diff --git a/checkpoint-1000/README.md b/checkpoint-1000/README.md new file mode 100644 index 0000000000000000000000000000000000000000..16b1eacdd9353dec380a08ee77ce6ed5ab50f12e --- /dev/null +++ b/checkpoint-1000/README.md @@ -0,0 +1,202 @@ +--- +library_name: peft +base_model: gotzmann/uni +--- + +# Model Card for Model ID + + + + + +## Model Details + +### Model Description + + + + + +- **Developed by:** [More Information Needed] +- **Funded by [optional]:** [More Information Needed] +- **Shared by [optional]:** [More Information Needed] +- **Model type:** [More Information Needed] +- **Language(s) (NLP):** [More Information Needed] +- **License:** [More Information Needed] +- **Finetuned from model [optional]:** [More Information Needed] + +### Model Sources [optional] + + + +- **Repository:** [More Information Needed] +- **Paper [optional]:** [More Information Needed] +- **Demo [optional]:** [More Information Needed] + +## Uses + + + +### Direct Use + + + +[More Information Needed] + +### Downstream Use [optional] + + + +[More Information Needed] + +### Out-of-Scope Use + + + +[More Information Needed] + +## Bias, Risks, and Limitations + + + +[More Information Needed] + +### Recommendations + + + +Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations. + +## How to Get Started with the Model + +Use the code below to get started with the model. + +[More Information Needed] + +## Training Details + +### Training Data + + + +[More Information Needed] + +### Training Procedure + + + +#### Preprocessing [optional] + +[More Information Needed] + + +#### Training Hyperparameters + +- **Training regime:** [More Information Needed] + +#### Speeds, Sizes, Times [optional] + + + +[More Information Needed] + +## Evaluation + + + +### Testing Data, Factors & Metrics + +#### Testing Data + + + +[More Information Needed] + +#### Factors + + + +[More Information Needed] + +#### Metrics + + + +[More Information Needed] + +### Results + +[More Information Needed] + +#### Summary + + + +## Model Examination [optional] + + + +[More Information Needed] + +## Environmental Impact + + + +Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700). + +- **Hardware Type:** [More Information Needed] +- **Hours used:** [More Information Needed] +- **Cloud Provider:** [More Information Needed] +- **Compute Region:** [More Information Needed] +- **Carbon Emitted:** [More Information Needed] + +## Technical Specifications [optional] + +### Model Architecture and Objective + +[More Information Needed] + +### Compute Infrastructure + +[More Information Needed] + +#### Hardware + +[More Information Needed] + +#### Software + +[More Information Needed] + +## Citation [optional] + + + +**BibTeX:** + +[More Information Needed] + +**APA:** + +[More Information Needed] + +## Glossary [optional] + + + +[More Information Needed] + +## More Information [optional] + +[More Information Needed] + +## Model Card Authors [optional] + +[More Information Needed] + +## Model Card Contact + +[More Information Needed] +### Framework versions + +- PEFT 0.10.0 \ No newline at end of file diff --git a/checkpoint-1000/adapter_config.json b/checkpoint-1000/adapter_config.json new file mode 100644 index 0000000000000000000000000000000000000000..832188d72d81e59dd2b5259e86f371199b441aca --- /dev/null +++ b/checkpoint-1000/adapter_config.json @@ -0,0 +1,31 @@ +{ + "alpha_pattern": {}, + "auto_mapping": null, + "base_model_name_or_path": "gotzmann/uni", + "bias": "none", + "fan_in_fan_out": false, + "inference_mode": true, + "init_lora_weights": true, + "layer_replication": null, + "layers_pattern": null, + "layers_to_transform": null, + "loftq_config": {}, + "lora_alpha": 128, + "lora_dropout": 0.0, + "megatron_config": null, + "megatron_core": "megatron.core", + "modules_to_save": null, + "peft_type": "LORA", + "r": 128, + "rank_pattern": {}, + "revision": null, + "target_modules": [ + "o_proj", + "k_proj", + "q_proj", + "v_proj" + ], + "task_type": "CAUSAL_LM", + "use_dora": false, + "use_rslora": true +} \ No newline at end of file diff --git a/checkpoint-1000/adapter_model.safetensors b/checkpoint-1000/adapter_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..634eb05b161ad2b8d63c690758e96ed6ad879566 --- /dev/null +++ b/checkpoint-1000/adapter_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e1ce644b4ea3cb5631a4ade4a68fa19557634516036410af8e088f5c2b51b3ad +size 1048664848 diff --git a/checkpoint-1000/global_step1000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt b/checkpoint-1000/global_step1000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..05c1fda4c6052faa8f7c60239ae248f06f6b8894 --- /dev/null +++ b/checkpoint-1000/global_step1000/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:c4f078801474443721e2577ddc62ff1c280744861d069c1bcb960c34196f9d9b +size 787270042 diff --git a/checkpoint-1000/global_step1000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt b/checkpoint-1000/global_step1000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..de5db04a98ca3d64d4e32c8d28ac7fc0c6b8c7ce --- /dev/null +++ b/checkpoint-1000/global_step1000/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:c735b7960263d865759c604d4661f0970bcbbba4f3f864b6cf7dd93357f669dc +size 787270042 diff --git a/checkpoint-1000/global_step1000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt b/checkpoint-1000/global_step1000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..9f74f5ab6a3f1386c711da953f10b5b6b8556bfb --- /dev/null +++ b/checkpoint-1000/global_step1000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:077981153dc39aef6468d5cae213da828adb6d6516dd1ecfd5e92126125e0e1e +size 787270042 diff --git a/checkpoint-1000/global_step1000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt b/checkpoint-1000/global_step1000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..3958dfe55e01fd06c0dfdba77776a5f6fe46ffca --- /dev/null +++ b/checkpoint-1000/global_step1000/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:f951e8a525cd2b57687d34505dfbe62ba707a1e7fa3b87351c6c4b5d089dd23c +size 787270042 diff --git a/checkpoint-1000/global_step1000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt b/checkpoint-1000/global_step1000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..3ea9e9a0fd06ca695efd22fa051301b913ce693e --- /dev/null +++ b/checkpoint-1000/global_step1000/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:eea35d69b8382c06ff895f2ca19189f9d36fbeaf8e2123f498fb431c2cf4dc8d +size 787270042 diff --git a/checkpoint-1000/global_step1000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt b/checkpoint-1000/global_step1000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..1315d20ae5964b65ffd456d2d136b7f50b9a77fa --- /dev/null +++ b/checkpoint-1000/global_step1000/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:adca0aeba92c1581a778b03dcebf647f4d80f9a304359aec2f7e8739e3d2dee9 +size 787270042 diff --git a/checkpoint-1000/global_step1000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt b/checkpoint-1000/global_step1000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..225d919d35bf6120bb37159ad7e7ad74aa4bdf42 --- /dev/null +++ b/checkpoint-1000/global_step1000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1be8b91f86d410b6d352f260da5223d0db64fe8a10f97da168db3465f59bc194 +size 787270042 diff --git a/checkpoint-1000/global_step1000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt b/checkpoint-1000/global_step1000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..de25629c43d14c771b3364e5252c77b058ae465f --- /dev/null +++ b/checkpoint-1000/global_step1000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6b7902bc2fc97ca5b26dfbeb1e093daa77464f605402b20f14f955aee413e021 +size 787270042 diff --git a/checkpoint-1000/global_step1000/zero_pp_rank_0_mp_rank_00_model_states.pt b/checkpoint-1000/global_step1000/zero_pp_rank_0_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..94eb802d0f82d382044acdb13c95baaf8808ac08 --- /dev/null +++ b/checkpoint-1000/global_step1000/zero_pp_rank_0_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:82c0af70dff5f96a3a06356646455999cb721f9c9c9cb6bffe441eccfe736008 +size 653742 diff --git a/checkpoint-1000/global_step1000/zero_pp_rank_1_mp_rank_00_model_states.pt b/checkpoint-1000/global_step1000/zero_pp_rank_1_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..59ccdb3a5d81c6d9326a89fef05824647a372a48 --- /dev/null +++ b/checkpoint-1000/global_step1000/zero_pp_rank_1_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:da48eb8b26f93823d0693833e5ac78f5cbe01579d9e23ff6b36a69489dca2610 +size 653742 diff --git a/checkpoint-1000/global_step1000/zero_pp_rank_2_mp_rank_00_model_states.pt b/checkpoint-1000/global_step1000/zero_pp_rank_2_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..a7564f51069055122aab1fddb7b91c4ee18b7ea2 --- /dev/null +++ b/checkpoint-1000/global_step1000/zero_pp_rank_2_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7436d2d4163ecc69a21cfb9c0aa9bb8fb04a6a0eb67326cc10e1abe243036e1f +size 653742 diff --git a/checkpoint-1000/global_step1000/zero_pp_rank_3_mp_rank_00_model_states.pt b/checkpoint-1000/global_step1000/zero_pp_rank_3_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..3acc1ba7bd92567083d7d22be80a70973550a416 --- /dev/null +++ b/checkpoint-1000/global_step1000/zero_pp_rank_3_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9739aad3885189418801ae6d471cad7ecc60d9dde6c5855acdbfe218e9c9eadd +size 653742 diff --git a/checkpoint-1000/global_step1000/zero_pp_rank_4_mp_rank_00_model_states.pt b/checkpoint-1000/global_step1000/zero_pp_rank_4_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..e5c938bd1f3417190d2c43d31d3cd842d38326e2 --- /dev/null +++ b/checkpoint-1000/global_step1000/zero_pp_rank_4_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:127f55a8a0ffe91f1b222203cb0dc4c57dc1a60cd0a546f3dfcc55b2a976fbca +size 653742 diff --git a/checkpoint-1000/global_step1000/zero_pp_rank_5_mp_rank_00_model_states.pt b/checkpoint-1000/global_step1000/zero_pp_rank_5_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..f5a4af1fb75f6c7a2626c035a339de66773228c5 --- /dev/null +++ b/checkpoint-1000/global_step1000/zero_pp_rank_5_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e3f1322e9f7cee7269435dc7f168dc235f65730c79bf762adc662a080c9ea720 +size 653742 diff --git a/checkpoint-1000/global_step1000/zero_pp_rank_6_mp_rank_00_model_states.pt b/checkpoint-1000/global_step1000/zero_pp_rank_6_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..60e77b0dc7069f3456267fa1adadda9dc41c8831 --- /dev/null +++ b/checkpoint-1000/global_step1000/zero_pp_rank_6_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:84675944246f5dc5aba60518fa340ee67c8efae1c8aa70b7fa2e9b6b9efeaf0a +size 653742 diff --git a/checkpoint-1000/global_step1000/zero_pp_rank_7_mp_rank_00_model_states.pt b/checkpoint-1000/global_step1000/zero_pp_rank_7_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..ac9dea05fe1c99698e039a35ae835e8444cc8e8f --- /dev/null +++ b/checkpoint-1000/global_step1000/zero_pp_rank_7_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5b0ad763b5bae57e03d281834711e7b66343bcd505012a0c2d92b05764fb7f3a +size 653742 diff --git a/checkpoint-1000/latest b/checkpoint-1000/latest new file mode 100644 index 0000000000000000000000000000000000000000..e2d3435fb1acf8913e6bd6c51b01adfc69b11ac6 --- /dev/null +++ b/checkpoint-1000/latest @@ -0,0 +1 @@ +global_step1000 \ No newline at end of file diff --git a/checkpoint-1000/rng_state_0.pth b/checkpoint-1000/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..9dd2a62da4ca83b3b986d96dbf0eaeb82207ca93 --- /dev/null +++ b/checkpoint-1000/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a0628a9017696045a3a29e9eaffc71e9262d855716e773c0c3be760a1fe85bc8 +size 15984 diff --git a/checkpoint-1000/rng_state_1.pth b/checkpoint-1000/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..1ba5f3aba4388a582cd47f7f9e57cd5879b1cbd2 --- /dev/null +++ b/checkpoint-1000/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:df342004a4d8e3626bf2a9f689fde7c8bfd6d995e14931f5496eda1f456cb6f2 +size 15984 diff --git a/checkpoint-1000/rng_state_2.pth b/checkpoint-1000/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..27b0f7845c2b9530c3e6ed3ce232ff4e86b86122 --- /dev/null +++ b/checkpoint-1000/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f02096eb4e8850b91490e80e4a042e2e60f71bd2abc6a269d62c271649cb77d2 +size 15984 diff --git a/checkpoint-1000/rng_state_3.pth b/checkpoint-1000/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..fcfb583fc43c6dd4395671708744cfd18c419970 --- /dev/null +++ b/checkpoint-1000/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:326c778d3d0e7e3d5665fa0a9ecd92986609c430da08b41611d6c05dc19815a8 +size 15984 diff --git a/checkpoint-1000/rng_state_4.pth b/checkpoint-1000/rng_state_4.pth new file mode 100644 index 0000000000000000000000000000000000000000..7a8c64b1f15ac655b2be2a42fe61cabe2a877704 --- /dev/null +++ b/checkpoint-1000/rng_state_4.