Hugging Face's logo

JackBAI
/
roberta-medium

Fill-Mask
Transformers
PyTorch
English
roberta
Inference Endpoints
Model card Files Community
Yuexiang Zhai commited on
Commit
39298c3
•
1 Parent(s): e5848eb

push

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. config.json +27 -0
  2. global_step30000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +3 -0
  3. global_step30000/bf16_zero_pp_rank_10_mp_rank_00_optim_states.pt +3 -0
  4. global_step30000/bf16_zero_pp_rank_11_mp_rank_00_optim_states.pt +3 -0
  5. global_step30000/bf16_zero_pp_rank_12_mp_rank_00_optim_states.pt +3 -0
  6. global_step30000/bf16_zero_pp_rank_13_mp_rank_00_optim_states.pt +3 -0
  7. global_step30000/bf16_zero_pp_rank_14_mp_rank_00_optim_states.pt +3 -0
  8. global_step30000/bf16_zero_pp_rank_15_mp_rank_00_optim_states.pt +3 -0
  9. global_step30000/bf16_zero_pp_rank_16_mp_rank_00_optim_states.pt +3 -0
  10. global_step30000/bf16_zero_pp_rank_17_mp_rank_00_optim_states.pt +3 -0
  11. global_step30000/bf16_zero_pp_rank_18_mp_rank_00_optim_states.pt +3 -0
  12. global_step30000/bf16_zero_pp_rank_19_mp_rank_00_optim_states.pt +3 -0
  13. global_step30000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt +3 -0
  14. global_step30000/bf16_zero_pp_rank_20_mp_rank_00_optim_states.pt +3 -0
  15. global_step30000/bf16_zero_pp_rank_21_mp_rank_00_optim_states.pt +3 -0
  16. global_step30000/bf16_zero_pp_rank_22_mp_rank_00_optim_states.pt +3 -0
  17. global_step30000/bf16_zero_pp_rank_23_mp_rank_00_optim_states.pt +3 -0
  18. global_step30000/bf16_zero_pp_rank_24_mp_rank_00_optim_states.pt +3 -0
  19. global_step30000/bf16_zero_pp_rank_25_mp_rank_00_optim_states.pt +3 -0
  20. global_step30000/bf16_zero_pp_rank_26_mp_rank_00_optim_states.pt +3 -0
  21. global_step30000/bf16_zero_pp_rank_27_mp_rank_00_optim_states.pt +3 -0
  22. global_step30000/bf16_zero_pp_rank_28_mp_rank_00_optim_states.pt +3 -0
  23. global_step30000/bf16_zero_pp_rank_29_mp_rank_00_optim_states.pt +3 -0
  24. global_step30000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt +3 -0
  25. global_step30000/bf16_zero_pp_rank_30_mp_rank_00_optim_states.pt +3 -0
  26. global_step30000/bf16_zero_pp_rank_31_mp_rank_00_optim_states.pt +3 -0
  27. global_step30000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt +3 -0
  28. global_step30000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt +3 -0
  29. global_step30000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt +3 -0
  30. global_step30000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt +3 -0
  31. global_step30000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt +3 -0
  32. global_step30000/bf16_zero_pp_rank_8_mp_rank_00_optim_states.pt +3 -0
  33. global_step30000/bf16_zero_pp_rank_9_mp_rank_00_optim_states.pt +3 -0
  34. global_step30000/mp_rank_00_model_states.pt +3 -0
  35. latest +1 -0
  36. merges.txt +0 -0
  37. pytorch_model.bin +3 -0
  38. rng_state_0.pth +3 -0
  39. rng_state_1.pth +3 -0
  40. rng_state_10.pth +3 -0
  41. rng_state_11.pth +3 -0
  42. rng_state_12.pth +3 -0
  43. rng_state_13.pth +3 -0
  44. rng_state_14.pth +3 -0
  45. rng_state_15.pth +3 -0
  46. rng_state_16.pth +3 -0
  47. rng_state_17.pth +3 -0
  48. rng_state_18.pth +3 -0
  49. rng_state_19.pth +3 -0
  50. rng_state_2.pth +3 -0
config.json ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "model_configs/roberta_medium.json",
3
+ "architectures": [
4
+ "RobertaForMaskedLM"
5
+ ],
6
+ "attention_probs_dropout_prob": 0.1,
7
+ "bos_token_id": 0,
8
+ "classifier_dropout": null,
9
+ "eos_token_id": 2,
10
+ "hidden_act": "gelu",
11
+ "hidden_dropout_prob": 0.1,
12
+ "hidden_size": 512,
13
+ "initializer_range": 0.02,
14
+ "intermediate_size": 2048,
15
+ "layer_norm_eps": 1e-05,
16
+ "max_position_embeddings": 514,
17
+ "model_type": "roberta",
18
+ "num_attention_heads": 8,
19
+ "num_hidden_layers": 8,
20
+ "pad_token_id": 1,
21
+ "position_embedding_type": "absolute",
22
+ "torch_dtype": "bfloat16",
23
+ "transformers_version": "4.32.0.dev0",
24
+ "type_vocab_size": 1,
25
+ "use_cache": true,
26
+ "vocab_size": 50265
27
+ }
global_step30000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:416ec51ca207b5626bce27c3dce5b2b1285e3312c1b30be3baf62426c21e90a8
3
+ size 19329335
global_step30000/bf16_zero_pp_rank_10_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6635ae1d61927f6627ad4d3534ab1e2b300fe3fac4e80c9cae86076b1b86e49f
3
+ size 19329471
global_step30000/bf16_zero_pp_rank_11_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1d232f93f3b90298965226b622e1504b898a28a74c10e6eaab1761bccc63ec73
3
+ size 19329151
global_step30000/bf16_zero_pp_rank_12_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:20553c65c019c8acc92f11835a6017dacde0b9aea6ea79cdf36414bfa9d44ae7
3
+ size 19329471
