diff --git a/README.md b/README.md index 3ecf66b011f3978a2dfcf4a0eba80029dcbb7cec..696ee1c9728410c176afb2b837850a07d9989d55 100644 --- a/README.md +++ b/README.md @@ -33,7 +33,7 @@ More information needed ### Training hyperparameters The following hyperparameters were used during training: -- learning_rate: 0.0001 +- learning_rate: 5e-05 - train_batch_size: 2 - eval_batch_size: 8 - seed: 42 @@ -43,8 +43,8 @@ The following hyperparameters were used during training: - total_eval_batch_size: 256 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear -- lr_scheduler_warmup_steps: 150 -- num_epochs: 3 +- lr_scheduler_warmup_steps: 100 +- num_epochs: 1 ### Training results diff --git a/adapter_config.json b/adapter_config.json index 1262421d7ae3daf679d445b35b8582d17a084013..97f785b4b0ab92aba6b516d28e02fa6041331d8d 100644 --- a/adapter_config.json +++ b/adapter_config.json @@ -12,26 +12,26 @@ "layers_pattern": null, "layers_to_transform": null, "loftq_config": {}, - "lora_alpha": 8, + "lora_alpha": 4, "lora_dropout": 0.1, "megatron_config": null, "megatron_core": "megatron.core", "modules_to_save": null, "peft_type": "LORA", - "r": 8, + "r": 4, "rank_pattern": {}, "revision": null, "target_modules": [ - "fc2", - "v_proj", "o_proj", + "gate_proj", + "fc1", + "down_proj", + "out_proj", "q_proj", "up_proj", - "out_proj", "k_proj", - "fc1", - "gate_proj", - "down_proj" + "fc2", + "v_proj" ], "task_type": null, "use_dora": true, diff --git a/global_step617/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..1eecfead68b8204159f7b3b32ea768f0ad329c84 --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bbd84e436df35b33e6edcc8eba1c5b2751394d75634b14d22c1086e1a74fb22c +size 5876663 diff --git a/global_step617/bf16_zero_pp_rank_10_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_10_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..2cea880e4890e44e9d7646a467799697e459abd7 --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_10_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:11d4eabf05d69836292253f5334a93eaee4d1f4be46464cd0026d191e0a2b702 +size 5876669 diff --git a/global_step617/bf16_zero_pp_rank_11_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_11_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..2018cdfb74235ef9768f230e14c61ce2ad2aeddb --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_11_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:24fc7bd53fae81a8e19e6d384464e3d3a0189b2af0f07b5a4481e1874f69a361 +size 5876669 diff --git a/global_step617/bf16_zero_pp_rank_12_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_12_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..ed2f011af40e70c5084bde5a7c68593e9c499511 --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_12_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:126e4005199fddf43db2d66aa181cc2f74b9eff0e33513f4919bb34a2dacf910 +size 5876669 diff --git a/global_step617/bf16_zero_pp_rank_13_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_13_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..9b06a13a6f1f9ee84eab29cb0201287a26092db2 --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_13_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b9b818434e86594166c3aca97a96e4884c6baf261345e5d4629e52a7d8a5268f +size 5876669 diff --git a/global_step617/bf16_zero_pp_rank_14_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_14_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..12b3409d28d49e1c719132deb3fcda1e4cdf4bbc --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_14_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f4045e410eb6f228a5cdf0d6080fd28f30bcd5b5f62d114ebf65f4d126fb96cf +size 5876669 diff --git a/global_step617/bf16_zero_pp_rank_15_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_15_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..a0924722659c99ecfedc0486b8b2a62b4c86f45a --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_15_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3f995f054c97456fc95222bab96fc8ac3e20582c20d07e370158d3ad8046b780 +size 5876669 diff --git a/global_step617/bf16_zero_pp_rank_16_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_16_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..cb9c867558e7d0f7df72f93499afa74645d72f58 --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_16_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1c5b3a781b9906afe6d0f48644ed80e405f6b3561ec588a2cbfa60de1bc81f83 +size 5876669 diff --git a/global_step617/bf16_zero_pp_rank_17_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_17_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..e8b1738962ed1887d26d4d2bda56a75de4fcafc9 --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_17_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aff3850c19d6f4e1d38e4304e4520871f9862f715be39b4e1ca220ecdd185c3d +size 5876669 diff --git a/global_step617/bf16_zero_pp_rank_18_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_18_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..db832cbde0653c14d57d418e8fd404a2bb3afe1d --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_18_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5728f9ea5ead4943be86bfa79a46a2ecd69450ce528fb09cb42a08578570afde +size 5876669 diff --git a/global_step617/bf16_zero_pp_rank_19_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_19_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..8f2d0813f03852d9aafdd6c74e5b82a4ef8a1ef4 --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_19_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9c427f189f4bb31ac43508757f8481ef38328eaeb0b986d2b373ca89e45172da +size 5876669 diff --git a/global_step617/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..b120123b3b1106928dd487d3ac13a2903b027d3b --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e06d86a99ad1ef585e6a9ce7af44b36110adfd627f027211b684fe61b547113d +size 5876663 diff --git a/global_step617/bf16_zero_pp_rank_20_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_20_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..abfbd907465597df7181718ae1543da01f063be3 --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_20_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a64b3d5954211f277daa176974437c83f40bf1d2bb710dfaab18581724559147 +size 5876669 diff --git a/global_step617/bf16_zero_pp_rank_21_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_21_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..c3bd95378a391af527fab34c84cf5caccfd9e53d --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_21_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9507b4a9a90fad0eae809cc96ea5d4a30bad6cbfd706ba7693bc168f94458aec +size 5876669 diff --git a/global_step617/bf16_zero_pp_rank_22_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_22_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..ac043481e26bddd21feee5a7c13516e83edfc869 --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_22_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:540446a9611f2cefcd88f53990fdcf92a265b8a740568de1ac9c182d8d2073d3 +size 5876669 diff --git a/global_step617/bf16_zero_pp_rank_23_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_23_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..0285ed2f9a19a1f31e597bc4df7cfb3b2cd89c4e --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_23_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8d70cf53d540b642ec77eb3926e26062b6298bfa4d31f8eb8417620e615bb493 +size 5876669 diff --git a/global_step617/bf16_zero_pp_rank_24_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_24_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..908314b0ab9467900db9abf8d6675ad82aa3501f --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_24_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9385ee9b217e5e022fc2ece597e2cb809b7f4d73ff3e6e9b3137256d9887e495 +size 5876669 diff --git a/global_step617/bf16_zero_pp_rank_25_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_25_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..ee1d213fce419f9f46fd78ba8ee010bb80a01c56 --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_25_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d881b8e796de9a1fb23c9408026baf277e07af83d2bfae9b5bdd97f0d09e6290 +size 5876669 diff --git a/global_step617/bf16_zero_pp_rank_26_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_26_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..5c6cdc2f537d005b78b586df00ceed1971e02fc4 --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_26_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1d0fe1b1a32bfaecfeb81a818bc57121cfffa49d96510f9819bf86ce59664645 +size 5876669 diff --git a/global_step617/bf16_zero_pp_rank_27_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_27_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..dbd2e07f10f5c6b551f8a94cf56a6895de783380 --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_27_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d90cd4df2bd14ff45b9e55a36383b3b9467ae1ddd9ba48f654c23778fd3a7e39 +size 5876669 diff --git a/global_step617/bf16_zero_pp_rank_28_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_28_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..7d6965e185f528ae7fca1b7faa3ecc67bd776d92 --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_28_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e50cbb9e8bf91a017e515e920e22475a2546a25513a106a37b29f6ea843200a9 +size 5876669 diff --git a/global_step617/bf16_zero_pp_rank_29_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_29_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..c4f72172fdc5c6a1ce89132bdcd1cb9364fc175b --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_29_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8d308dcf06cb24f33e4d485269f08bbe04f764f9465cf44b18c2803e40a226bf +size 5876669 diff --git a/global_step617/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..2e9c9706a99889d37a2f779eec6a3ea6d073d69f --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2333faff5d517a129bd9045a725649a7b063d7a804373daade434826187fbc70 +size 5876663 diff --git a/global_step617/bf16_zero_pp_rank_30_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_30_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..1460be7908e1f9e04fbb0b09621ecf687b26e623 --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_30_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8edde03ed899030be09e83b4fc32efa0fb41c5505dff86c984837e27752d670e +size 5876669 diff --git a/global_step617/bf16_zero_pp_rank_31_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_31_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..e4922f8efa6957722fed060341aa9f4083fcfe30 --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_31_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1acf244496607c1965b454b818c200d9050fc95c0ccc64a1ab160f2f9b5d5b8f +size 5876669 diff --git a/global_step617/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..0d3472a6050bb586d5443811ece74e45d0f35352 --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a73b1ec5fb02d20782048dc035e34efea9bee4433dc5ceb7f25c223f870b76f3 +size 5876663 diff --git a/global_step617/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..71f76551304601542c132e28604e7e93f1e43310 --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c34164a3a58fe7d51d18f5a2db2abd7ca655c96189ba33ee06a902322eda83e8 +size 5876663 diff --git a/global_step617/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..f4d64b8d6a920743340d3d93eedb45688e7ee89c --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ce7e227d0fa52de5db7ba5b6c74a1ae0f0e437f9e789fe9c14a3245664c24f1 +size 5876663 diff --git a/global_step617/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..bae32a01d40d551528a96c0595180bdf8af14b31 --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:84cf1f1393d1cd8a3ec0b5710e5461de896cda88edc0d7a530b080ce4c87fdce +size 5876663 diff --git a/global_step617/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..76a543aa566131fe5c2e8c7b60f40e69dd22b791 --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c93f752e6ed78e26d7901a375e2a62d3ccf6ec90af28da48475eaf8e36f0951e +size 5876663 diff --git a/global_step617/bf16_zero_pp_rank_8_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_8_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..9fd59a09012407d8829e318dfcd1e0b4090b7db6 --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_8_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a1b896709fb82dc1c176127efb6e6b46c5c0e92a7185534e117bf2647a5664e2 +size 5876663 diff --git a/global_step617/bf16_zero_pp_rank_9_mp_rank_00_optim_states.pt b/global_step617/bf16_zero_pp_rank_9_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..e65c049683797daa2dc09d2f6da27aed50135893 --- /dev/null +++ b/global_step617/bf16_zero_pp_rank_9_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8a299dce16fa0210378e9156e6f3838cca85b1ccdbe94610422bbb660f7735dd +size 5876663 diff --git a/global_step617/zero_pp_rank_0_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_0_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..185ceab4b489e9ffb4cf4f50771b4c66408a063c --- /dev/null +++ b/global_step617/zero_pp_rank_0_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2648d72a036d2e021237bf73df96dc7ea24d8406fb51b1439e2fd82a48ec5478 +size 526932687 diff --git a/global_step617/zero_pp_rank_10_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_10_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..367e51c6f08e397642104557108282acda34ba7e --- /dev/null +++ b/global_step617/zero_pp_rank_10_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:38cc0db99c8e0e99543f7dd8aa6ce66f5a350f9b95d5a62a3fd4ea6692173fa9 +size 526935445 diff --git a/global_step617/zero_pp_rank_11_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_11_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..403a75b0a55db0c222ef5510c3153eaed86ba46b --- /dev/null +++ b/global_step617/zero_pp_rank_11_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:76424c2cef09fb837a98377999eb87a483f9a03f86073a5fd0a324abb35c7a31 +size 526935445 diff --git a/global_step617/zero_pp_rank_12_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_12_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..8f89af6a78f53392e244a0abaf24fba1c24a3a0c --- /dev/null +++ b/global_step617/zero_pp_rank_12_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:121df840b8716ed0b95b2d5052d0b2a12f5be956cc0574af74ff001ed605fdb9 +size 526935445 diff --git a/global_step617/zero_pp_rank_13_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_13_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..91a32eb0dd8909b637facf5519b8d332e01fa7fc --- /dev/null +++ b/global_step617/zero_pp_rank_13_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ca174c3599380ece8385ebe7f6c7fc6549f7af36f7ae532a6e190612344f919c +size 526935445 diff --git a/global_step617/zero_pp_rank_14_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_14_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..15283e6d7fb2610d6e0629a95deffa99398937d9 --- /dev/null +++ b/global_step617/zero_pp_rank_14_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0aae76a99efbd38a9fb50d548b5b9e593afe3245a353555e063db3fe8e669dfa +size 526935445 diff --git a/global_step617/zero_pp_rank_15_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_15_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..c60263dc1ba3637fb6597ffaa2ccba8bb8b35fc7 --- /dev/null +++ b/global_step617/zero_pp_rank_15_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a8b9e3f96b6dc515a16500f1b4ce4d7dd5d0e13ccf810370fbbd9234e8403fe7 +size 526935445 diff --git a/global_step617/zero_pp_rank_16_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_16_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..0418a95634596ee9c54ad4177acc13ad36742a96 --- /dev/null +++ b/global_step617/zero_pp_rank_16_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ef039d4bf2ebc432fead574307def58f2c416116b69305e0adbf97c889235686 +size 526935445 diff --git a/global_step617/zero_pp_rank_17_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_17_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..bd0ad9520b6e5d9a66f938f09a099bffc9dad1d7 --- /dev/null +++ b/global_step617/zero_pp_rank_17_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b6415f2f08f99b80d4c20f4c318d1d96cb4b59dad70c4f6aa94ef71270a36e52 +size 526935445 diff --git a/global_step617/zero_pp_rank_18_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_18_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..560a2bbf049dd5927e51db39967b02bf4799ce8c --- /dev/null +++ b/global_step617/zero_pp_rank_18_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fbcd7716d0379c587dd6b2a8388de8cbc92deaae72f862783bdecb835079f931 +size 526935445 diff --git a/global_step617/zero_pp_rank_19_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_19_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..78a22f16fda0ade90c4d7de17c3a08c99eedc6a1 --- /dev/null +++ b/global_step617/zero_pp_rank_19_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fdf5647a1c7c32e9432e678e87afac2a1b4390dc8194b291a7617d104ee19d1f +size 526935445 diff --git a/global_step617/zero_pp_rank_1_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_1_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..0780c8d1a6e398b6001ef2ea2748be17a10794a3 --- /dev/null +++ b/global_step617/zero_pp_rank_1_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cf3d1ec3db38a0ade0d4b5f45cbc2f75f3ef71194ab2fdcc5d4c3bcd43ce5e6f +size 526932687 diff --git a/global_step617/zero_pp_rank_20_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_20_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..7e9dd2e6d5abcd494878ba05cd31b9f8198bbb04 --- /dev/null +++ b/global_step617/zero_pp_rank_20_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:478bc995351b6d8070d2c36999a7240186768af043376814d58bb240ac8477b7 +size 526935445 diff --git a/global_step617/zero_pp_rank_21_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_21_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..cd86af3750b67e9ab35f18b52243117d8abbb85f --- /dev/null +++ b/global_step617/zero_pp_rank_21_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d58ab3564ebe7c5056c8761476be1a7d4cfebc4840320b4650311c42c77ffaae +size 526935445 diff --git a/global_step617/zero_pp_rank_22_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_22_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..ffc52848ccf63136729ae30c3a4490421adea3a0 --- /dev/null +++ b/global_step617/zero_pp_rank_22_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7ebabcc5800c98991dce48bf6f0e6407d82406152f24fa69cfad5017d65948e0 +size 526935445 diff --git a/global_step617/zero_pp_rank_23_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_23_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..50d2a4183c49097e9e09ead6985af763d1528aff --- /dev/null +++ b/global_step617/zero_pp_rank_23_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c0ff03d88777677b9aff6927ed6c60df1d17edc39d8d3cf912453d62e1988e87 +size 526935445 diff --git a/global_step617/zero_pp_rank_24_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_24_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..a4c80fb079791799690d9f972b7dde6e2ec84608 --- /dev/null +++ b/global_step617/zero_pp_rank_24_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d54a5fc72629ae6591d89ef2c4f9ddb3cc453bf03cd9f82c34e10883003b4277 +size 526935445 diff --git a/global_step617/zero_pp_rank_25_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_25_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..90ab9eb95e1f4dfc203ebca1ab5331a50605ac8e --- /dev/null +++ b/global_step617/zero_pp_rank_25_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d20dc099cb7a17d555eed8758e9ccc5bccb1f3292de8ac0ebdc8b6d493b21a13 +size 526935445 diff --git a/global_step617/zero_pp_rank_26_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_26_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..ce91e26b65113eacc4156c48def757b5325ec716 --- /dev/null +++ b/global_step617/zero_pp_rank_26_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ede8386e767606f9f24ce0af5e14d861b98551fba35b05bc2c281b1006f57c56 +size 526935445 diff --git a/global_step617/zero_pp_rank_27_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_27_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..caa78c87bf34267d9c0b95d0d5f47b50ab84e0f1 --- /dev/null +++ b/global_step617/zero_pp_rank_27_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c2ddfdce861aa664a75c3bfcec8e67735c65a2b7464d61e889340cff53a733e1 +size 526935445 diff --git