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d978dcb0c34e022ee6750e9d86814b8c82e4965d7e07662f35f06eeac12938f3 +size 15984 diff --git a/checkpoint-1000/rng_state_5.pth b/checkpoint-1000/rng_state_5.pth new file mode 100644 index 0000000000000000000000000000000000000000..262e8187e6caeca12ef3b0aa923b12afd697e03d --- /dev/null +++ b/checkpoint-1000/rng_state_5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:01e83399aed1d9d173c3e07b2efa8530c956b62b2b68394c2ed0d43bd8bba9d1 +size 15984 diff --git a/checkpoint-1000/rng_state_6.pth b/checkpoint-1000/rng_state_6.pth new file mode 100644 index 0000000000000000000000000000000000000000..72f794e31f8d3e0c63972e5076e1ed90c52087ba --- /dev/null +++ b/checkpoint-1000/rng_state_6.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:606ab3ca92e3d20c327c69fdcce7f7e39bec2f2c3538b036088b255f917e3ba4 +size 15984 diff --git a/checkpoint-1000/rng_state_7.pth b/checkpoint-1000/rng_state_7.pth new file mode 100644 index 0000000000000000000000000000000000000000..244e7fdaa1cef2e82bd4e16afb10f32f68318bcc --- /dev/null +++ b/checkpoint-1000/rng_state_7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1276a987dd22c9093fec58921ba19f340a28f18bff635cc01324e09a3c37ac3a +size 15984 diff --git a/checkpoint-1000/scheduler.pt b/checkpoint-1000/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..01f479c24cd827042085d519a3c91073dad5efba --- /dev/null +++ b/checkpoint-1000/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:269f3c1d01e53e47ac89c5d48686db3c5bd5052b5dbb75807c9cffc4b0ab99ae +size 1064 diff --git a/checkpoint-1000/special_tokens_map.json b/checkpoint-1000/special_tokens_map.json new file mode 100644 index 0000000000000000000000000000000000000000..72ecfeeb7e14d244c936169d2ed139eeae235ef1 --- /dev/null +++ b/checkpoint-1000/special_tokens_map.json @@ -0,0 +1,24 @@ +{ + "bos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "eos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "pad_token": "", + "unk_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + } +} diff --git a/checkpoint-1000/tokenizer.model b/checkpoint-1000/tokenizer.model new file mode 100644 index 0000000000000000000000000000000000000000..6c00c742ce03c627d6cd5b795984876fa49fa899 --- /dev/null +++ b/checkpoint-1000/tokenizer.model @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347 +size 499723 diff --git a/checkpoint-1000/tokenizer_config.json b/checkpoint-1000/tokenizer_config.json new file mode 100644 index 0000000000000000000000000000000000000000..bb5a9f09d8c0f3c32c66fc6118fe5c76c5c6fd90 --- /dev/null +++ b/checkpoint-1000/tokenizer_config.json @@ -0,0 +1,45 @@ +{ + "add_bos_token": true, + "add_eos_token": false, + "add_prefix_space": false, + "added_tokens_decoder": { + "0": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "1": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "2": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + } + }, + "bos_token": "", + "chat_template": "{% if messages[0]['role'] == 'system' %}{% set system_message = messages[0]['content'] %}{% endif %}{% if system_message is defined %}{{ '' + '### System:\\n\\n' + system_message }}{% endif %}{% for message in messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ '\\n\\n### Human:\\n\\n' + content }}{% elif message['role'] == 'assistant' %}{{ '\\n\\n### Assistant:\\n\\n' + content + '' }}{% endif %}{% endfor %}", + "clean_up_tokenization_spaces": false, + "eos_token": "", + "legacy": false, + "model_max_length": 1000000000000000019884624838656, + "pad_token": "", + "padding_side": "right", + "sp_model_kwargs": {}, + "spaces_between_special_tokens": false, + "split_special_tokens": false, + "tokenizer_class": "LlamaTokenizer", + "unk_token": "", + "use_default_system_prompt": false +} diff --git a/checkpoint-1000/trainer_state.json b/checkpoint-1000/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..1c8b4a95e4310fa237b7a233fad45f643cbdd090 --- /dev/null +++ b/checkpoint-1000/trainer_state.json @@ -0,0 +1,7021 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 1.365797896662094, + "eval_steps": 500, + "global_step": 1000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "grad_norm": 0.849355824164473, + "learning_rate": 4.878048780487805e-07, + "loss": 1.3655, + "step": 1 + }, + { + "epoch": 0.0, + "grad_norm": 10.01567518957158, + "learning_rate": 9.75609756097561e-07, + "loss": 1.5767, + "step": 2 + }, + { + "epoch": 0.01, + "grad_norm": 0.6466000875559635, + "learning_rate": 1.4634146341463414e-06, + "loss": 1.3913, + "step": 3 + }, + { + "epoch": 0.01, + "grad_norm": 0.6644565932010504, + "learning_rate": 1.951219512195122e-06, + "loss": 1.3218, + "step": 4 + }, + { + "epoch": 0.01, + "grad_norm": 0.571354207588475, + "learning_rate": 2.4390243902439027e-06, + "loss": 1.3597, + "step": 5 + }, + { + "epoch": 0.01, + "grad_norm": 0.31036262839244955, + "learning_rate": 2.926829268292683e-06, + "loss": 1.2832, + "step": 6 + }, + { + "epoch": 0.01, + "grad_norm": 0.2622135027188184, + "learning_rate": 3.414634146341464e-06, + "loss": 1.2161, + "step": 7 + }, + { + "epoch": 0.01, + "grad_norm": 0.296824630261661, + "learning_rate": 3.902439024390244e-06, + "loss": 1.2985, + "step": 8 + }, + { + "epoch": 0.02, + "grad_norm": 0.2557267467361569, + "learning_rate": 4.390243902439025e-06, + "loss": 1.3175, + "step": 9 + }, + { + "epoch": 0.02, + "grad_norm": 0.23418939513890769, + "learning_rate": 4.8780487804878055e-06, + "loss": 1.2617, + "step": 10 + }, + { + "epoch": 0.02, + "grad_norm": 0.2364760983285843, + "learning_rate": 5.365853658536586e-06, + "loss": 1.3103, + "step": 11 + }, + { + "epoch": 0.02, + "grad_norm": 0.23893034721889, + "learning_rate": 5.853658536585366e-06, + "loss": 1.2405, + "step": 12 + }, + { + "epoch": 0.02, + "grad_norm": 0.25563593295485887, + "learning_rate": 6.341463414634147e-06, + "loss": 1.2831, + "step": 13 + }, + { + "epoch": 0.03, + "grad_norm": 0.23239975352661665, + "learning_rate": 6.829268292682928e-06, + "loss": 1.3125, + "step": 14 + }, + { + "epoch": 0.03, + "grad_norm": 0.3092813858209507, + "learning_rate": 7.317073170731707e-06, + "loss": 1.2422, + "step": 15 + }, + { + "epoch": 0.03, + "grad_norm": 0.282563380367434, + "learning_rate": 7.804878048780489e-06, + "loss": 1.2453, + "step": 16 + }, + { + "epoch": 0.03, + "grad_norm": 0.22065680088315018, + "learning_rate": 8.292682926829268e-06, + "loss": 1.2491, + "step": 17 + }, + { + "epoch": 0.03, + "grad_norm": 0.22777800877980184, + "learning_rate": 8.78048780487805e-06, + "loss": 1.2655, + "step": 18 + }, + { + "epoch": 0.03, + "grad_norm": 0.22145212540177928, + "learning_rate": 9.268292682926831e-06, + "loss": 1.2413, + "step": 19 + }, + { + "epoch": 0.04, + "grad_norm": 0.22482351883112714, + "learning_rate": 9.756097560975611e-06, + "loss": 1.2653, + "step": 20 + }, + { + "epoch": 0.04, + "grad_norm": 0.20823080508385733, + "learning_rate": 1.024390243902439e-05, + "loss": 1.2374, + "step": 21 + }, + { + "epoch": 0.04, + "grad_norm": 0.26025492562935737, + "learning_rate": 1.0731707317073172e-05, + "loss": 1.2065, + "step": 22 + }, + { + "epoch": 0.04, + "grad_norm": 0.2150252124176173, + "learning_rate": 1.1219512195121953e-05, + "loss": 1.2782, + "step": 23 + }, + { + "epoch": 0.04, + "grad_norm": 0.2505915177425618, + "learning_rate": 1.1707317073170731e-05, + "loss": 1.2742, + "step": 24 + }, + { + "epoch": 0.05, + "grad_norm": 0.20129223044786942, + "learning_rate": 1.2195121951219513e-05, + "loss": 1.3366, + "step": 25 + }, + { + "epoch": 0.05, + "grad_norm": 0.1973508510397107, + "learning_rate": 1.2682926829268294e-05, + "loss": 1.2476, + "step": 26 + }, + { + "epoch": 0.05, + "grad_norm": 0.27103325392437194, + "learning_rate": 1.3170731707317076e-05, + "loss": 1.2325, + "step": 27 + }, + { + "epoch": 0.05, + "grad_norm": 0.17954976411006285, + "learning_rate": 1.3658536585365855e-05, + "loss": 1.2523, + "step": 28 + }, + { + "epoch": 0.05, + "grad_norm": 0.22216997851088888, + "learning_rate": 1.4146341463414635e-05, + "loss": 1.3297, + "step": 29 + }, + { + "epoch": 0.05, + "grad_norm": 0.2071458864548587, + "learning_rate": 1.4634146341463415e-05, + "loss": 1.2127, + "step": 30 + }, + { + "epoch": 0.06, + "grad_norm": 0.18039422081622164, + "learning_rate": 1.5121951219512196e-05, + "loss": 1.2509, + "step": 31 + }, + { + "epoch": 0.06, + "grad_norm": 0.18631254372974412, + "learning_rate": 1.5609756097560978e-05, + "loss": 1.2247, + "step": 32 + }, + { + "epoch": 0.06, + "grad_norm": 0.18843872523649827, + "learning_rate": 1.6097560975609757e-05, + "loss": 1.195, + "step": 33 + }, + { + "epoch": 0.06, + "grad_norm": 0.2163847267778325, + "learning_rate": 1.6585365853658537e-05, + "loss": 1.2179, + "step": 34 + }, + { + "epoch": 0.06, + "grad_norm": 0.19687688475496104, + "learning_rate": 1.7073170731707317e-05, + "loss": 1.2763, + "step": 35 + }, + { + "epoch": 0.07, + "grad_norm": 0.20409643064887947, + "learning_rate": 1.75609756097561e-05, + "loss": 1.253, + "step": 36 + }, + { + "epoch": 0.07, + "grad_norm": 0.1879182661759335, + "learning_rate": 1.804878048780488e-05, + "loss": 1.2586, + "step": 37 + }, + { + "epoch": 0.07, + "grad_norm": 0.19400648948514373, + "learning_rate": 1.8536585365853663e-05, + "loss": 1.2154, + "step": 38 + }, + { + "epoch": 0.07, + "grad_norm": 0.1878879343148452, + "learning_rate": 1.902439024390244e-05, + "loss": 1.2304, + "step": 39 + }, + { + "epoch": 0.07, + "grad_norm": 0.17687475469924052, + "learning_rate": 1.9512195121951222e-05, + "loss": 1.2351, + "step": 40 + }, + { + "epoch": 0.07, + "grad_norm": 0.18223935625384885, + "learning_rate": 2e-05, + "loss": 1.2222, + "step": 41 + }, + { + "epoch": 0.08, + "grad_norm": 0.1943061629408338, + "learning_rate": 2.048780487804878e-05, + "loss": 1.2044, + "step": 42 + }, + { + "epoch": 0.08, + "grad_norm": 0.17027514338700078, + "learning_rate": 2.0975609756097564e-05, + "loss": 1.1548, + "step": 43 + }, + { + "epoch": 0.08, + "grad_norm": 0.18553769630586192, + "learning_rate": 2.1463414634146344e-05, + "loss": 1.2721, + "step": 44 + }, + { + "epoch": 0.08, + "grad_norm": 0.19732826914228765, + "learning_rate": 2.1951219512195124e-05, + "loss": 1.3097, + "step": 45 + }, + { + "epoch": 0.08, + "grad_norm": 0.18714230986631472, + "learning_rate": 2.2439024390243907e-05, + "loss": 1.2662, + "step": 46 + }, + { + "epoch": 0.09, + "grad_norm": 0.19988987568002223, + "learning_rate": 2.2926829268292683e-05, + "loss": 1.2904, + "step": 47 + }, + { + "epoch": 0.09, + "grad_norm": 0.17744650133390918, + "learning_rate": 2.3414634146341463e-05, + "loss": 1.1825, + "step": 48 + }, + { + "epoch": 0.09, + "grad_norm": 0.16576734763834533, + "learning_rate": 2.3902439024390246e-05, + "loss": 1.1858, + "step": 49 + }, + { + "epoch": 0.09, + "grad_norm": 0.179591794065527, + "learning_rate": 2.4390243902439026e-05, + "loss": 1.2711, + "step": 50 + }, + { + "epoch": 0.09, + "grad_norm": 0.17923464471176911, + "learning_rate": 2.4878048780487805e-05, + "loss": 1.2289, + "step": 51 + }, + { + "epoch": 0.1, + "grad_norm": 0.18991742907836837, + "learning_rate": 2.536585365853659e-05, + "loss": 1.3097, + "step": 52 + }, + { + "epoch": 0.1, + "grad_norm": 0.19849796137254636, + "learning_rate": 2.5853658536585368e-05, + "loss": 1.2489, + "step": 53 + }, + { + "epoch": 0.1, + "grad_norm": 0.17452371110976383, + "learning_rate": 2.634146341463415e-05, + "loss": 1.2461, + "step": 54 + }, + { + "epoch": 0.1, + "grad_norm": 0.17671022353085036, + "learning_rate": 2.682926829268293e-05, + "loss": 1.153, + "step": 55 + }, + { + "epoch": 0.1, + "grad_norm": 0.36820559192096686, + "learning_rate": 2.731707317073171e-05, + "loss": 1.2431, + "step": 56 + }, + { + "epoch": 0.1, + "grad_norm": 0.20331468526494198, + "learning_rate": 2.7804878048780487e-05, + "loss": 1.2575, + "step": 57 + }, + { + "epoch": 0.11, + "grad_norm": 0.2402486598118377, + "learning_rate": 2.829268292682927e-05, + "loss": 1.2538, + "step": 58 + }, + { + "epoch": 0.11, + "grad_norm": 0.2549409484173144, + "learning_rate": 2.878048780487805e-05, + "loss": 1.2065, + "step": 59 + }, + { + "epoch": 0.11, + "grad_norm": 