global_step30000/bf16_zero_pp_rank_13_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d066b7436276c0ce0d18964e2446cba7eda0c8699bde2941c7f487e72db2ad39
3
+ size 19329151
global_step30000/bf16_zero_pp_rank_14_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:570d2d2a352c903b4de49bb1a60922bc24134a100b3f4d1a27894ebbdbc6b79c
3
+ size 19329471
global_step30000/bf16_zero_pp_rank_15_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c577fe5c777d4b494b0cc71355ae4905e79237c4597824291b9442cd0bbcbc37
3
+ size 19329151
global_step30000/bf16_zero_pp_rank_16_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:87d9fc60231623eb32abd7f7eb4aa51f1ebebd82adb16d6dd3c38c0855269a0d
3
+ size 19329663
global_step30000/bf16_zero_pp_rank_17_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:22a44f3560efcbfee78049dc9839ed9429296a724c70738635d558b9526320e5
3
+ size 19328959
global_step30000/bf16_zero_pp_rank_18_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:343ce030e1865add771d779c6887e6365549045d847f404526c03fec92f4668f
3
+ size 19329279
global_step30000/bf16_zero_pp_rank_19_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f0dccd4d2aa2a70a5019dddbc01e407ed1e47d9a0c0df668a5699862d12b2637
3
+ size 19328959
global_step30000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c709013239f11a1f88e43d8fedfe4626a1c0a433298f0823d05493e079b0556f
3
+ size 19329143
global_step30000/bf16_zero_pp_rank_20_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:10daab721d813a407253445810d075b04a939f0e03ea874ea5e5c88a0f0b0610
3
+ size 19329279
global_step30000/bf16_zero_pp_rank_21_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e433b272f6a1f533bbbe00f489d26878d4446a1032235c95d31faa8b6f050cb2
3
+ size 19328959
global_step30000/bf16_zero_pp_rank_22_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:59d0ea5d31f3e6d159ad9bd89b9206e853efe0cf8e21dab94cb8a297ddd756f6
3
+ size 19329279
global_step30000/bf16_zero_pp_rank_23_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0988846d93a2ebfa9b58af2115703fac8f3903eb3d399cff3c5911e6b51fe4ba
3
+ size 19328959
global_step30000/bf16_zero_pp_rank_24_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2e368a3e8da908392478b27208b9d61670ee1caae5a8b02387cfffba4c24de06
3
+ size 19329279
global_step30000/bf16_zero_pp_rank_25_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4336522e9720e0e4faac9b1a76096306f41e68f561a78d5dd6995eaed2e7b5fc
3
+ size 19329023
global_step30000/bf16_zero_pp_rank_26_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3e6f6f8d21a0f6fcec7e079a714c55eeedeaf05182b5b9cea73d1ec2a1c8546f
3
+ size 19329215
global_step30000/bf16_zero_pp_rank_27_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8d9e97bd397f3e1ae2138bb034eb59fa2ff1f51acdf3ae0745d96220b2d1063c
3
+ size 19329023
global_step30000/bf16_zero_pp_rank_28_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8817482b822b3d644efa67a9ddff4817248f1fa6f9617e9167b47fd624c4db02
3
+ size 19329215
global_step30000/bf16_zero_pp_rank_29_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:39103efbfc0eface3d53aa17f5b784bd7d0e154707b24192d8f4b9481ecae577
3
+ size 19329023
global_step30000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:408ef4921b6ea8143d8c310b8e298464eb230789554c6982a520aae8a9634574
3
+ size 19329463
global_step30000/bf16_zero_pp_rank_30_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fc0a7aa4cdd530975368cc498841a8cea5afbce3b26bee8011bca93f34e551e2
3
+ size 19329215
global_step30000/bf16_zero_pp_rank_31_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4217c1fbfbbc1aa38fdd7d002b0b067654fd6d3882ef5fa67580114abd82804f
3
+ size 19329151
global_step30000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f6ce482ea85b4d03d9bdba1912a08bbc2060ee189c060cbcf04de56188854129
3
+ size 19329143
global_step30000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ac5f1e47eee84c98e3f2a95cf0fcb66f7bb988c2145ea3247b7c5dff6bed1212
3
+ size 19329463
global_step30000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:974a0365d2186b32e4123ef33901102afe45129594baa677c5c8c2652d4616b7
3
+ size 19329143
global_step30000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6ca07d81a4f2ea2cc835b2b2eeda047cb05cb0816163d03bbc37046d5878290f
3
+ size 19329463
global_step30000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7c2c54f79728ad72e9594b0e3283ab3d204a7c5ce7b1faff7dde944f2f42ed61
3
+ size 19329143
global_step30000/bf16_zero_pp_rank_8_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:018b34b379bbcf096a3378eba9645e340703b16fcad27b5be78381e38f579ffa
3
+ size 19329463
global_step30000/bf16_zero_pp_rank_9_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e2c59521377f23025f9bcf33db538a6e03df47ba8ebcbdadf2e50acae675c617
3
+ size 19329143
global_step30000/mp_rank_00_model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9848809431a3c21c54a4acb80edb5465905443b2c085b5f273b182521fb624f9
3
+ size 103109543
latest ADDED
@@ -0,0 +1 @@
 
