + "model.layers.31.mlp.down_proj.weight": "pytorch_model-00044-of-00056.bin", + "model.layers.31.mlp.gate_proj.weight": "pytorch_model-00043-of-00056.bin", + "model.layers.31.mlp.up_proj.weight": "pytorch_model-00044-of-00056.bin", + "model.layers.31.post_attention_layernorm.weight": "pytorch_model-00044-of-00056.bin", + "model.layers.31.self_attn.k_proj.weight": "pytorch_model-00043-of-00056.bin", + "model.layers.31.self_attn.o_proj.weight": "pytorch_model-00043-of-00056.bin", + "model.layers.31.self_attn.q_proj.weight": "pytorch_model-00043-of-00056.bin", + "model.layers.31.self_attn.v_proj.weight": "pytorch_model-00043-of-00056.bin", + "model.layers.32.input_layernorm.weight": "pytorch_model-00045-of-00056.bin", + "model.layers.32.mlp.down_proj.weight": "pytorch_model-00045-of-00056.bin", + "model.layers.32.mlp.gate_proj.weight": "pytorch_model-00045-of-00056.bin", + "model.layers.32.mlp.up_proj.weight": "pytorch_model-00045-of-00056.bin", + "model.layers.32.post_attention_layernorm.weight": "pytorch_model-00045-of-00056.bin", + "model.layers.32.self_attn.k_proj.weight": "pytorch_model-00044-of-00056.bin", + "model.layers.32.self_attn.o_proj.weight": "pytorch_model-00044-of-00056.bin", + "model.layers.32.self_attn.q_proj.weight": "pytorch_model-00044-of-00056.bin", + "model.layers.32.self_attn.v_proj.weight": "pytorch_model-00044-of-00056.bin", + "model.layers.33.input_layernorm.weight": "pytorch_model-00047-of-00056.bin", + "model.layers.33.mlp.down_proj.weight": "pytorch_model-00047-of-00056.bin", + "model.layers.33.mlp.gate_proj.weight": "pytorch_model-00046-of-00056.bin", + "model.layers.33.mlp.up_proj.weight": "pytorch_model-00046-of-00056.bin", + "model.layers.33.post_attention_layernorm.weight": "pytorch_model-00047-of-00056.bin", + "model.layers.33.self_attn.k_proj.weight": "pytorch_model-00046-of-00056.bin", + "model.layers.33.self_attn.o_proj.weight": "pytorch_model-00046-of-00056.bin", + "model.layers.33.self_attn.q_proj.weight": "pytorch_model-00045-of-00056.bin", + "model.layers.33.self_attn.v_proj.weight": "pytorch_model-00046-of-00056.bin", + "model.layers.34.input_layernorm.weight": "pytorch_model-00048-of-00056.bin", + "model.layers.34.mlp.down_proj.weight": "pytorch_model-00048-of-00056.bin", + "model.layers.34.mlp.gate_proj.weight": "pytorch_model-00047-of-00056.bin", + "model.layers.34.mlp.up_proj.weight": "pytorch_model-00048-of-00056.bin", + "model.layers.34.post_attention_layernorm.weight": "pytorch_model-00048-of-00056.bin", + "model.layers.34.self_attn.k_proj.weight": "pytorch_model-00047-of-00056.bin", + "model.layers.34.self_attn.o_proj.weight": "pytorch_model-00047-of-00056.bin", + "model.layers.34.self_attn.q_proj.weight": "pytorch_model-00047-of-00056.bin", + "model.layers.34.self_attn.v_proj.weight": "pytorch_model-00047-of-00056.bin", + "model.layers.35.input_layernorm.weight": "pytorch_model-00049-of-00056.bin", + "model.layers.35.mlp.down_proj.weight": "pytorch_model-00049-of-00056.bin", + "model.layers.35.mlp.gate_proj.weight": "pytorch_model-00049-of-00056.bin", + "model.layers.35.mlp.up_proj.weight": "pytorch_model-00049-of-00056.bin", + "model.layers.35.post_attention_layernorm.weight": "pytorch_model-00049-of-00056.bin", + "model.layers.35.self_attn.k_proj.weight": "pytorch_model-00048-of-00056.bin", + "model.layers.35.self_attn.o_proj.weight": "pytorch_model-00048-of-00056.bin", + "model.layers.35.self_attn.q_proj.weight": "pytorch_model-00048-of-00056.bin", + "model.layers.35.self_attn.v_proj.weight": "pytorch_model-00048-of-00056.bin", + "model.layers.36.input_layernorm.weight": "pytorch_model