0.2053105349872685, + "learning_rate": 2.926829268292683e-05, + "loss": 1.2094, + "step": 60 + }, + { + "epoch": 0.11, + "grad_norm": 0.17971910872957886, + "learning_rate": 2.9756097560975613e-05, + "loss": 1.228, + "step": 61 + }, + { + "epoch": 0.11, + "grad_norm": 0.1885853654992973, + "learning_rate": 3.0243902439024392e-05, + "loss": 1.2286, + "step": 62 + }, + { + "epoch": 0.12, + "grad_norm": 0.1848524571968613, + "learning_rate": 3.073170731707317e-05, + "loss": 1.2718, + "step": 63 + }, + { + "epoch": 0.12, + "grad_norm": 0.18734105883548513, + "learning_rate": 3.1219512195121955e-05, + "loss": 1.2357, + "step": 64 + }, + { + "epoch": 0.12, + "grad_norm": 0.17774668052121825, + "learning_rate": 3.170731707317074e-05, + "loss": 1.1509, + "step": 65 + }, + { + "epoch": 0.12, + "grad_norm": 0.17890968008080646, + "learning_rate": 3.2195121951219514e-05, + "loss": 1.1924, + "step": 66 + }, + { + "epoch": 0.12, + "grad_norm": 0.18249273371332375, + "learning_rate": 3.268292682926829e-05, + "loss": 1.2545, + "step": 67 + }, + { + "epoch": 0.12, + "grad_norm": 0.21064122671902577, + "learning_rate": 3.3170731707317074e-05, + "loss": 1.2832, + "step": 68 + }, + { + "epoch": 0.13, + "grad_norm": 0.1820064171955093, + "learning_rate": 3.365853658536586e-05, + "loss": 1.2071, + "step": 69 + }, + { + "epoch": 0.13, + "grad_norm": 0.16996662800553433, + "learning_rate": 3.414634146341463e-05, + "loss": 1.2073, + "step": 70 + }, + { + "epoch": 0.13, + "grad_norm": 0.1618669302922445, + "learning_rate": 3.4634146341463416e-05, + "loss": 1.1289, + "step": 71 + }, + { + "epoch": 0.13, + "grad_norm": 0.18948744950985544, + "learning_rate": 3.51219512195122e-05, + "loss": 1.2915, + "step": 72 + }, + { + "epoch": 0.13, + "grad_norm": 0.18326143691603383, + "learning_rate": 3.5609756097560976e-05, + "loss": 1.2238, + "step": 73 + }, + { + "epoch": 0.14, + "grad_norm": 0.17410704510700503, + "learning_rate": 3.609756097560976e-05, + "loss": 1.1784, + "step": 74 + }, + { + "epoch": 0.14, + "grad_norm": 0.1983667344995625, + "learning_rate": 3.658536585365854e-05, + "loss": 1.2452, + "step": 75 + }, + { + "epoch": 0.14, + "grad_norm": 0.3416310763369357, + "learning_rate": 3.7073170731707325e-05, + "loss": 1.1972, + "step": 76 + }, + { + "epoch": 0.14, + "grad_norm": 0.2776466983511955, + "learning_rate": 3.75609756097561e-05, + "loss": 1.3121, + "step": 77 + }, + { + "epoch": 0.14, + "grad_norm": 0.20026129636576834, + "learning_rate": 3.804878048780488e-05, + "loss": 1.2436, + "step": 78 + }, + { + "epoch": 0.14, + "grad_norm": 0.21064549243917835, + "learning_rate": 3.853658536585366e-05, + "loss": 1.2064, + "step": 79 + }, + { + "epoch": 0.15, + "grad_norm": 0.22119482175714267, + "learning_rate": 3.9024390243902444e-05, + "loss": 1.2715, + "step": 80 + }, + { + "epoch": 0.15, + "grad_norm": 0.23047133748844142, + "learning_rate": 3.951219512195122e-05, + "loss": 1.2888, + "step": 81 + }, + { + "epoch": 0.15, + "grad_norm": 0.18741863156973176, + "learning_rate": 4e-05, + "loss": 1.248, + "step": 82 + }, + { + "epoch": 0.15, + "grad_norm": 0.1747859810629604, + "learning_rate": 4.0487804878048786e-05, + "loss": 1.1683, + "step": 83 + }, + { + "epoch": 0.15, + "grad_norm": 0.1896944798413341, + "learning_rate": 4.097560975609756e-05, + "loss": 1.2155, + "step": 84 + }, + { + "epoch": 0.16, + "grad_norm": 0.18724128114363303, + "learning_rate": 4.1463414634146346e-05, + "loss": 1.2273, + "step": 85 + }, + { + "epoch": 0.16, + "grad_norm": 0.17368125504855478, + "learning_rate": 4.195121951219513e-05, + "loss": 1.224, + "step": 86 + }, + { + "epoch": 0.16, + "grad_norm": 0.18371141013625703, + "learning_rate": 4.2439024390243905e-05, + "loss": 1.2294, + "step": 87 + }, + { + "epoch": 0.16, + "grad_norm": 0.1791029365673714, + "learning_rate": 4.292682926829269e-05, + "loss": 1.2895, + "step": 88 + }, + { + "epoch": 0.16, + "grad_norm": 0.20259974283859655, + "learning_rate": 4.341463414634147e-05, + "loss": 1.1841, + "step": 89 + }, + { + "epoch": 0.16, + "grad_norm": 0.17457456183272174, + "learning_rate": 4.390243902439025e-05, + "loss": 1.2357, + "step": 90 + }, + { + "epoch": 0.17, + "grad_norm": 0.1815824380789748, + "learning_rate": 4.439024390243903e-05, + "loss": 1.2304, + "step": 91 + }, + { + "epoch": 0.17, + "grad_norm": 0.17566480599583392, + "learning_rate": 4.4878048780487814e-05, + "loss": 1.242, + "step": 92 + }, + { + "epoch": 0.17, + "grad_norm": 0.18422975005984474, + "learning_rate": 4.536585365853658e-05, + "loss": 1.2177, + "step": 93 + }, + { + "epoch": 0.17, + "grad_norm": 0.16796781877940678, + "learning_rate": 4.5853658536585366e-05, + "loss": 1.1482, + "step": 94 + }, + { + "epoch": 0.17, + "grad_norm": 0.18636131653783305, + "learning_rate": 4.634146341463415e-05, + "loss": 1.1758, + "step": 95 + }, + { + "epoch": 0.18, + "grad_norm": 0.1823665700289814, + "learning_rate": 4.6829268292682926e-05, + "loss": 1.289, + "step": 96 + }, + { + "epoch": 0.18, + "grad_norm": 0.1719900691262439, + "learning_rate": 4.731707317073171e-05, + "loss": 1.1626, + "step": 97 + }, + { + "epoch": 0.18, + "grad_norm": 0.17937994168039778, + "learning_rate": 4.780487804878049e-05, + "loss": 1.175, + "step": 98 + }, + { + "epoch": 0.18, + "grad_norm": 0.16631851422106986, + "learning_rate": 4.829268292682927e-05, + "loss": 1.2177, + "step": 99 + }, + { + "epoch": 0.18, + "grad_norm": 0.19143696232800309, + "learning_rate": 4.878048780487805e-05, + "loss": 1.3071, + "step": 100 + }, + { + "epoch": 0.18, + "grad_norm": 0.17859506638780318, + "learning_rate": 4.9268292682926835e-05, + "loss": 1.2351, + "step": 101 + }, + { + "epoch": 0.19, + "grad_norm": 0.18381520321248196, + "learning_rate": 4.975609756097561e-05, + "loss": 1.2342, + "step": 102 + }, + { + "epoch": 0.19, + "grad_norm": 0.17968218683773912, + "learning_rate": 5.0243902439024394e-05, + "loss": 1.2074, + "step": 103 + }, + { + "epoch": 0.19, + "grad_norm": 0.18139489969339018, + "learning_rate": 5.073170731707318e-05, + "loss": 1.1558, + "step": 104 + }, + { + "epoch": 0.19, + "grad_norm": 0.17366624842514394, + "learning_rate": 5.121951219512195e-05, + "loss": 1.1897, + "step": 105 + }, + { + "epoch": 0.19, + "grad_norm": 0.16034845455223745, + "learning_rate": 5.1707317073170736e-05, + "loss": 1.179, + "step": 106 + }, + { + "epoch": 0.2, + "grad_norm": 0.17583069577827776, + "learning_rate": 5.219512195121952e-05, + "loss": 1.1856, + "step": 107 + }, + { + "epoch": 0.2, + "grad_norm": 0.1853758076989552, + "learning_rate": 5.26829268292683e-05, + "loss": 1.2072, + "step": 108 + }, + { + "epoch": 0.2, + "grad_norm": 0.19597443965936462, + "learning_rate": 5.317073170731708e-05, + "loss": 1.2271, + "step": 109 + }, + { + "epoch": 0.2, + "grad_norm": 0.1899206334098331, + "learning_rate": 5.365853658536586e-05, + "loss": 1.1961, + "step": 110 + }, + { + "epoch": 0.2, + "grad_norm": 0.17463763837757018, + "learning_rate": 5.4146341463414645e-05, + "loss": 1.2049, + "step": 111 + }, + { + "epoch": 0.2, + "grad_norm": 0.20431371701229986, + "learning_rate": 5.463414634146342e-05, + "loss": 1.2891, + "step": 112 + }, + { + "epoch": 0.21, + "grad_norm": 0.1814475107638498, + "learning_rate": 5.51219512195122e-05, + "loss": 1.2346, + "step": 113 + }, + { + "epoch": 0.21, + "grad_norm": 0.1883849423207823, + "learning_rate": 5.5609756097560974e-05, + "loss": 1.244, + "step": 114 + }, + { + "epoch": 0.21, + "grad_norm": 0.1857258128640568, + "learning_rate": 5.609756097560976e-05, + "loss": 1.2669, + "step": 115 + }, + { + "epoch": 0.21, + "grad_norm": 0.1740768514118401, + "learning_rate": 5.658536585365854e-05, + "loss": 1.2414, + "step": 116 + }, + { + "epoch": 0.21, + "grad_norm": 0.1919320335584178, + "learning_rate": 5.7073170731707317e-05, + "loss": 1.2886, + "step": 117 + }, + { + "epoch": 0.22, + "grad_norm": 0.18288775167828136, + "learning_rate": 5.75609756097561e-05, + "loss": 1.1875, + "step": 118 + }, + { + "epoch": 0.22, + "grad_norm": 0.18208588867750863, + "learning_rate": 5.804878048780488e-05, + "loss": 1.2388, + "step": 119 + }, + { + "epoch": 0.22, + "grad_norm": 0.1743260015658331, + "learning_rate": 5.853658536585366e-05, + "loss": 1.1762, + "step": 120 + }, + { + "epoch": 0.22, + "grad_norm": 0.17856046291517946, + "learning_rate": 5.902439024390244e-05, + "loss": 1.2888, + "step": 121 + }, + { + "epoch": 0.22, + "grad_norm": 0.17493794870966536, + "learning_rate": 5.9512195121951225e-05, + "loss": 1.2222, + "step": 122 + }, + { + "epoch": 0.22, + "grad_norm": 0.1909202655203384, + "learning_rate": 6.000000000000001e-05, + "loss": 1.2414, + "step": 123 + }, + { + "epoch": 0.23, + "grad_norm": 0.18345819482834988, + "learning_rate": 6.0487804878048785e-05, + "loss": 1.2756, + "step": 124 + }, + { + "epoch": 0.23, + "grad_norm": 0.2057069352956621, + "learning_rate": 6.097560975609757e-05, + "loss": 1.261, + "step": 125 + }, + { + "epoch": 0.23, + "grad_norm": 0.299775882469108, + "learning_rate": 6.146341463414634e-05, + "loss": 1.2566, + "step": 126 + }, + { + "epoch": 0.23, + "grad_norm": 0.1869687633018095, + "learning_rate": 6.195121951219513e-05, + "loss": 1.3039, + "step": 127 + }, + { + "epoch": 0.23, + "grad_norm": 0.17747149926197442, + "learning_rate": 6.243902439024391e-05, + "loss": 1.2524, + "step": 128 + }, + { + "epoch": 0.24, + "grad_norm": 0.17885157788044242, + "learning_rate": 6.29268292682927e-05, + "loss": 1.2455, + "step": 129 + }, + { + "epoch": 0.24, + "grad_norm": 0.17617298187845123, + "learning_rate": 6.341463414634148e-05, + "loss": 1.2009, + "step": 130 + }, + { + "epoch": 0.24, + "grad_norm": 0.20164176323497066, + "learning_rate": 6.390243902439025e-05, + "loss": 1.2634, + "step": 131 + }, + { + "epoch": 0.24, + "grad_norm": 0.20459903417307612, + "learning_rate": 6.439024390243903e-05, + "loss": 1.1963, + "step": 132 + }, + { + "epoch": 0.24, + "grad_norm": 0.1863755486334296, + "learning_rate": 6.487804878048781e-05, + "loss": 1.2387, + "step": 133 + }, + { + "epoch": 0.25, + "grad_norm": 0.19265866140295207, + "learning_rate": 6.536585365853658e-05, + "loss": 1.2688, + "step": 134 + }, + { + "epoch": 0.25, + "grad_norm": 0.1823425868969493, + "learning_rate": 6.585365853658536e-05, + "loss": 1.2041, + "step": 135 + }, + { + "epoch": 0.25, + "grad_norm": 0.2016853266472781, + "learning_rate": 6.634146341463415e-05, + "loss": 1.1223, + "step": 136 + }, + { + "epoch": 0.25, + "grad_norm": 0.17282675192463448, + "learning_rate": 6.682926829268293e-05, + "loss": 1.1879, + "step": 137 + }, + { + "epoch": 0.25, + "grad_norm": 0.17398811693399288, + "learning_rate": 6.731707317073171e-05, + "loss": 1.2682, + "step": 138 + }, + { + "epoch": 0.25, + "grad_norm": 0.18516916965434696, + "learning_rate": 6.78048780487805e-05, + "loss": 1.1666, + "step": 139 + }, + { + "epoch": 0.26, + "grad_norm": 0.1852213129647933, + "learning_rate": 6.829268292682927e-05, + "loss": 1.2501, + "step": 140 + }, + { + "epoch": 0.26, + "grad_norm": 0.17915948766591883, + "learning_rate": 6.878048780487805e-05, + "loss": 1.2264, + "step": 141 + }, + { + "epoch": 0.26, + "grad_norm": 0.21599939417233183, + "learning_rate": 6.926829268292683e-05, + "loss": 1.2376, + "step": 142 + }, + { + "epoch": 0.26, + "grad_norm": 0.17839304459521851, + "learning_rate": 6.975609756097562e-05, + "loss": 1.2353, + "step": 143 + }, + { + "epoch": 0.26, + "grad_norm": 0.20826913231380875, + "learning_rate": 7.02439024390244e-05, + "loss": 1.1901, + "step": 144 + }, + { + "epoch": 0.27, + "grad_norm": 0.20788894913361589, + "learning_rate": 7.073170731707318e-05, + "loss": 1.2577, + "step": 145 + }, + { + "epoch": 0.27, + "grad_norm": 0.18420055842301297, + "learning_rate": 7.121951219512195e-05, + "loss": 1.1393, + "step": 146 + }, + { + "epoch": 0.27, + "grad_norm": 0.19903048468685589, + "learning_rate": 7.170731707317073e-05, + "loss": 1.2321, + "step": 147 + }, + { + "epoch": 0.27, + "grad_norm": 0.19074116314985748, + "learning_rate": 7.219512195121952e-05, + "loss": 1.1912, + "step": 148 + }, + { + "epoch": 0.27, + "grad_norm": 0.2353816469403903, + "learning_rate": 7.26829268292683e-05, + "loss": 1.28, + "step": 149 + }, + { + "epoch": 0.27, + "grad_norm": 0.21634875684769345, + "learning_rate": 7.317073170731708e-05, + "loss": 1.3312, + "step": 150 + }, + { + "epoch": 0.28, + "grad_norm": 0.18290969006743918, + "learning_rate": 7.365853658536587e-05, + "loss": 1.2214, + "step": 151 + }, + { + "epoch": 0.28, + "grad_norm": 0.18484243897545208, + "learning_rate": 7.414634146341465e-05, + "loss": 1.1895, + "step": 152 + }, + { + "epoch": 