 
1
+ global_step30000
merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:df9d2aec3b51894913abad54e93ece9e5197d644d480870ef3d4d5b95f8500da
3
+ size 154684413
rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6469549ec6e92d5607a72ff345d2eddf5ca471ea5bb2d49fc058778733ecad77
3
+ size 21687
rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c0fafd385807e4678dedd6c84244aab9e071de5de64fdf04b352cea364fe7c1c
3
+ size 21687
rng_state_10.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2cd9ad0aa78084e5e3a192f60cbafee3b8cb4959b211952b65d79e8441715e59
3
+ size 21698
rng_state_11.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eea42f6f08b54e6bfe83b13777912d75404da9dd9b42deae4def374fa61b5286
3
+ size 21698
rng_state_12.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3c840b9dbb021c4fc1b7d6e8457f349722748ccb38f69f647cdb8f468379bd9b
3
+ size 21698
rng_state_13.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9c5b6a7166ca751f241a3e570b26eee3999f40251a42b863f608b17da3e01387
3
+ size 21698
rng_state_14.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c9f3f6a580fdee419ba90237d8221d755b1e3a234176d2c96c6bc852100b73a3
3
+ size 21698
rng_state_15.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f56e4684216d72eeba1332e6317b55cac09e372d1750704e1c5204c02cb86b8e
3
+ size 21698
rng_state_16.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9cdb09e20205c8c4708bcfd00b13793303eede638da87f23dbd365068fc77bb0
3
+ size 21698
rng_state_17.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:68c5ce4eabfe25b073b7f4bdff1754d601db5669ab1597050243d17d455642d1
3
+ size 21698
rng_state_18.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3727c9e53671ba713f502865f1d5039f6e4dca068a656e0dcaedd1f78f0cbc5b
3
+ size 21698
rng_state_19.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:572caf8a893f46f063b13a6d64a443f384ba5eb37be753832af948f79c37256b
3
+ size 21698
rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c747ac806c925aab1f39ab62b3c64eba03184a664c33b205ab56a3cd6520d1bc
3
+ size 21687