-00051-of-00056.bin", + "model.layers.36.mlp.down_proj.weight": "pytorch_model-00051-of-00056.bin", + "model.layers.36.mlp.gate_proj.weight": "pytorch_model-00050-of-00056.bin", + "model.layers.36.mlp.up_proj.weight": "pytorch_model-00050-of-00056.bin", + "model.layers.36.post_attention_layernorm.weight": "pytorch_model-00051-of-00056.bin", + "model.layers.36.self_attn.k_proj.weight": "pytorch_model-00050-of-00056.bin", + "model.layers.36.self_attn.o_proj.weight": "pytorch_model-00050-of-00056.bin", + "model.layers.36.self_attn.q_proj.weight": "pytorch_model-00049-of-00056.bin", + "model.layers.36.self_attn.v_proj.weight": "pytorch_model-00050-of-00056.bin", + "model.layers.37.input_layernorm.weight": "pytorch_model-00052-of-00056.bin", + "model.layers.37.mlp.down_proj.weight": "pytorch_model-00052-of-00056.bin", + "model.layers.37.mlp.gate_proj.weight": "pytorch_model-00051-of-00056.bin", + "model.layers.37.mlp.up_proj.weight": "pytorch_model-00052-of-00056.bin", + "model.layers.37.post_attention_layernorm.weight": "pytorch_model-00052-of-00056.bin", + "model.layers.37.self_attn.k_proj.weight": "pytorch_model-00051-of-00056.bin", + "model.layers.37.self_attn.o_proj.weight": "pytorch_model-00051-of-00056.bin", + "model.layers.37.self_attn.q_proj.weight": "pytorch_model-00051-of-00056.bin", + "model.layers.37.self_attn.v_proj.weight": "pytorch_model-00051-of-00056.bin", + "model.layers.38.input_layernorm.weight": "pytorch_model-00053-of-00056.bin", + "model.layers.38.mlp.down_proj.weight": "pytorch_model-00053-of-00056.bin", + "model.layers.38.mlp.gate_proj.weight": "pytorch_model-00053-of-00056.bin", + "model.layers.38.mlp.up_proj.weight": "pytorch_model-00053-of-00056.bin", + "model.layers.38.post_attention_layernorm.weight": "pytorch_model-00053-of-00056.bin", + "model.layers.38.self_attn.k_proj.weight": "pytorch_model-00052-of-00056.bin", + "model.layers.38.self_attn.o_proj.weight": "pytorch_model-00052-of-00056.bin", + "model.layers.38.self_attn.q_proj.weight": "pytorch_model-00052-of-00056.bin", + "model.layers.38.self_attn.v_proj.weight": "pytorch_model-00052-of-00056.bin", + "model.layers.39.input_layernorm.weight": "pytorch_model-00055-of-00056.bin", + "model.layers.39.mlp.down_proj.weight": "pytorch_model-00055-of-00056.bin", + "model.layers.39.mlp.gate_proj.weight": "pytorch_model-00054-of-00056.bin", + "model.layers.39.mlp.up_proj.weight": "pytorch_model-00054-of-00056.bin", + "model.layers.39.post_attention_layernorm.weight": "pytorch_model-00055-of-00056.bin", + "model.layers.39.self_attn.k_proj.weight": "pytorch_model-00054-of-00056.bin", + "model.layers.39.self_attn.o_proj.weight": "pytorch_model-00054-of-00056.bin", + "model.layers.39.self_attn.q_proj.weight": "pytorch_model-00053-of-00056.bin", + "model.layers.39.self_attn.v_proj.weight": "pytorch_model-00054-of-00056.bin", + "model.layers.4.input_layernorm.weight": "pytorch_model-00008-of-00056.bin", + "model.layers.4.mlp.down_proj.weight": "pytorch_model-00008-of-00056.bin", + "model.layers.4.mlp.gate_proj.weight": "pytorch_model-00007-of-00056.bin", + "model.layers.4.mlp.up_proj.weight": "pytorch_model-00008-of-00056.bin", + "model.layers.4.post_attention_layernorm.weight": "pytorch_model-00008-of-00056.bin", + "model.layers.4.self_attn.k_proj.weight": "pytorch_model-00007-of-00056.bin", + "model.layers.4.self_attn.o_proj.weight": "pytorch_model-00007-of-00056.bin", + "model.layers.4.self_attn.q_proj.weight": "pytorch_model-00007-of-00056.bin", + "model.layers.4.self_attn.v_proj.weight": "pytorch_model-00007-of-00056.bin", + "model.layers.5.input_layernorm.weight": "pytorch_model-00009-of-00056.bin", + "model.layers.5.mlp.down_proj.weight": "pytorch_model-00009-of-00056.bin", + "model.layers.5.mlp.gate_proj.weight": "pytorch_model-00009-of-00056.bin", + "model.layers.5.mlp.up_proj.weight": "pytorch_model-00009-of-00056.bin", + "model.layers.5.post_attention_layernorm.weight": "pytorch_model-00009-of-00056.bin", + "model.layers.5.self_attn.k_proj.weight": "pytorch_model-00008-of-00056.bin", + "model.layers.5.self_attn.o_proj.weight": "pytorch_model-00008-of-00056.bin", + "model.layers.5.self_attn.q_proj.weight": "pytorch_model-00008-of-00056.bin", + "model.layers.5.self_attn.v_proj.weight": "pytorch_model-00008-of-00056.bin", + "model.layers.6.input_layernorm.weight": "pytorch_model-00011-of-00056.bin", + "model.layers.6.mlp.down_proj.weight": "pytorch_model-00011-of-00056.bin", + "model.layers.6.mlp.gate_proj.weight": "pytorch_model-00010-of-00056.bin", + "model.layers.6.mlp.up_proj.weight": "pytorch_model-00010-of-00056.bin", + "model.layers.6.post_attention_layernorm.weight": "pytorch_model-00011-of-00056.bin", + "model.layers.6.self_attn.k_proj.weight": "pytorch_model-00010-of-00056.bin", + "model.layers.6.self_attn.o_proj.weight": "pytorch_model-00010-of-00056.bin", + "model.layers.6.self_attn.q_proj.weight": "pytorch_model-00009-of-00056.bin", + "model.layers.6.self_attn.v_proj.weight": "pytorch_model-00010-of-00056.bin", + "model.layers.7.input_layernorm.weight": "pytorch_model-00012-of-00056.bin", + "model.layers.7.mlp.down_proj.weight": "pytorch_model-00012-of-00056.bin", + "model.layers.7.mlp.gate_proj.weight": "pytorch_model-00011-of-00056.bin", + "model.layers.7.mlp.up_proj.weight": "pytorch_model-00012-of-00056.bin", + "model.layers.7.post_attention_layernorm.weight": "pytorch_model-00012-of-00056.bin", + "model.layers.7.self_attn.k_proj.weight": "pytorch_model-00011-of-00056.bin", + "model.layers.7.self_attn.o_proj.weight": "pytorch_model-00011-of-00056.bin", + "model.layers.7.self_attn.q_proj.weight": "pytorch_model-00011-of-00056.bin", + "model.layers.7.self_attn.v_proj.weight": "pytorch_model-00011-of-00056.bin", + "model.layers.8.input_layernorm.weight": "pytorch_model-00013-of-00056.bin", + "model.layers.8.mlp.down_proj.weight": "pytorch_model-00013-of-00056.bin", + "model.layers.8.mlp.gate_proj.weight": "pytorch_model-00013-of-00056.bin", + "model.layers.8.mlp.up_proj.weight": "pytorch_model-00013-of-00056.bin", + "model.layers.8.post_attention_layernorm.weight": "pytorch_model-00013-of-00056.bin", + "model.layers.8.self_attn.k_proj.weight": "pytorch_model-00012-of-00056.bin", + "model.layers.8.self_attn.o_proj.weight": "pytorch_model-00012-of-00056.bin", + "model.layers.8.self_attn.q_proj.weight": "pytorch_model-00012-of-00056.bin", + "model.layers.8.self_attn.v_proj.weight": "pytorch_model-00012-of-00056.bin", + "model.layers.9.input_layernorm.weight": "pytorch_model-00015-of-00056.bin", + "model.layers.9.mlp.down_proj.weight": "pytorch_model-00015-of-00056.bin", + "model.layers.9.mlp.gate_proj.weight": "pytorch_model-00014-of-00056.bin", + "model.layers.9.mlp.up_proj.weight": "pytorch_model-00014-of-00056.bin", + "model.layers.9.post_attention_layernorm.weight": "pytorch_model-00015-of-00056.bin", + "model.layers.9.self_attn.k_proj.weight": "pytorch_model-00014-of-00056.bin", + "model.layers.9.self_attn.o_proj.weight": "pytorch_model-00014-of-00056.bin", + "model.layers.9.self_attn.q_proj.weight": "pytorch_model-00013-of-00056.bin", + "model.layers.9.self_attn.v_proj.weight": "pytorch_model-00014-of-00056.bin", + "model.norm.weight": "pytorch_model-00055-of-00056.bin" + } +}