0.28, + "grad_norm": 0.21882343112978872, + "learning_rate": 7.463414634146342e-05, + "loss": 1.2219, + "step": 153 + }, + { + "epoch": 0.28, + "grad_norm": 0.19868284379241205, + "learning_rate": 7.51219512195122e-05, + "loss": 1.2176, + "step": 154 + }, + { + "epoch": 0.28, + "grad_norm": 0.20912516312950613, + "learning_rate": 7.560975609756097e-05, + "loss": 1.242, + "step": 155 + }, + { + "epoch": 0.29, + "grad_norm": 0.23811880045549916, + "learning_rate": 7.609756097560976e-05, + "loss": 1.2838, + "step": 156 + }, + { + "epoch": 0.29, + "grad_norm": 0.19511077122033713, + "learning_rate": 7.658536585365854e-05, + "loss": 1.1594, + "step": 157 + }, + { + "epoch": 0.29, + "grad_norm": 0.20094129399534238, + "learning_rate": 7.707317073170732e-05, + "loss": 1.2966, + "step": 158 + }, + { + "epoch": 0.29, + "grad_norm": 0.19366245038292418, + "learning_rate": 7.75609756097561e-05, + "loss": 1.2246, + "step": 159 + }, + { + "epoch": 0.29, + "grad_norm": 0.19409570223867306, + "learning_rate": 7.804878048780489e-05, + "loss": 1.2312, + "step": 160 + }, + { + "epoch": 0.29, + "grad_norm": 0.2087258457033805, + "learning_rate": 7.853658536585366e-05, + "loss": 1.2169, + "step": 161 + }, + { + "epoch": 0.3, + "grad_norm": 0.18765223996270428, + "learning_rate": 7.902439024390244e-05, + "loss": 1.2383, + "step": 162 + }, + { + "epoch": 0.3, + "grad_norm": 0.20734180224147242, + "learning_rate": 7.951219512195122e-05, + "loss": 1.2587, + "step": 163 + }, + { + "epoch": 0.3, + "grad_norm": 0.24690929540287834, + "learning_rate": 8e-05, + "loss": 1.1951, + "step": 164 + }, + { + "epoch": 0.3, + "grad_norm": 0.2003538797619543, + "learning_rate": 7.999990914797545e-05, + "loss": 1.1982, + "step": 165 + }, + { + "epoch": 0.3, + "grad_norm": 0.22469075613510484, + "learning_rate": 7.99996365923145e-05, + "loss": 1.2355, + "step": 166 + }, + { + "epoch": 0.31, + "grad_norm": 0.21870100788336058, + "learning_rate": 7.999918233425526e-05, + "loss": 1.1103, + "step": 167 + }, + { + "epoch": 0.31, + "grad_norm": 0.20939989594131886, + "learning_rate": 7.999854637586122e-05, + "loss": 1.1966, + "step": 168 + }, + { + "epoch": 0.31, + "grad_norm": 0.43108211416237796, + "learning_rate": 7.999772872002132e-05, + "loss": 1.2882, + "step": 169 + }, + { + "epoch": 0.31, + "grad_norm": 0.27045413432174487, + "learning_rate": 7.999672937044984e-05, + "loss": 1.2399, + "step": 170 + }, + { + "epoch": 0.31, + "grad_norm": 0.19700483036740515, + "learning_rate": 7.999554833168642e-05, + "loss": 1.202, + "step": 171 + }, + { + "epoch": 0.31, + "grad_norm": 0.3335979493370708, + "learning_rate": 7.999418560909604e-05, + "loss": 1.1995, + "step": 172 + }, + { + "epoch": 0.32, + "grad_norm": 0.3165803974474567, + "learning_rate": 7.999264120886902e-05, + "loss": 1.1569, + "step": 173 + }, + { + "epoch": 0.32, + "grad_norm": 0.1951699080346223, + "learning_rate": 7.999091513802093e-05, + "loss": 1.1778, + "step": 174 + }, + { + "epoch": 0.32, + "grad_norm": 0.2087559121749787, + "learning_rate": 7.998900740439265e-05, + "loss": 1.1736, + "step": 175 + }, + { + "epoch": 0.32, + "grad_norm": 0.20345180977460478, + "learning_rate": 7.998691801665024e-05, + "loss": 1.2281, + "step": 176 + }, + { + "epoch": 0.32, + "grad_norm": 0.24617644827252333, + "learning_rate": 7.998464698428495e-05, + "loss": 1.2072, + "step": 177 + }, + { + "epoch": 0.33, + "grad_norm": 0.2469050959356265, + "learning_rate": 7.998219431761318e-05, + "loss": 1.2242, + "step": 178 + }, + { + "epoch": 0.33, + "grad_norm": 0.19529317748460623, + "learning_rate": 7.997956002777642e-05, + "loss": 1.2567, + "step": 179 + }, + { + "epoch": 0.33, + "grad_norm": 0.19048389491381376, + "learning_rate": 7.99767441267412e-05, + "loss": 1.2982, + "step": 180 + }, + { + "epoch": 0.33, + "grad_norm": 0.2085799116493225, + "learning_rate": 7.997374662729904e-05, + "loss": 1.1254, + "step": 181 + }, + { + "epoch": 0.33, + "grad_norm": 0.20636853256378995, + "learning_rate": 7.997056754306636e-05, + "loss": 1.2435, + "step": 182 + }, + { + "epoch": 0.33, + "grad_norm": 0.20590016382290252, + "learning_rate": 7.99672068884845e-05, + "loss": 1.2658, + "step": 183 + }, + { + "epoch": 0.34, + "grad_norm": 0.1931166169764433, + "learning_rate": 7.996366467881955e-05, + "loss": 1.1637, + "step": 184 + }, + { + "epoch": 0.34, + "grad_norm": 0.18873318157988098, + "learning_rate": 7.995994093016237e-05, + "loss": 1.1335, + "step": 185 + }, + { + "epoch": 0.34, + "grad_norm": 0.19210254625199108, + "learning_rate": 7.995603565942846e-05, + "loss": 1.1928, + "step": 186 + }, + { + "epoch": 0.34, + "grad_norm": 0.2130986479765664, + "learning_rate": 7.995194888435792e-05, + "loss": 1.2158, + "step": 187 + }, + { + "epoch": 0.34, + "grad_norm": 0.22003854501814088, + "learning_rate": 7.994768062351532e-05, + "loss": 1.2288, + "step": 188 + }, + { + "epoch": 0.35, + "grad_norm": 0.20330803191993058, + "learning_rate": 7.994323089628968e-05, + "loss": 1.2426, + "step": 189 + }, + { + "epoch": 0.35, + "grad_norm": 0.20567314642208634, + "learning_rate": 7.993859972289434e-05, + "loss": 1.2649, + "step": 190 + }, + { + "epoch": 0.35, + "grad_norm": 0.21556663727342962, + "learning_rate": 7.993378712436686e-05, + "loss": 1.2545, + "step": 191 + }, + { + "epoch": 0.35, + "grad_norm": 0.20309165469109888, + "learning_rate": 7.992879312256897e-05, + "loss": 1.3338, + "step": 192 + }, + { + "epoch": 0.35, + "grad_norm": 0.19574356669421325, + "learning_rate": 7.992361774018641e-05, + "loss": 1.278, + "step": 193 + }, + { + "epoch": 0.35, + "grad_norm": 0.2763613746722313, + "learning_rate": 7.991826100072891e-05, + "loss": 1.2571, + "step": 194 + }, + { + "epoch": 0.36, + "grad_norm": 0.19346552479915102, + "learning_rate": 7.991272292852996e-05, + "loss": 1.2027, + "step": 195 + }, + { + "epoch": 0.36, + "grad_norm": 0.2281167812123908, + "learning_rate": 7.990700354874683e-05, + "loss": 1.2586, + "step": 196 + }, + { + "epoch": 0.36, + "grad_norm": 0.19699013712137542, + "learning_rate": 7.990110288736042e-05, + "loss": 1.1371, + "step": 197 + }, + { + "epoch": 0.36, + "grad_norm": 0.21768209981475933, + "learning_rate": 7.989502097117503e-05, + "loss": 1.2522, + "step": 198 + }, + { + "epoch": 0.36, + "grad_norm": 0.21335427847754582, + "learning_rate": 7.988875782781838e-05, + "loss": 1.2437, + "step": 199 + }, + { + "epoch": 0.37, + "grad_norm": 0.21856710629066897, + "learning_rate": 7.988231348574147e-05, + "loss": 1.2135, + "step": 200 + }, + { + "epoch": 0.37, + "grad_norm": 0.20482062658774797, + "learning_rate": 7.987568797421836e-05, + "loss": 1.1755, + "step": 201 + }, + { + "epoch": 0.37, + "grad_norm": 0.2017756813960897, + "learning_rate": 7.986888132334608e-05, + "loss": 1.1699, + "step": 202 + }, + { + "epoch": 0.37, + "grad_norm": 0.20496443848153809, + "learning_rate": 7.986189356404458e-05, + "loss": 1.2125, + "step": 203 + }, + { + "epoch": 0.37, + "grad_norm": 0.2134603800558358, + "learning_rate": 7.985472472805643e-05, + "loss": 1.2391, + "step": 204 + }, + { + "epoch": 0.37, + "grad_norm": 0.2364175573420861, + "learning_rate": 7.98473748479468e-05, + "loss": 1.2384, + "step": 205 + }, + { + "epoch": 0.38, + "grad_norm": 0.1872419861598724, + "learning_rate": 7.983984395710326e-05, + "loss": 1.1457, + "step": 206 + }, + { + "epoch": 0.38, + "grad_norm": 0.28222194007095774, + "learning_rate": 7.983213208973566e-05, + "loss": 1.2952, + "step": 207 + }, + { + "epoch": 0.38, + "grad_norm": 0.1916094851162064, + "learning_rate": 7.982423928087593e-05, + "loss": 1.1763, + "step": 208 + }, + { + "epoch": 0.38, + "grad_norm": 0.18446245256166657, + "learning_rate": 7.981616556637795e-05, + "loss": 1.1863, + "step": 209 + }, + { + "epoch": 0.38, + "grad_norm": 0.195191961022491, + "learning_rate": 7.980791098291737e-05, + "loss": 1.2036, + "step": 210 + }, + { + "epoch": 0.39, + "grad_norm": 0.2652439657825496, + "learning_rate": 7.979947556799151e-05, + "loss": 1.2834, + "step": 211 + }, + { + "epoch": 0.39, + "grad_norm": 0.24308438957843412, + "learning_rate": 7.979085935991906e-05, + "loss": 1.234, + "step": 212 + }, + { + "epoch": 0.39, + "grad_norm": 0.21294701043622016, + "learning_rate": 7.978206239784004e-05, + "loss": 1.3006, + "step": 213 + }, + { + "epoch": 0.39, + "grad_norm": 0.25809277041859524, + "learning_rate": 7.977308472171553e-05, + "loss": 1.2272, + "step": 214 + }, + { + "epoch": 0.39, + "grad_norm": 0.193463860107294, + "learning_rate": 7.976392637232754e-05, + "loss": 1.2295, + "step": 215 + }, + { + "epoch": 0.4, + "grad_norm": 0.2150023760609626, + "learning_rate": 7.975458739127877e-05, + "loss": 1.2135, + "step": 216 + }, + { + "epoch": 0.4, + "grad_norm": 0.22590495955605894, + "learning_rate": 7.974506782099253e-05, + "loss": 1.2532, + "step": 217 + }, + { + "epoch": 0.4, + "grad_norm": 0.21023744668403702, + "learning_rate": 7.973536770471242e-05, + "loss": 1.2472, + "step": 218 + }, + { + "epoch": 0.4, + "grad_norm": 0.2345749799511543, + "learning_rate": 7.972548708650218e-05, + "loss": 1.1791, + "step": 219 + }, + { + "epoch": 0.4, + "grad_norm": 0.2158876734005217, + "learning_rate": 7.971542601124553e-05, + "loss": 1.2483, + "step": 220 + }, + { + "epoch": 0.4, + "grad_norm": 0.29455339949432446, + "learning_rate": 7.970518452464593e-05, + "loss": 1.2894, + "step": 221 + }, + { + "epoch": 0.41, + "grad_norm": 0.23983708730626851, + "learning_rate": 7.969476267322636e-05, + "loss": 1.271, + "step": 222 + }, + { + "epoch": 0.41, + "grad_norm": 0.1922400905426158, + "learning_rate": 7.968416050432912e-05, + "loss": 1.2139, + "step": 223 + }, + { + "epoch": 0.41, + "grad_norm": 0.2238136844422931, + "learning_rate": 7.967337806611568e-05, + "loss": 1.2655, + "step": 224 + }, + { + "epoch": 0.41, + "grad_norm": 0.21230292828267672, + "learning_rate": 7.966241540756631e-05, + "loss": 1.2406, + "step": 225 + }, + { + "epoch": 0.41, + "grad_norm": 0.26656119419070456, + "learning_rate": 7.965127257848004e-05, + "loss": 1.2595, + "step": 226 + }, + { + "epoch": 0.42, + "grad_norm": 0.22381385502992684, + "learning_rate": 7.963994962947426e-05, + "loss": 1.1737, + "step": 227 + }, + { + "epoch": 0.42, + "grad_norm": 0.20056702203994298, + "learning_rate": 7.962844661198462e-05, + "loss": 1.1969, + "step": 228 + }, + { + "epoch": 0.42, + "grad_norm": 0.20148701321526885, + "learning_rate": 7.961676357826478e-05, + "loss": 1.2151, + "step": 229 + }, + { + "epoch": 0.42, + "grad_norm": 0.20034834807028637, + "learning_rate": 7.960490058138604e-05, + "loss": 1.1455, + "step": 230 + }, + { + "epoch": 0.42, + "grad_norm": 0.21050838521846033, + "learning_rate": 7.959285767523732e-05, + "loss": 1.2223, + "step": 231 + }, + { + "epoch": 0.42, + "grad_norm": 0.20904772138969777, + "learning_rate": 7.95806349145247e-05, + "loss": 1.2534, + "step": 232 + }, + { + "epoch": 0.43, + "grad_norm": 0.20307877304792957, + "learning_rate": 7.956823235477134e-05, + "loss": 1.1352, + "step": 233 + }, + { + "epoch": 0.43, + "grad_norm": 0.20501105270897094, + "learning_rate": 7.95556500523171e-05, + "loss": 1.2031, + "step": 234 + }, + { + "epoch": 0.43, + "grad_norm": 0.19800586972038586, + "learning_rate": 7.954288806431838e-05, + "loss": 1.2567, + "step": 235 + }, + { + "epoch": 0.43, + "grad_norm": 0.2175102450594135, + "learning_rate": 7.952994644874777e-05, + "loss": 1.2538, + "step": 236 + }, + { + "epoch": 0.43, + "grad_norm": 0.22698189300067595, + "learning_rate": 7.951682526439391e-05, + "loss": 1.3088, + "step": 237 + }, + { + "epoch": 0.44, + "grad_norm": 0.19208392014975315, + "learning_rate": 7.950352457086109e-05, + "loss": 1.2336, + "step": 238 + }, + { + "epoch": 0.44, + "grad_norm": 0.27004086334319655, + "learning_rate": 7.949004442856905e-05, + "loss": 1.2012, + "step": 239 + }, + { + "epoch": 0.44, + "grad_norm": 0.23420974954538043, + "learning_rate": 7.947638489875272e-05, + "loss": 1.2244, + "step": 240 + }, + { + "epoch": 0.44, + "grad_norm": 0.20514399124802024, + "learning_rate": 7.946254604346186e-05, + "loss": 1.2548, + "step": 241 + }, + { + "epoch": 0.44, + "grad_norm": 0.19334973602372896, + "learning_rate": 7.944852792556092e-05, + "loss": 1.2104, + "step": 242 + }, + { + "epoch": 0.44, + "grad_norm": 0.1992640714537956, + "learning_rate": 7.943433060872858e-05, + "loss": 1.2628, + "step": 243 + }, + { + "epoch": 0.45, + "grad_norm": 0.203284617090413, + "learning_rate": 7.941995415745761e-05, + "loss": 1.2002, + "step": 244 + }, + { + "epoch": 0.45, + "grad_norm": 0.22795306969682058, + "learning_rate": 7.94053986370545e-05, + "loss": 1.2215, + "step": 