a/global_step617/zero_pp_rank_28_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_28_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..c8432b7361d4a2646cc556ed532a452e29898710 --- /dev/null +++ b/global_step617/zero_pp_rank_28_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:516054386641a0d255d8f858712c7bb81d7b2ef1cb160c18d4c1b858a2149773 +size 526935445 diff --git a/global_step617/zero_pp_rank_29_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_29_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..de1725eec2594708ff6898d006ef1c4a6d2fc591 --- /dev/null +++ b/global_step617/zero_pp_rank_29_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:130d0b5d7582280b4a9ebf0089cc3b778d32c2d2c402b866588db3e53a25ab87 +size 526935445 diff --git a/global_step617/zero_pp_rank_2_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_2_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..a1bec6968f7e999f4f186bfcbba90f3e308a17bc --- /dev/null +++ b/global_step617/zero_pp_rank_2_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:018704edfc2a6f60e450cdf75babb436d7483e59840cd2972dab3390c913ada8 +size 526932687 diff --git a/global_step617/zero_pp_rank_30_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_30_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..df7eb56eb672e7f567a1996f7572049e458beba3 --- /dev/null +++ b/global_step617/zero_pp_rank_30_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:336a97c950405255d0ebcd14acb4c0c0f9f6573d0da6f3149a99e3c2eb8c565d +size 526935445 diff --git a/global_step617/zero_pp_rank_31_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_31_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..ce507c9738eb63ce51e9efad89ec979235e2b30c --- /dev/null +++ b/global_step617/zero_pp_rank_31_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e19a00c495c22b2252db14c8c9be09c48b2fb416e38a73dd16bc267f729769cf +size 526935445 diff --git a/global_step617/zero_pp_rank_3_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_3_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..fb9b193d16aa751b6c2ec860c2d9d04a82fd939c --- /dev/null +++ b/global_step617/zero_pp_rank_3_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:616a25f5ed75b572fed7bd99344f07a94138d8df18446298fcbdcdb214475dd9 +size 526932687 diff --git a/global_step617/zero_pp_rank_4_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_4_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..4e5452bdad671e8678d28f2e92577c7e6477c109 --- /dev/null +++ b/global_step617/zero_pp_rank_4_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:48323c7403e7cbdc45584ef619776ee8b53fc89bf16c4f783aae62e4cda8b7bd +size 526932687 diff --git a/global_step617/zero_pp_rank_5_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_5_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..0b85f43fc8455b3c5f0bc4c0235495beb4d4d299 --- /dev/null +++ b/global_step617/zero_pp_rank_5_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:637e7a13c0e5c7e916e956dacd5dd0f4e936805882967bbed3dbbad5cba90887 +size 526932687 diff --git a/global_step617/zero_pp_rank_6_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_6_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..952bcbb697026426bbd7dadbc7063cb71669ac8b --- /dev/null +++ b/global_step617/zero_pp_rank_6_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9b12999b165357d980f2642902b71f737f25ebe17d7afc762338bf898e191340 +size 526932687 diff --git a/global_step617/zero_pp_rank_7_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_7_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..97f2ed8f1b883f43487f50fd0f2be652ed65711b --- /dev/null +++ b/global_step617/zero_pp_rank_7_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a56e9c8658da96db70ac05cfaeab2ec5a362e0645a1c06dff906708ca414e5ed +size 526932687 diff --git a/global_step617/zero_pp_rank_8_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_8_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..84398f92bbf48dd086e7f3da3853ed248e420356 --- /dev/null +++ b/global_step617/zero_pp_rank_8_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ed94db9b4df426d62887d85a7752de1432c8d431f000c00822d80f431e5ed832 +size 526932687 diff --git a/global_step617/zero_pp_rank_9_mp_rank_00_model_states.pt b/global_step617/zero_pp_rank_9_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..93b1db02f3f1f3d2b27606789f290e9463aa10b7 --- /dev/null +++ b/global_step617/zero_pp_rank_9_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7fc15a0561e80216c812887833ef84b488198c680ee5b492171f84dc5d3b20c8 +size 526932687 diff --git a/latest b/latest index e943533aea1ffb04655dc58d931ae8c9b0a271e3..e3591976dc02bcf876afbbae81c6b15992f0b7cc 100644 --- a/latest +++ b/latest @@ -1 +1 @@ -global_step1851 \ No newline at end of file +global_step617 \ No newline at end of file diff --git a/training_args.bin b/training_args.bin index 8c6bfea32f706ff09901e4b16a277df410a110dd..37c58fd5e6fbafea1e60ac80603fe54917e7b383 100644 --- a/training_args.bin +++ b/training_args.bin @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:175919cb25877e1996cc155cc0993789d718845c4e5743fb29fe45e4e6c4c586 +oid sha256:473e29d9baac3e0364c98f3b27fced8c3f94ce855cb03c87151ab7bcabc5fd1b size 5819