245 + }, + { + "epoch": 0.45, + "grad_norm": 0.20789041346838505, + "learning_rate": 7.939066411363915e-05, + "loss": 1.0998, + "step": 246 + }, + { + "epoch": 0.45, + "grad_norm": 0.22354868884742066, + "learning_rate": 7.937575065414464e-05, + "loss": 1.2564, + "step": 247 + }, + { + "epoch": 0.45, + "grad_norm": 0.21176392726647736, + "learning_rate": 7.936065832631687e-05, + "loss": 1.2816, + "step": 248 + }, + { + "epoch": 0.46, + "grad_norm": 0.19967179557235587, + "learning_rate": 7.934538719871427e-05, + "loss": 1.1961, + "step": 249 + }, + { + "epoch": 0.46, + "grad_norm": 0.210819577350627, + "learning_rate": 7.932993734070747e-05, + "loss": 1.2167, + "step": 250 + }, + { + "epoch": 0.46, + "grad_norm": 0.21537794551756187, + "learning_rate": 7.931430882247903e-05, + "loss": 1.2341, + "step": 251 + }, + { + "epoch": 0.46, + "grad_norm": 0.22850872387256574, + "learning_rate": 7.929850171502304e-05, + "loss": 1.1686, + "step": 252 + }, + { + "epoch": 0.46, + "grad_norm": 0.22380366415076383, + "learning_rate": 7.928251609014493e-05, + "loss": 1.1462, + "step": 253 + }, + { + "epoch": 0.46, + "grad_norm": 0.22426923149036065, + "learning_rate": 7.926635202046102e-05, + "loss": 1.1792, + "step": 254 + }, + { + "epoch": 0.47, + "grad_norm": 0.42082703321103965, + "learning_rate": 7.925000957939822e-05, + "loss": 1.2718, + "step": 255 + }, + { + "epoch": 0.47, + "grad_norm": 0.2235432774854074, + "learning_rate": 7.92334888411937e-05, + "loss": 1.2598, + "step": 256 + }, + { + "epoch": 0.47, + "grad_norm": 0.281644028934108, + "learning_rate": 7.92167898808946e-05, + "loss": 1.2205, + "step": 257 + }, + { + "epoch": 0.47, + "grad_norm": 0.2037705143888748, + "learning_rate": 7.919991277435763e-05, + "loss": 1.1737, + "step": 258 + }, + { + "epoch": 0.47, + "grad_norm": 0.20917419230028977, + "learning_rate": 7.918285759824879e-05, + "loss": 1.2035, + "step": 259 + }, + { + "epoch": 0.48, + "grad_norm": 0.20510847570635518, + "learning_rate": 7.916562443004292e-05, + "loss": 1.2135, + "step": 260 + }, + { + "epoch": 0.48, + "grad_norm": 0.25172483071092466, + "learning_rate": 7.914821334802342e-05, + "loss": 1.2218, + "step": 261 + }, + { + "epoch": 0.48, + "grad_norm": 0.21102706700634313, + "learning_rate": 7.91306244312819e-05, + "loss": 1.1738, + "step": 262 + }, + { + "epoch": 0.48, + "grad_norm": 0.22626060872645815, + "learning_rate": 7.911285775971781e-05, + "loss": 1.238, + "step": 263 + }, + { + "epoch": 0.48, + "grad_norm": 0.22448567539778486, + "learning_rate": 7.909491341403805e-05, + "loss": 1.2404, + "step": 264 + }, + { + "epoch": 0.48, + "grad_norm": 0.2019099786139193, + "learning_rate": 7.907679147575661e-05, + "loss": 1.213, + "step": 265 + }, + { + "epoch": 0.49, + "grad_norm": 0.24307234839096267, + "learning_rate": 7.905849202719422e-05, + "loss": 1.2322, + "step": 266 + }, + { + "epoch": 0.49, + "grad_norm": 0.19801890521743487, + "learning_rate": 7.904001515147802e-05, + "loss": 1.2448, + "step": 267 + }, + { + "epoch": 0.49, + "grad_norm": 0.2102742273575385, + "learning_rate": 7.902136093254106e-05, + "loss": 1.1657, + "step": 268 + }, + { + "epoch": 0.49, + "grad_norm": 0.2173464476815016, + "learning_rate": 7.900252945512201e-05, + "loss": 1.2549, + "step": 269 + }, + { + "epoch": 0.49, + "grad_norm": 0.20957275458699595, + "learning_rate": 7.898352080476479e-05, + "loss": 1.2536, + "step": 270 + }, + { + "epoch": 0.5, + "grad_norm": 0.20691966388952363, + "learning_rate": 7.896433506781811e-05, + "loss": 1.2661, + "step": 271 + }, + { + "epoch": 0.5, + "grad_norm": 0.2276662275112648, + "learning_rate": 7.894497233143509e-05, + "loss": 1.2409, + "step": 272 + }, + { + "epoch": 0.5, + "grad_norm": 0.23854109569301263, + "learning_rate": 7.892543268357297e-05, + "loss": 1.2681, + "step": 273 + }, + { + "epoch": 0.5, + "grad_norm": 0.2233864156677627, + "learning_rate": 7.890571621299252e-05, + "loss": 1.1687, + "step": 274 + }, + { + "epoch": 0.5, + "grad_norm": 0.20114129147925475, + "learning_rate": 7.888582300925787e-05, + "loss": 1.2184, + "step": 275 + }, + { + "epoch": 0.5, + "grad_norm": 0.2154654670569462, + "learning_rate": 7.886575316273586e-05, + "loss": 1.1982, + "step": 276 + }, + { + "epoch": 0.51, + "grad_norm": 0.2292982209343639, + "learning_rate": 7.884550676459583e-05, + "loss": 1.2129, + "step": 277 + }, + { + "epoch": 0.51, + "grad_norm": 0.21302713135229548, + "learning_rate": 7.882508390680908e-05, + "loss": 1.1605, + "step": 278 + }, + { + "epoch": 0.51, + "grad_norm": 0.2123661020671048, + "learning_rate": 7.88044846821485e-05, + "loss": 1.2308, + "step": 279 + }, + { + "epoch": 0.51, + "grad_norm": 0.2080577410800404, + "learning_rate": 7.878370918418818e-05, + "loss": 1.2195, + "step": 280 + }, + { + "epoch": 0.51, + "grad_norm": 0.19663901881127385, + "learning_rate": 7.876275750730289e-05, + "loss": 1.1591, + "step": 281 + }, + { + "epoch": 0.52, + "grad_norm": 0.20534502031312163, + "learning_rate": 7.874162974666776e-05, + "loss": 1.2664, + "step": 282 + }, + { + "epoch": 0.52, + "grad_norm": 0.23240445399513837, + "learning_rate": 7.872032599825779e-05, + "loss": 1.2151, + "step": 283 + }, + { + "epoch": 0.52, + "grad_norm": 0.2672527316717507, + "learning_rate": 7.86988463588474e-05, + "loss": 1.2406, + "step": 284 + }, + { + "epoch": 0.52, + "grad_norm": 0.19893903058743695, + "learning_rate": 7.867719092601003e-05, + "loss": 1.1291, + "step": 285 + }, + { + "epoch": 0.52, + "grad_norm": 0.33275268109930917, + "learning_rate": 7.865535979811768e-05, + "loss": 1.1406, + "step": 286 + }, + { + "epoch": 0.52, + "grad_norm": 0.2373619455690358, + "learning_rate": 7.863335307434045e-05, + "loss": 1.2799, + "step": 287 + }, + { + "epoch": 0.53, + "grad_norm": 0.263235735390858, + "learning_rate": 7.861117085464612e-05, + "loss": 1.2415, + "step": 288 + }, + { + "epoch": 0.53, + "grad_norm": 0.25884281780784324, + "learning_rate": 7.858881323979965e-05, + "loss": 1.3919, + "step": 289 + }, + { + "epoch": 0.53, + "grad_norm": 0.25426288332255736, + "learning_rate": 7.85662803313628e-05, + "loss": 1.174, + "step": 290 + }, + { + "epoch": 0.53, + "grad_norm": 0.26655405527881243, + "learning_rate": 7.854357223169356e-05, + "loss": 1.2806, + "step": 291 + }, + { + "epoch": 0.53, + "grad_norm": 0.20909844432349833, + "learning_rate": 7.852068904394579e-05, + "loss": 1.2627, + "step": 292 + }, + { + "epoch": 0.54, + "grad_norm": 0.21307115068935759, + "learning_rate": 7.849763087206866e-05, + "loss": 1.1879, + "step": 293 + }, + { + "epoch": 0.54, + "grad_norm": 0.25009949471398946, + "learning_rate": 7.847439782080628e-05, + "loss": 1.2881, + "step": 294 + }, + { + "epoch": 0.54, + "grad_norm": 0.20960783418679174, + "learning_rate": 7.845098999569712e-05, + "loss": 1.2723, + "step": 295 + }, + { + "epoch": 0.54, + "grad_norm": 0.24968832437925104, + "learning_rate": 7.842740750307362e-05, + "loss": 1.2029, + "step": 296 + }, + { + "epoch": 0.54, + "grad_norm": 0.22981196585125677, + "learning_rate": 7.84036504500616e-05, + "loss": 1.1695, + "step": 297 + }, + { + "epoch": 0.55, + "grad_norm": 0.2320606844751365, + "learning_rate": 7.837971894457991e-05, + "loss": 1.2317, + "step": 298 + }, + { + "epoch": 0.55, + "grad_norm": 0.23051459673906124, + "learning_rate": 7.835561309533981e-05, + "loss": 1.2046, + "step": 299 + }, + { + "epoch": 0.55, + "grad_norm": 0.2510027231060586, + "learning_rate": 7.833133301184457e-05, + "loss": 1.199, + "step": 300 + }, + { + "epoch": 0.55, + "grad_norm": 0.23601180466018787, + "learning_rate": 7.830687880438895e-05, + "loss": 1.1755, + "step": 301 + }, + { + "epoch": 0.55, + "grad_norm": 0.24740820934385369, + "learning_rate": 7.828225058405864e-05, + "loss": 1.2054, + "step": 302 + }, + { + "epoch": 0.55, + "grad_norm": 0.23065372979111173, + "learning_rate": 7.825744846272984e-05, + "loss": 1.2066, + "step": 303 + }, + { + "epoch": 0.56, + "grad_norm": 0.22385077334838213, + "learning_rate": 7.823247255306866e-05, + "loss": 1.2147, + "step": 304 + }, + { + "epoch": 0.56, + "grad_norm": 0.42981213948386104, + "learning_rate": 7.820732296853074e-05, + "loss": 1.2314, + "step": 305 + }, + { + "epoch": 0.56, + "grad_norm": 0.21122844902751076, + "learning_rate": 7.818199982336058e-05, + "loss": 1.1462, + "step": 306 + }, + { + "epoch": 0.56, + "grad_norm": 0.23374869692118933, + "learning_rate": 7.815650323259117e-05, + "loss": 1.2051, + "step": 307 + }, + { + "epoch": 0.56, + "grad_norm": 0.21662363795962128, + "learning_rate": 7.813083331204332e-05, + "loss": 1.1575, + "step": 308 + }, + { + "epoch": 0.57, + "grad_norm": 0.2088315773384112, + "learning_rate": 7.810499017832526e-05, + "loss": 1.1316, + "step": 309 + }, + { + "epoch": 0.57, + "grad_norm": 0.2095238410730976, + "learning_rate": 7.807897394883203e-05, + "loss": 1.2087, + "step": 310 + }, + { + "epoch": 0.57, + "grad_norm": 0.22672932127256515, + "learning_rate": 7.805278474174499e-05, + "loss": 1.2512, + "step": 311 + }, + { + "epoch": 0.57, + "grad_norm": 0.21873052340922736, + "learning_rate": 7.802642267603126e-05, + "loss": 1.1909, + "step": 312 + }, + { + "epoch": 0.57, + "grad_norm": 0.219814521916342, + "learning_rate": 7.79998878714432e-05, + "loss": 1.1669, + "step": 313 + }, + { + "epoch": 0.57, + "grad_norm": 0.3049426027257317, + "learning_rate": 7.797318044851786e-05, + "loss": 1.1797, + "step": 314 + }, + { + "epoch": 0.58, + "grad_norm": 0.22309435690065985, + "learning_rate": 7.794630052857638e-05, + "loss": 1.1417, + "step": 315 + }, + { + "epoch": 0.58, + "grad_norm": 0.3891885169154885, + "learning_rate": 7.791924823372354e-05, + "loss": 1.2369, + "step": 316 + }, + { + "epoch": 0.58, + "grad_norm": 0.24780269452456372, + "learning_rate": 7.789202368684711e-05, + "loss": 1.2521, + "step": 317 + }, + { + "epoch": 0.58, + "grad_norm": 0.21660460720269362, + "learning_rate": 7.786462701161738e-05, + "loss": 1.2151, + "step": 318 + }, + { + "epoch": 0.58, + "grad_norm": 0.23635409466561857, + "learning_rate": 7.783705833248649e-05, + "loss": 1.2363, + "step": 319 + }, + { + "epoch": 0.59, + "grad_norm": 0.2616135839903218, + "learning_rate": 7.780931777468797e-05, + "loss": 1.2428, + "step": 320 + }, + { + "epoch": 0.59, + "grad_norm": 0.21461059159245083, + "learning_rate": 7.77814054642361e-05, + "loss": 1.1434, + "step": 321 + }, + { + "epoch": 0.59, + "grad_norm": 0.25348824286656163, + "learning_rate": 7.775332152792539e-05, + "loss": 1.2368, + "step": 322 + }, + { + "epoch": 0.59, + "grad_norm": 0.22275034726331247, + "learning_rate": 7.772506609332995e-05, + "loss": 1.1827, + "step": 323 + }, + { + "epoch": 0.59, + "grad_norm": 0.25030821228147526, + "learning_rate": 7.769663928880298e-05, + "loss": 1.2428, + "step": 324 + }, + { + "epoch": 0.59, + "grad_norm": 0.22251804398745534, + "learning_rate": 7.766804124347608e-05, + "loss": 1.1889, + "step": 325 + }, + { + "epoch": 0.6, + "grad_norm": 0.23381455520411995, + "learning_rate": 7.763927208725879e-05, + "loss": 1.2115, + "step": 326 + }, + { + "epoch": 0.6, + "grad_norm": 0.27341902651946226, + "learning_rate": 7.761033195083791e-05, + "loss": 1.2535, + "step": 327 + }, + { + "epoch": 0.6, + "grad_norm": 0.24862471659814522, + "learning_rate": 7.758122096567694e-05, + "loss": 1.2128, + "step": 328 + }, + { + "epoch": 0.6, + "grad_norm": 0.2251357082045494, + "learning_rate": 7.755193926401547e-05, + "loss": 1.2334, + "step": 329 + }, + { + "epoch": 0.6, + "grad_norm": 0.3173274941622932, + "learning_rate": 7.752248697886857e-05, + "loss": 1.226, + "step": 330 + }, + { + "epoch": 0.61, + "grad_norm": 0.23056440717672175, + "learning_rate": 7.74928642440263e-05, + "loss": 1.2339, + "step": 331 + }, + { + "epoch": 0.61, + "grad_norm": 0.2801507500859342, + "learning_rate": 7.746307119405286e-05, + "loss": 1.287, + "step": 332 + }, + { + "epoch": 0.61, + "grad_norm": 0.2267818430426272, + "learning_rate": 7.743310796428622e-05, + "loss": 1.1916, + "step": 333 + }, + { + "epoch": 0.61, + "grad_norm": 0.2777329160365585, + "learning_rate": 7.74029746908374e-05, + "loss": 1.252, + "step": 334 + }, + { + "epoch": 0.61, + "grad_norm": 0.25289169762353, + "learning_rate": 7.737267151058983e-05, + "loss": 1.2153, + "step": 335 + }, + { + "epoch": 0.61, + "grad_norm": 0.2424670686901653, + "learning_rate": 7.734219856119875e-05, + "loss": 1.2227, + "step": 336 + }, + { + "epoch": 0.62, + "grad_norm": 0.22747092217441645, + "learning_rate": 7.731155598109067e-05, + "loss": 1.19, + "step": 337 + }, + { + "epoch": 0.62, + "grad_norm": 0.2307810940100189, + "learning_rate": 7.728074390946257e-05, + "loss": 1.1818, + "step": 338 + }, + { + "epoch": 0.62, + "grad_norm": 0.2583402574655623, + "learning_rate": 7.724976248628142e-05, + "loss": 1.1608, + "step": 339 + }, + { + "epoch": 0.62, + "grad_norm": 0.22140209760890694, + "learning_rate": 7.721861185228347e-05, + "loss": 1.1245, + "step": 340 + }, + { + "epoch": 0.62, + "grad_norm": 0.25859310758244686, + "learning_rate": 7.718729214897362e-05, + "loss": 1.2247, + "step": 341 + }, + { + "epoch": 0.63, + "grad_norm": 0.26371179531372124, + "learning_rate": 7.715580351862482e-05, + "loss": 1.2128, + "step": 342 + }, + { + "epoch": 0.63, + "grad_norm": 0.26575541302851047, + "learning_rate": 7.712414610427733e-05, + "loss": 1.2443, + "step": 343 + }, + { + "epoch": 0.63, + "grad_norm": 0.269978305197599, + "learning_rate": 7.709232004973816e-05, + "loss": 1.2231, + "step": 344 + }, + { + "epoch": 0.63, + "grad_norm": 0.26583998705977047, + "learning_rate": 7.70603254995804e-05, + "loss": 1.2476, + "step": 345 + }, + { + "epoch": 0.63, + "grad_norm": 0.24256062164066097, + "learning_rate": 7.702816259914253e-05, + "loss": 1.2901, + "step": 346 + }, + { + "epoch": 0.63, + "grad_norm": 0.3463123472658915, + "learning_rate": 7.699583149452779e-05, + "loss": 1.3277, + "step": 347 + }, + { + "epoch": 0.64, + "grad_norm": 0.2269096590531878, + "learning_rate": 7.696333233260345e-05, + "loss": 1.2047, + "step": 348 + }, + { + "epoch": 0.64, + "grad_norm": 0.25136883001050025, + "learning_rate": 7.693066526100031e-05, + "loss": 1.1619, + "step": 349 + }, + { + "epoch": 0.64, + "grad_norm": 0.2565112571116145, + "learning_rate": 7.68978304281118e-05, + "loss": 1.2389, + "step": 350 + }, + { + "epoch": 0.64, + "grad_norm": 0.22175779550828703, + "learning_rate": 7.686482798309349e-05, + "loss": 1.2238, + "step": 351 + }, + { + "epoch": 0.64, + "grad_norm": 0.22588304332216555, + "learning_rate": 7.683165807586234e-05, + "loss": 1.174, + "step": 352 + }, + { + "epoch": 0.65, + "grad_norm": 0.24889474296529737, + "learning_rate": 7.6798320857096e-05, + "loss": 1.2366, + "step": 353 + }, + { + "epoch": 0.65, + "grad_norm": 0.27339703806525034, + "learning_rate": 7.676481647823214e-05, + "loss": 1.2356, + "step": 354 + }, + { + "epoch": 0.65, + "grad_norm": 0.23424666722888365, + "learning_rate": 7.673114509146782e-05, + "loss": 1.2089, + "step": 355 + }, + { + "epoch": 0.65, + "grad_norm": 0.27978285392461766, + "learning_rate": 7.66973068497587e-05, + "loss": 1.2609, + "step": 356 + }, + { + "epoch": 0.65, + "grad_norm": 0.2509423350138824, + "learning_rate": 7.666330190681844e-05, + "loss": 1.1777, + "step": 357 + }, + { + "epoch": 0.65, + "grad_norm": 0.23007730927468031, + "learning_rate": 7.662913041711793e-05, + "loss": 1.154, + "step": 358 + }, + { + "epoch": 0.66, + "grad_norm": 0.2438648674953112, + "learning_rate": 7.659479253588462e-05, + "loss": 1.2257, + "step": 359 + }, + { + "epoch": 0.66, + "grad_norm": 0.28816093242092233, + "learning_rate": 7.65602884191018e-05, + "loss": 1.2558, + "step": 360 + }, + { + "epoch": 0.66, + "grad_norm": 0.24972815300596035, + "learning_rate": 7.652561822350793e-05, + "loss": 1.2837, + "step": 361 + }, + { + "epoch": 0.66, + "grad_norm": 0.2543189139697063, + "learning_rate": 7.649078210659587e-05, + "loss": 1.2193, + "step": 362 + }, + { + "epoch": 0.66, + "grad_norm": 0.2237937956718952, + "learning_rate": 7.645578022661224e-05, + "loss": 1.2237, + "step": 363 + }, + { + "epoch": 0.67, + "grad_norm": 0.29742029408787396, + "learning_rate": 7.642061274255657e-05, + "loss": 1.2116, + "step": 364 + }, + { + "epoch": 0.67, + "grad_norm": 0.2462883147335493, + "learning_rate": 7.638527981418075e-05, + "loss": 1.1827, + "step": 365 + }, + { + "epoch": 0.67, + "grad_norm": 0.2647802498907096, + "learning_rate": 7.634978160198817e-05, + "loss": 1.2739, + "step": 366 + }, + { + "epoch": 0.67, + "grad_norm": 0.22360398779217264, + "learning_rate": 7.631411826723306e-05, + "loss": 1.2185, + "step": 367 + }, + { + "epoch": 0.67, + "grad_norm": 0.2635048004593543, + "learning_rate": 7.627828997191973e-05, + "loss": 1.2317, + "step": 368 + }, + { + "epoch": 0.67, + "grad_norm": 0.2764803449917684, + "learning_rate": 7.624229687880184e-05, + "loss": 1.1923, + "step": 369 + }, + { + "epoch": 0.68, + "grad_norm": 0.25724943233414527, + "learning_rate": 7.620613915138166e-05, + "loss": 1.2218, + "step": 370 + }, + { + "epoch": 0.68, + "grad_norm": 0.2858318045794755, + "learning_rate": 7.61698169539093e-05, + "loss": 1.1496, + "step": 371 + }, + { + "epoch": 0.68, + "grad_norm": 0.23547216647460364, + "learning_rate": 7.613333045138206e-05, + "loss": 1.1905, + "step": 372 + }, + { + "epoch": 0.68, + "grad_norm": 0.22984814903684375, + "learning_rate": 7.609667980954355e-05, + "loss": 1.2009, + "step": 373 + }, + { + "epoch": 0.68, + "grad_norm": 0.2551903754079084, + "learning_rate": 7.605986519488301e-05, + "loss": 1.2042, + "step": 374 + }, + { + "epoch": 0.69, + "grad_norm": 0.2508257410125616, + "learning_rate": 7.602288677463457e-05, + "loss": 1.2468, + "step": 375 + }, + { + "epoch": 0.69, + "grad_norm": 0.25324577774935964, + "learning_rate": 7.598574471677644e-05, + "loss": 1.2603, + "step": 376 + }, + { + "epoch": 0.69, + "grad_norm": 0.35888776531769967, + "learning_rate": 7.59484391900302e-05, + "loss": 1.1929, + "step": 377 + }, + { + "epoch": 0.69, + "grad_norm": 0.22048517191014724, + "learning_rate": 7.591097036385994e-05, + "loss": 1.1783, + "step": 378 + }, + { + "epoch": 0.69, + "grad_norm": 0.2781160412746083, + "learning_rate": 7.587333840847162e-05, + "loss": 1.3397, + "step": 379 + }, + { + "epoch": 0.7, + "grad_norm": 0.24033046830332258, + "learning_rate": 7.583554349481222e-05, + "loss": 1.2436, + "step": 380 + }, + { + "epoch": 0.7, + "grad_norm": 0.26413762380260003, + "learning_rate": 7.579758579456893e-05, + "loss": 1.1917, + "step": 381 + }, + { + "epoch": 0.7, + "grad_norm": 0.2390937887338632, + "learning_rate": 7.575946548016847e-05, + "loss": 1.2186, + "step": 382 + }, + { + "epoch": 0.7, + "grad_norm": 0.25131263043429275, + "learning_rate": 7.572118272477622e-05, + "loss": 1.2538, + "step": 383 + }, + { + "epoch": 0.7, + "grad_norm": 0.223974104870702, + "learning_rate": 7.568273770229546e-05, + "loss": 1.2165, + "step": 384 + }, + { + "epoch": 0.7, + "grad_norm": 0.25840356830252875, + "learning_rate": 7.564413058736663e-05, + "loss": 1.1848, + "step": 385 + }, + { + "epoch": 0.71, + "grad_norm": 0.2723156683076603, + "learning_rate": 7.560536155536641e-05, + "loss": 1.1982, + "step": 386 + }, + { + "epoch": 0.71, + "grad_norm": 0.265687427976889, + "learning_rate": 7.556643078240708e-05, + "loss": 1.231, + "step": 387 + }, + { + "epoch": 0.71, + "grad_norm": 0.25152762080976077, + "learning_rate": 7.552733844533562e-05, + "loss": 1.1974, + "step": 388 + }, + { + "epoch": 0.71, + "grad_norm": 0.2366049485053541, + "learning_rate": 7.548808472173292e-05, + "loss": 1.3119, + "step": 389 + }, + { + "epoch": 0.71, + "grad_norm": 0.22092196577077122, + "learning_rate": 7.5448669789913e-05, + "loss": 1.195, + "step": 390 + }, + { + "epoch": 0.72, + "grad_norm": 0.22667521540462374, + "learning_rate": 7.540909382892217e-05, + "loss": 1.1431, + "step": 391 + }, + { + "epoch": 0.72, + "grad_norm": 0.25432207282646513, + "learning_rate": 7.536935701853823e-05, + "loss": 1.2173, + "step": 392 + }, + { + "epoch": 0.72, + "grad_norm": 0.29950506457923864, + "learning_rate": 7.53294595392697e-05, + "loss": 1.1962, + "step": 393 + }, + { + "epoch": 0.72, + "grad_norm": 0.24735689607229913, + "learning_rate": 7.528940157235487e-05, + "loss": 1.2053, + "step": 394 + }, + { + "epoch": 0.72, + "grad_norm": 0.24394198607459663, + "learning_rate": 7.524918329976114e-05, + "loss": 1.1979, + "step": 395 + }, + { + "epoch": 0.72, + "grad_norm": 0.2630369372689188, + "learning_rate": 7.520880490418409e-05, + "loss": 1.2111, + "step": 396 + }, + { + "epoch": 0.73, + "grad_norm": 0.26275028416291457, + "learning_rate": 7.516826656904664e-05, + "loss": 1.2133, + "step": 397 + }, + { + "epoch": 0.73, + "grad_norm": 0.23938074620956928, + "learning_rate": 7.512756847849831e-05, + "loss": 1.1355, + "step": 398 + }, + { + "epoch": 0.73, + "grad_norm": 0.3724960610098138, + "learning_rate": 7.508671081741428e-05, + "loss": 1.2572, + "step": 399 + }, + { + "epoch": 0.73, + "grad_norm": 0.24161685847894723, + "learning_rate": 7.504569377139462e-05, + "loss": 1.1706, + "step": 400 + }, + { + "epoch": 0.73, + "grad_norm": 0.26121591322670523, + "learning_rate": 7.50045175267634e-05, + "loss": 1.2135, + "step": 401 + }, + { + "epoch": 0.74, + "grad_norm": 0.2465579498164775, + "learning_rate": 7.496318227056788e-05, + "loss": 1.1641, + "step": 402 + }, + { + "epoch": 0.74, + "grad_norm": 0.2556288696122787, + "learning_rate": 7.492168819057767e-05, + "loss": 1.2939, + "step": 403 + }, + { + "epoch": 0.74, + "grad_norm": 0.261481216336303, + "learning_rate": 7.488003547528382e-05, + "loss": 1.2026, + "step": 404 + }, + { + "epoch": 0.74, + "grad_norm": 0.2389415135676362, + "learning_rate": 7.483822431389799e-05, + "loss": 1.2131, + "step": 405 + }, + { + "epoch": 0.74, + "grad_norm": 0.2559201956627192, + "learning_rate": 7.479625489635162e-05, + "loss": 1.1246, + "step": 406 + }, + { + "epoch": 0.74, + "grad_norm": 0.27127932491822604, + "learning_rate": 7.475412741329504e-05, + "loss": 1.2429, + "step": 407 + }, + { + "epoch": 0.75, + "grad_norm": 0.27006004008695594, + "learning_rate": 7.47118420560966e-05, + "loss": 1.2388, + "step": 408 + }, + { + "epoch": 0.75, + "grad_norm": 0.23716823297200537, + "learning_rate": 7.466939901684182e-05, + "loss": 1.1264, + "step": 409 + }, + { + "epoch": 0.75, + "grad_norm": 0.2885373898669248, + "learning_rate": 7.462679848833252e-05, + "loss": 1.2786, + "step": 410 + }, + { + "epoch": 0.75, + "grad_norm": 0.49215227598639927, + "learning_rate": 7.458404066408588e-05, + "loss": 1.2386, + "step": 411 + }, + { + "epoch": 0.75, + "grad_norm": 0.24235735604947403, + "learning_rate": 7.454112573833368e-05, + "loss": 1.1423, + "step": 412 + }, + { + "epoch": 0.76, + "grad_norm": 0.2584614748054343, + "learning_rate": 7.449805390602127e-05, + "loss": 1.2669, + "step": 413 + }, + { + "epoch": 0.76, + "grad_norm": 0.23806123085998873, + "learning_rate": 7.445482536280684e-05, + "loss": 1.1763, + "step": 414 + }, + { + "epoch": 0.76, + "grad_norm": 0.24459517607786851, + "learning_rate": 7.441144030506043e-05, + "loss": 1.198, + "step": 415 + }, + { + "epoch": 0.76, + "grad_norm": 0.25801616402700395, + "learning_rate": 7.436789892986304e-05, + "loss": 1.2136, + "step": 416 + }, + { + "epoch": 0.76, + "grad_norm": 0.2814819942392514, + "learning_rate": 7.432420143500578e-05, + "loss": 1.2398, + "step": 417 + }, + { + "epoch": 0.76, + "grad_norm": 0.22134709322606153, + "learning_rate": 7.428034801898893e-05, + "loss": 1.1592, + "step": 418 + }, + { + "epoch": 0.77, + "grad_norm": 0.2899677536995633, + "learning_rate": 7.42363388810211e-05, + "loss": 1.2296, + "step": 419 + }, + { + "epoch": 0.77, + "grad_norm": 0.24005943230262294, + "learning_rate": 7.419217422101822e-05, + "loss": 1.2223, + "step": 420 + }, + { + "epoch": 0.77, + "grad_norm": 0.26417562369496167, + "learning_rate": 7.414785423960275e-05, + "loss": 1.2261, + "step": 421 + }, + { + "epoch": 0.77, + "grad_norm": 0.2580815883535521, + "learning_rate": 7.410337913810271e-05, + "loss": 1.2021, + "step": 422 + }, + { + "epoch": 0.77, + "grad_norm": 0.25242217589496435, + "learning_rate": 7.405874911855071e-05, + "loss": 1.239, + "step": 423 + }, + { + "epoch": 0.78, + "grad_norm": 0.21991733999839932, + "learning_rate": 7.401396438368315e-05, + "loss": 1.1716, + "step": 424 + }, + { + "epoch": 0.78, + "grad_norm": 0.40116538322720213, + "learning_rate": 7.396902513693924e-05, + "loss": 1.2773, + "step": 425 + }, + { + "epoch": 0.78, + "grad_norm": 0.277333939455099, + "learning_rate": 7.392393158246002e-05, + "loss": 1.2574, + "step": 426 + }, + { + "epoch": 0.78, + "grad_norm": 0.27146087746385755, + "learning_rate": 7.387868392508756e-05, + "loss": 1.2243, + "step": 427 + }, + { + "epoch": 0.78, + "grad_norm": 0.255881055620786, + "learning_rate": 7.38332823703639e-05, + "loss": 1.223, + "step": 428 + }, + { + "epoch": 0.78, + "grad_norm": 0.24807364856677255, + "learning_rate": 7.378772712453021e-05, + "loss": 1.1985, + "step": 429 + }, + { + "epoch": 0.79, + "grad_norm": 0.25746257617764423, + "learning_rate": 7.37420183945258e-05, + "loss": 1.2502, + "step": 430 + }, + { + "epoch": 0.79, + "grad_norm": 0.28851991049982234, + "learning_rate": 7.369615638798722e-05, + "loss": 1.2535, + "step": 431 + }, + { + "epoch": 0.79, + "grad_norm": 0.24113389811604363, + "learning_rate": 7.365014131324725e-05, + "loss": 1.2227, + "step": 432 + }, + { + "epoch": 0.79, + "grad_norm": 0.2414465151257969, + "learning_rate": 7.360397337933405e-05, + "loss": 1.1884, + "step": 433 + }, + { + "epoch": 0.79, + "grad_norm": 0.2735463134699831, + "learning_rate": 7.355765279597011e-05, + "loss": 1.2756, + "step": 434 + }, + { + "epoch": 0.8, + "grad_norm": 0.2588437452987293, + "learning_rate": 7.351117977357139e-05, + "loss": 1.2108, + "step": 435 + }, + { + "epoch": 0.8, + "grad_norm": 0.26573294117796553, + "learning_rate": 7.346455452324629e-05, + "loss": 1.1821, + "step": 436 + }, + { + "epoch": 0.8, + "grad_norm": 0.2555476577827304, + "learning_rate": 7.341777725679473e-05, + "loss": 1.1937, + "step": 437 + }, + { + "epoch": 0.8, + "grad_norm": 0.2867704132108098, + "learning_rate": 7.337084818670716e-05, + "loss": 1.2272, + "step": 438 + }, + { + "epoch": 0.8, + "grad_norm": 0.27726678115981157, + "learning_rate": 7.332376752616367e-05, + "loss": 1.2331, + "step": 439 + }, + { + "epoch": 0.8, + "grad_norm": 0.26955338021079955, + "learning_rate": 7.32765354890329e-05, + "loss": 1.1731, + "step": 440 + }, + { + "epoch": 0.81, + "grad_norm": 0.25250321202536524, + "learning_rate": 7.322915228987116e-05, + "loss": 1.2653, + "step": 441 + }, + { + "epoch": 0.81, + "grad_norm": 0.24748844179765395, + "learning_rate": 7.318161814392143e-05, + "loss": 1.24, + "step": 442 + }, + { + "epoch": 0.81, + "grad_norm": 0.28177805247356325, + "learning_rate": 7.313393326711239e-05, + "loss": 1.185, + "step": 443 + }, + { + "epoch": 0.81, + "grad_norm": 0.24093242000396312, + "learning_rate": 7.30860978760574e-05, + "loss": 1.1994, + "step": 444 + }, + { + "epoch": 0.81, + "grad_norm": 0.26277803901457075, + "learning_rate": 7.30381121880536e-05, + "loss": 1.212, + "step": 445 + }, + { + "epoch": 0.82, + "grad_norm": 0.2506524258682433, + "learning_rate": 7.298997642108079e-05, + "loss": 1.2421, + "step": 446 + }, + { + "epoch": 0.82, + "grad_norm": 0.2840599700015824, + "learning_rate": 7.294169079380061e-05, + "loss": 1.1818, + "step": 447 + }, + { + "epoch": 0.82, + "grad_norm": 0.24892184038117549, + "learning_rate": 7.289325552555538e-05, + "loss": 1.1916, + "step": 448 + }, + { + "epoch": 0.82, + "grad_norm": 0.2700898428541357, + "learning_rate": 7.284467083636722e-05, + "loss": 1.2517, + "step": 449 + }, + { + "epoch": 0.82, + "grad_norm": 0.2617848546539419, + "learning_rate": 7.279593694693698e-05, + "loss": 1.2063, + "step": 450 + }, + { + "epoch": 0.82, + "grad_norm": 0.2698278585334131, + "learning_rate": 7.274705407864332e-05, + "loss": 1.194, + "step": 451 + }, + { + "epoch": 0.83, + "grad_norm": 0.23678313024953834, + "learning_rate": 7.26980224535416e-05, + "loss": 1.2349, + "step": 452 + }, + { + "epoch": 0.83, + "grad_norm": 0.24851875792002978, + "learning_rate": 7.264884229436293e-05, + "loss": 1.1758, + "step": 453 + }, + { + "epoch": 0.83, + "grad_norm": 0.24122080121681125, + "learning_rate": 7.259951382451318e-05, + "loss": 1.1962, + "step": 454 + }, + { + "epoch": 0.83, + "grad_norm": 0.22741322959884405, + "learning_rate": 7.25500372680719e-05, + "loss": 1.1702, + "step": 455 + }, + { + "epoch": 0.83, + "grad_norm": 0.2297475610861458, + "learning_rate": 7.250041284979137e-05, + "loss": 1.1466, + "step": 456 + }, + { + "epoch": 0.84, + "grad_norm": 0.3057605989721467, + "learning_rate": 7.245064079509553e-05, + "loss": 1.246, + "step": 457 + }, + { + "epoch": 0.84, + "grad_norm": 0.2719638501597136, + "learning_rate": 7.240072133007899e-05, + "loss": 1.2184, + "step": 458 + }, + { + "epoch": 0.84, + "grad_norm": 0.2436807816414479, + "learning_rate": 7.235065468150593e-05, + "loss": 1.2324, + "step": 459 + }, + { + "epoch": 0.84, + "grad_norm": 0.23436349430255515, + "learning_rate": 7.23004410768092e-05, + "loss": 1.1813, + "step": 460 + }, + { + "epoch": 0.84, + "grad_norm": 0.2398940990211377, + "learning_rate": 7.22500807440892e-05, + "loss": 1.1924, + "step": 461 + }, + { + "epoch": 0.84, + "grad_norm": 0.2605716625062531, + "learning_rate": 7.219957391211281e-05, + "loss": 1.182, + "step": 462 + }, + { + "epoch": 0.85, + "grad_norm": 0.260462524570941, + "learning_rate": 7.214892081031244e-05, + "loss": 1.2136, + "step": 463 + }, + { + "epoch": 0.85, + "grad_norm": 0.21979766512306334, + "learning_rate": 7.209812166878491e-05, + "loss": 1.2066, + "step": 464 + }, + { + "epoch": 0.85, + "grad_norm": 0.23324453647530663, + "learning_rate": 7.204717671829051e-05, + "loss": 1.1657, + "step": 465 + }, + { + "epoch": 0.85, + "grad_norm": 0.2529434935507481, + "learning_rate": 7.199608619025177e-05, + "loss": 1.2093, + "step": 466 + }, + { + "epoch": 0.85, + "grad_norm": 0.25371701891720116, + "learning_rate": 7.194485031675265e-05, + "loss": 1.2225, + "step": 467 + }, + { + "epoch": 0.86, + "grad_norm": 0.23272423066292103, + "learning_rate": 7.189346933053725e-05, + "loss": 1.1721, + "step": 468 + }, + { + "epoch": 0.86, + "grad_norm": 0.25122928735587546, + "learning_rate": 7.184194346500892e-05, + "loss": 1.2537, + "step": 469 + }, + { + "epoch": 0.86, + "grad_norm": 0.2159270875490409, + "learning_rate": 7.179027295422913e-05, + "loss": 1.197, + "step": 470 + }, + { + "epoch": 0.86, + "grad_norm": 0.2633111059076544, + "learning_rate": 7.173845803291636e-05, + "loss": 1.1721, + "step": 471 + }, + { + "epoch": 0.86, + "grad_norm": 0.30555936322098703, + "learning_rate": 7.168649893644517e-05, + "loss": 1.3011, + "step": 472 + }, + { + "epoch": 0.87, + "grad_norm": 0.23492670111453726, + "learning_rate": 7.163439590084502e-05, + "loss": 1.1601, + "step": 473 + }, + { + "epoch": 0.87, + "grad_norm": 0.26602734263721806, + "learning_rate": 7.158214916279923e-05, + "loss": 1.2808, + "step": 474 + }, + { + "epoch": 0.87, + "grad_norm": 0.3182695007856262, + "learning_rate": 7.152975895964386e-05, + "loss": 1.2967, + "step": 475 + }, + { + "epoch": 0.87, + "grad_norm": 0.2785021674736721, + "learning_rate": 7.147722552936673e-05, + "loss": 1.1789, + "step": 476 + }, + { + "epoch": 0.87, + "grad_norm": 0.279474303138652, + "learning_rate": 7.142454911060627e-05, + "loss": 1.2596, + "step": 477 + }, + { + "epoch": 0.87, + "grad_norm": 0.2556980144910755, + "learning_rate": 7.137172994265044e-05, + "loss": 1.2426, + "step": 478 + }, + { + "epoch": 0.88, + "grad_norm": 0.3311256331993533, + "learning_rate": 7.131876826543565e-05, + "loss": 1.2059, + "step": 479 + }, + { + "epoch": 0.88, + "grad_norm": 0.26467296197775253, + "learning_rate": 7.12656643195457e-05, + "loss": 1.2482, + "step": 480 + }, + { + "epoch": 0.88, + "grad_norm": 0.27444885274652553, + "learning_rate": 7.121241834621064e-05, + "loss": 1.2528, + "step": 481 + }, + { + "epoch": 0.88, + "grad_norm": 0.2572283861115396, + "learning_rate": 7.115903058730567e-05, + "loss": 1.1849, + "step": 482 + }, + { + "epoch": 0.88, + "grad_norm": 0.2677065778235683, + "learning_rate": 7.11055012853501e-05, + "loss": 1.2011, + "step": 483 + }, + { + "epoch": 0.89, + "grad_norm": 0.29470622036742816, + "learning_rate": 7.105183068350619e-05, + "loss": 1.2398, + "step": 484 + }, + { + "epoch": 0.89, + "grad_norm": 0.27609230248969197, + "learning_rate": 7.099801902557811e-05, + "loss": 1.2259, + "step": 485 + }, + { + "epoch": 0.89, + "grad_norm": 0.24248634168099284, + "learning_rate": 7.094406655601073e-05, + "loss": 1.2282, + "step": 486 + }, + { + "epoch": 0.89, + "grad_norm": 0.2765941767688746, + "learning_rate": 7.088997351988865e-05, + "loss": 1.2319, + "step": 487 + }, + { + "epoch": 0.89, + "grad_norm": 0.29347776909858947, + "learning_rate": 7.083574016293493e-05, + "loss": 1.1765, + "step": 488 + }, + { + "epoch": 0.89, + "grad_norm": 0.285370295424537, + "learning_rate": 7.078136673151008e-05, + "loss": 1.26, + "step": 489 + }, + { + "epoch": 0.9, + "grad_norm": 0.29408734903836536, + "learning_rate": 7.072685347261093e-05, + "loss": 1.226, + "step": 490 + }, + { + "epoch": 0.9, + "grad_norm": 0.27437470239205813, + "learning_rate": 7.067220063386947e-05, + "loss": 1.1976, + "step": 491 + }, + { + "epoch": 0.9, + "grad_norm": 0.2680770258777871, + "learning_rate": 7.061740846355176e-05, + "loss": 1.1915, + "step": 492 + }, + { + "epoch": 0.9, + "grad_norm": 0.27200362879502954, + "learning_rate": 7.056247721055678e-05, + "loss": 1.2002, + "step": 493 + }, + { + "epoch": 0.9, + "grad_norm": 0.2637811092577037, + "learning_rate": 7.050740712441528e-05, + "loss": 1.287, + "step": 494 + }, + { + "epoch": 0.91, + "grad_norm": 0.24657959209271266, + "learning_rate": 7.045219845528875e-05, + "loss": 1.2284, + "step": 495 + }, + { + "epoch": 0.91, + "grad_norm": 0.25311992110358666, + "learning_rate": 7.039685145396812e-05, + "loss": 1.1616, + "step": 496 + }, + { + "epoch": 0.91, + "grad_norm": 0.2564633694193358, + "learning_rate": 7.034136637187275e-05, + "loss": 1.2067, + "step": 497 + }, + { + "epoch": 0.91, + "grad_norm": 0.2446797651174144, + "learning_rate": 7.028574346104926e-05, + "loss": 1.2284, + "step": 498 + }, + { + "epoch": 0.91, + "grad_norm": 0.2592751463399255, + "learning_rate": 7.022998297417034e-05, + "loss": 1.2371, + "step": 499 + }, + { + "epoch": 0.91, + "grad_norm": 0.2500713943206808, + "learning_rate": 7.017408516453365e-05, + "loss": 1.1061, + "step": 500 + }, + { + "epoch": 0.92, + "grad_norm": 0.2812266276040743, + "learning_rate": 7.011805028606064e-05, + "loss": 1.1949, + "step": 501 + }, + { + "epoch": 0.92, + "grad_norm": 0.298829667668083, + "learning_rate": 7.006187859329544e-05, + "loss": 1.2313, + "step": 502 + }, + { + "epoch": 0.92, + "grad_norm": 0.26518768159745104, + "learning_rate": 7.000557034140361e-05, + "loss": 1.2246, + "step": 503 + }, + { + "epoch": 0.92, + "grad_norm": 0.3037280360760458, + "learning_rate": 6.994912578617113e-05, + "loss": 1.1617, + "step": 504 + }, + { + "epoch": 0.92, + "grad_norm": 0.2726903109255714, + "learning_rate": 6.989254518400309e-05, + "loss": 1.2415, + "step": 505 + }, + { + "epoch": 0.93, + "grad_norm": 0.25568082003046966, + "learning_rate": 6.98358287919226e-05, + "loss": 1.1817, + "step": 506 + }, + { + "epoch": 0.93, + "grad_norm": 0.25633294893705044, + "learning_rate": 6.97789768675696e-05, + "loss": 1.2149, + "step": 507 + }, + { + "epoch": 0.93, + "grad_norm": 0.28291439435087123, + "learning_rate": 6.972198966919972e-05, + "loss": 1.1578, + "step": 508 + }, + { + "epoch": 0.93, + "grad_norm": 0.27195184756655516, + "learning_rate": 6.966486745568308e-05, + "loss": 1.2355, + "step": 509 + }, + { + "epoch": 0.93, + "grad_norm": 0.239159568376005, + "learning_rate": 6.960761048650312e-05, + "loss": 1.1688, + "step": 510 + }, + { + "epoch": 0.93, + "grad_norm": 0.22961475425949177, + "learning_rate": 6.955021902175543e-05, + "loss": 1.2094, + "step": 511 + }, + { + "epoch": 0.94, + "grad_norm": 0.27443773600741117, + "learning_rate": 6.949269332214651e-05, + "loss": 1.2559, + "step": 512 + }, + { + "epoch": 0.94, + "grad_norm": 0.26230551832002097, + "learning_rate": 6.94350336489927e-05, + "loss": 1.2121, + "step": 513 + }, + { + "epoch": 0.94, + "grad_norm": 0.2716742985303849, + "learning_rate": 6.937724026421892e-05, + "loss": 1.2444, + "step": 514 + }, + { + "epoch": 0.94, + "grad_norm": 0.2537850139439542, + "learning_rate": 6.931931343035742e-05, + "loss": 1.1327, + "step": 515 + }, + { + "epoch": 0.94, + "grad_norm": 0.28599587967496826, + "learning_rate": 6.926125341054676e-05, + "loss": 1.2236, + "step": 516 + }, + { + "epoch": 0.95, + "grad_norm": 0.26780654378470103, + "learning_rate": 6.920306046853043e-05, + "loss": 1.2295, + "step": 517 + }, + { + "epoch": 0.95, + "grad_norm": 0.23606296888412015, + "learning_rate": 6.914473486865577e-05, + "loss": 1.1543, + "step": 518 + }, + { + "epoch": 0.95, + "grad_norm": 0.34976881174240837, + "learning_rate": 6.90862768758727e-05, + "loss": 1.2067, + "step": 519 + }, + { + "epoch": 0.95, + "grad_norm": 0.2481257873494882, + "learning_rate": 6.902768675573258e-05, + "loss": 1.2188, + "step": 520 + }, + { + "epoch": 0.95, + "grad_norm": 0.2996395778117021, + "learning_rate": 6.896896477438699e-05, + "loss": 1.2326, + "step": 521 + }, + { + "epoch": 0.95, + "grad_norm": 0.8839768816333193, + "learning_rate": 6.891011119858643e-05, + "loss": 1.2435, + "step": 522 + }, + { + "epoch": 0.96, + "grad_norm": 0.2851882482058998, + "learning_rate": 6.885112629567927e-05, + "loss": 1.2644, + "step": 523 + }, + { + "epoch": 0.96, + "grad_norm": 0.2813663482913699, + "learning_rate": 6.879201033361035e-05, + "loss": 1.2309, + "step": 524 + }, + { + "epoch": 0.96, + "grad_norm": 0.3257551560135454, + "learning_rate": 6.873276358091996e-05, + "loss": 1.2755, + "step": 525 + }, + { + "epoch": 0.96, + "grad_norm": 0.28930479952494365, + "learning_rate": 6.867338630674247e-05, + "loss": 1.1962, + "step": 526 + }, + { + "epoch": 0.96, + "grad_norm": 0.3077462996938649, + "learning_rate": 6.861387878080511e-05, + "loss": 1.2402, + "step": 527 + }, + { + "epoch": 0.97, + "grad_norm": 0.2848900193452761, + "learning_rate": 6.855424127342688e-05, + "loss": 1.2748, + "step": 528 + }, + { + "epoch": 0.97, + "grad_norm": 0.4765938812802202, + "learning_rate": 6.849447405551718e-05, + "loss": 1.2226, + "step": 529 + }, + { + "epoch": 0.97, + "grad_norm": 0.53184473292579, + "learning_rate": 6.843457739857467e-05, + "loss": 1.2347, + "step": 530 + }, + { + "epoch": 0.97, + "grad_norm": 0.6416239346492343, + "learning_rate": 6.837455157468596e-05, + "loss": 1.2429, + "step": 531 + }, + { + "epoch": 0.97, + "grad_norm": 0.3188092712502773, + "learning_rate": 6.831439685652442e-05, + "loss": 1.216, + "step": 532 + }, + { + "epoch": 0.97, + "grad_norm": 0.3527495731006385, + "learning_rate": 6.825411351734895e-05, + "loss": 1.1682, + "step": 533 + }, + { + "epoch": 0.98, + "grad_norm": 0.29603753744741856, + "learning_rate": 6.819370183100274e-05, + "loss": 1.1434, + "step": 534 + }, + { + "epoch": 0.98, + "grad_norm": 0.5252450389976622, + "learning_rate": 6.813316207191198e-05, + "loss": 1.1943, + "step": 535 + }, + { + "epoch": 0.98, + "grad_norm": 0.32999419558659937, + "learning_rate": 6.807249451508466e-05, + "loss": 1.192, + "step": 536 + }, + { + "epoch": 0.98, + "grad_norm": 0.3650175469778724, + "learning_rate": 6.801169943610929e-05, + "loss": 1.2141, + "step": 537 + }, + { + "epoch": 0.98, + "grad_norm": 1.0643532150783557, + "learning_rate": 6.795077711115368e-05, + "loss": 1.2253, + "step": 538 + }, + { + "epoch": 0.99, + "grad_norm": 0.5041310609130145, + "learning_rate": 6.788972781696363e-05, + "loss": 1.278, + "step": 539 + }, + { + "epoch": 0.99, + "grad_norm": 0.5123058164360991, + "learning_rate": 6.782855183086177e-05, + "loss": 1.2231, + "step": 540 + }, + { + "epoch": 0.99, + "grad_norm": 0.3533015702394419, + "learning_rate": 6.776724943074619e-05, + "loss": 1.2072, + "step": 541 + }, + { + "epoch": 0.99, + "grad_norm": 0.30253964625417207, + "learning_rate": 6.770582089508927e-05, + "loss": 1.1382, + "step": 542 + }, + { + "epoch": 0.99, + "grad_norm": 0.348991618828202, + "learning_rate": 6.764426650293633e-05, + "loss": 1.2079, + "step": 543 + }, + { + "epoch": 0.99, + "grad_norm": 0.46017440578788743, + "learning_rate": 6.758258653390444e-05, + "loss": 1.1813, + "step": 544 + }, + { + "epoch": 1.0, + "grad_norm": 0.31962101755594885, + "learning_rate": 6.75207812681811e-05, + "loss": 1.1339, + "step": 545 + }, + { + "epoch": 1.0, + "grad_norm": 0.37092024548285923, + "learning_rate": 6.745885098652298e-05, + "loss": 1.2591, + "step": 546 + }, + { + "epoch": 1.0, + "grad_norm": 0.32347106450715835, + "learning_rate": 6.739679597025466e-05, + "loss": 1.2017, + "step": 547 + }, + { + "epoch": 1.0, + "grad_norm": 0.39250187112342494, + "learning_rate": 6.733461650126733e-05, + "loss": 1.0933, + "step": 548 + }, + { + "epoch": 1.0, + "grad_norm": 0.473522452217324, + "learning_rate": 6.727231286201752e-05, + "loss": 1.1124, + "step": 549 + }, + { + "epoch": 1.01, + "grad_norm": 0.4809062179622052, + "learning_rate": 6.720988533552582e-05, + "loss": 1.1585, + "step": 550 + }, + { + "epoch": 1.01, + "grad_norm": 0.3529662801059162, + "learning_rate": 6.714733420537559e-05, + "loss": 1.0501, + "step": 551 + }, + { + "epoch": 1.01, + "grad_norm": 0.5958247214391118, + "learning_rate": 6.708465975571168e-05, + "loss": 1.1086, + "step": 552 + }, + { + "epoch": 1.01, + "grad_norm": 0.5341364205022454, + "learning_rate": 6.70218622712391e-05, + "loss": 1.0518, + "step": 553 + }, + { + "epoch": 1.01, + "grad_norm": 0.3601805724462006, + "learning_rate": 6.695894203722181e-05, + "loss": 1.1779, + "step": 554 + }, + { + "epoch": 1.02, + "grad_norm": 0.43410190338280613, + "learning_rate": 6.68958993394813e-05, + "loss": 1.093, + "step": 555 + }, + { + "epoch": 1.02, + "grad_norm": 0.46217742572873594, + "learning_rate": 6.683273446439546e-05, + "loss": 1.0117, + "step": 556 + }, + { + "epoch": 1.02, + "grad_norm": 0.8591682373623357, + "learning_rate": 6.676944769889708e-05, + "loss": 1.1002, + "step": 557 + }, + { + "epoch": 1.02, + "grad_norm": 0.7383229487622726, + "learning_rate": 6.670603933047272e-05, + "loss": 1.0779, + "step": 558 + }, + { + "epoch": 1.02, + "grad_norm": 0.5965305891207813, + "learning_rate": 6.664250964716131e-05, + "loss": 1.0889, + "step": 559 + }, + { + "epoch": 1.02, + "grad_norm": 0.6030858606684543, + "learning_rate": 6.657885893755288e-05, + "loss": 1.0982, + "step": 560 + }, + { + "epoch": 1.03, + "grad_norm": 0.4644510682398409, + "learning_rate": 6.65150874907872e-05, + "loss": 1.1004, + "step": 561 + }, + { + "epoch": 1.03, + "grad_norm": 0.43943285132452564, + "learning_rate": 6.645119559655254e-05, + "loss": 1.0536, + "step": 562 + }, + { + "epoch": 1.03, + "grad_norm": 0.4456395978600012, + "learning_rate": 6.638718354508427e-05, + "loss": 1.0733, + "step": 563 + }, + { + "epoch": 1.03, + "grad_norm": 0.3303824433217466, + "learning_rate": 6.632305162716365e-05, + "loss": 1.0552, + "step": 564 + }, + { + "epoch": 1.03, + "grad_norm": 0.3617704823170143, + "learning_rate": 6.62588001341164e-05, + "loss": 1.1092, + "step": 565 + }, + { + "epoch": 1.04, + "grad_norm": 0.4465013349903427, + "learning_rate": 6.619442935781141e-05, + "loss": 1.0781, + "step": 566 + }, + { + "epoch": 1.04, + "grad_norm": 0.48516780613791277, + "learning_rate": 6.612993959065947e-05, + "loss": 1.0686, + "step": 567 + }, + { + "epoch": 1.04, + "grad_norm": 0.38867820318536633, + "learning_rate": 6.606533112561186e-05, + "loss": 1.1215, + "step": 568 + }, + { + "epoch": 1.04, + "grad_norm": 0.38566119820378336, + "learning_rate": 6.600060425615907e-05, + "loss": 1.1213, + "step": 569 + }, + { + "epoch": 1.04, + "grad_norm": 0.35534855445058544, + "learning_rate": 6.593575927632947e-05, + "loss": 1.0955, + "step": 570 + }, + { + "epoch": 1.04, + "grad_norm": 0.38124406233349717, + "learning_rate": 6.587079648068795e-05, + "loss": 1.0659, + "step": 571 + }, + { + "epoch": 1.05, + "grad_norm": 0.454750160923548, + "learning_rate": 6.580571616433457e-05, + "loss": 1.1149, + "step": 572 + }, + { + "epoch": 1.05, + "grad_norm": 0.35353190088025255, + "learning_rate": 6.574051862290325e-05, + "loss": 1.0388, + "step": 573 + }, + { + "epoch": 1.05, + "grad_norm": 0.3249395594793626, + "learning_rate": 6.567520415256045e-05, + "loss": 1.0784, + "step": 574 + }, + { + "epoch": 1.05, + "grad_norm": 0.40078898818247227, + "learning_rate": 6.560977305000375e-05, + "loss": 1.0859, + "step": 575 + }, + { + "epoch": 1.05, + "grad_norm": 0.4115264795060035, + "learning_rate": 6.554422561246054e-05, + "loss": 1.1828, + "step": 576 + }, + { + "epoch": 1.06, + "grad_norm": 0.30090229228069215, + "learning_rate": 6.54785621376867e-05, + "loss": 1.0901, + "step": 577 + }, + { + "epoch": 1.06, + "grad_norm": 0.28827860350299206, + "learning_rate": 6.541278292396523e-05, + "loss": 1.0277, + "step": 578 + }, + { + "epoch": 1.06, + "grad_norm": 0.34690404488996757, + "learning_rate": 6.534688827010484e-05, + "loss": 1.048, + "step": 579 + }, + { + "epoch": 1.06, + "grad_norm": 0.29943113556644785, + "learning_rate": 6.528087847543867e-05, + "loss": 1.0646, + "step": 580 + }, + { + "epoch": 1.06, + "grad_norm": 0.37318202575874415, + "learning_rate": 6.521475383982291e-05, + "loss": 1.1091, + "step": 581 + }, + { + "epoch": 1.06, + "grad_norm": 0.3049663659203959, + "learning_rate": 6.51485146636354e-05, + "loss": 1.0552, + "step": 582 + }, + { + "epoch": 1.07, + "grad_norm": 0.3342407867509692, + "learning_rate": 6.508216124777431e-05, + "loss": 1.2227, + "step": 583 + }, + { + "epoch": 1.07, + "grad_norm": 0.3348396047855952, + "learning_rate": 6.501569389365674e-05, + "loss": 1.0861, + "step": 584 + }, + { + "epoch": 1.07, + "grad_norm": 0.30951429367513383, + "learning_rate": 6.494911290321737e-05, + "loss": 1.0461, + "step": 585 + }, + { + "epoch": 1.07, + "grad_norm": 0.33898401361064606, + "learning_rate": 6.488241857890711e-05, + "loss": 1.0854, + "step": 586 + }, + { + "epoch": 1.07, + "grad_norm": 0.4901462068263497, + "learning_rate": 6.481561122369164e-05, + "loss": 1.1012, + "step": 587 + }, + { + "epoch": 1.08, + "grad_norm": 0.3179574879809652, + "learning_rate": 6.474869114105018e-05, + "loss": 1.0451, + "step": 588 + }, + { + "epoch": 1.08, + "grad_norm": 0.32159328915060714, + "learning_rate": 6.468165863497395e-05, + "loss": 1.0458, + "step": 589 + }, + { + "epoch": 1.08, + "grad_norm": 0.36462235008537297, + "learning_rate": 6.461451400996491e-05, + "loss": 1.1247, + "step": 590 + }, + { + "epoch": 1.08, + "grad_norm": 0.5373862753611778, + "learning_rate": 6.454725757103432e-05, + "loss": 1.0542, + "step": 591 + }, + { + "epoch": 1.08, + "grad_norm": 0.3160409270291303, + "learning_rate": 6.447988962370133e-05, + "loss": 1.0829, + "step": 592 + }, + { + "epoch": 1.08, + "grad_norm": 0.390452102978435, + "learning_rate": 6.441241047399169e-05, + "loss": 1.192, + "step": 593 + }, + { + "epoch": 1.09, + "grad_norm": 0.3802122712014928, + "learning_rate": 6.434482042843627e-05, + "loss": 1.1153, + "step": 594 + }, + { + "epoch": 1.09, + "grad_norm": 0.4081584328242501, + "learning_rate": 6.427711979406966e-05, + "loss": 1.1635, + "step": 595 + }, + { + "epoch": 1.09, + "grad_norm": 0.3791962989638633, + "learning_rate": 6.420930887842889e-05, + "loss": 1.1581, + "step": 596 + }, + { + "epoch": 1.09, + "grad_norm": 0.33239440056484193, + "learning_rate": 6.414138798955189e-05, + "loss": 1.0926, + "step": 597 + }, + { + "epoch": 1.09, + "grad_norm": 0.3279881540815014, + "learning_rate": 6.407335743597616e-05, + "loss": 1.1386, + "step": 598 + }, + { + "epoch": 1.1, + "grad_norm": 0.30309644763750837, + "learning_rate": 6.40052175267374e-05, + "loss": 1.0523, + "step": 599 + }, + { + "epoch": 1.1, + "grad_norm": 0.3349097308403333, + "learning_rate": 6.393696857136801e-05, + "loss": 1.0815, + "step": 600 + }, + { + "epoch": 1.1, + "grad_norm": 0.3288227593556618, + "learning_rate": 6.386861087989581e-05, + "loss": 1.015, + "step": 601 + }, + { + "epoch": 1.1, + "grad_norm": 0.36685586740843157, + "learning_rate": 6.380014476284255e-05, + "loss": 1.1232, + "step": 602 + }, + { + "epoch": 1.1, + "grad_norm": 0.3620977714204643, + "learning_rate": 6.373157053122243e-05, + "loss": 1.1138, + "step": 603 + }, + { + "epoch": 1.1, + "grad_norm": 0.3130587018197183, + "learning_rate": 6.366288849654091e-05, + "loss": 1.1255, + "step": 604 + }, + { + "epoch": 1.11, + "grad_norm": 0.3602737087072766, + "learning_rate": 6.359409897079303e-05, + "loss": 1.0282, + "step": 605 + }, + { + "epoch": 1.11, + "grad_norm": 0.31168852571991945, + "learning_rate": 6.352520226646222e-05, + "loss": 1.0779, + "step": 606 + }, + { + "epoch": 1.11, + "grad_norm": 0.3516045580189353, + "learning_rate": 6.345619869651871e-05, + "loss": 1.1028, + "step": 607 + }, + { + "epoch": 1.11, + "grad_norm": 0.3231857927563657, + "learning_rate": 6.33870885744182e-05, + "loss": 1.1202, + "step": 608 + }, + { + "epoch": 1.11, + "grad_norm": 0.30205205129701157, + "learning_rate": 6.331787221410041e-05, + "loss": 1.1369, + "step": 609 + }, + { + "epoch": 1.12, + "grad_norm": 0.3198359813888166, + "learning_rate": 6.32485499299877e-05, + "loss": 1.1763, + "step": 610 + }, + { + "epoch": 1.12, + "grad_norm": 0.3128641370321787, + "learning_rate": 6.31791220369835e-05, + "loss": 1.0223, + "step": 611 + }, + { + "epoch": 1.12, + "grad_norm": 0.2989105616213649, + "learning_rate": 6.31095888504711e-05, + "loss": 1.0358, + "step": 612 + }, + { + "epoch": 1.12, + "grad_norm": 0.3103537906853337, + "learning_rate": 6.303995068631203e-05, + "loss": 1.1261, + "step": 613 + }, + { + "epoch": 1.12, + "grad_norm": 0.28598715532508207, + "learning_rate": 6.297020786084467e-05, + "loss": 1.0629, + "step": 614 + }, + { + "epoch": 1.12, + "grad_norm": 0.29809789918093255, + "learning_rate": 6.290036069088288e-05, + "loss": 1.035, + "step": 615 + }, + { + "epoch": 1.13, + "grad_norm": 0.33765270252261453, + "learning_rate": 6.283040949371451e-05, + "loss": 1.1221, + "step": 616 + }, + { + "epoch": 1.13, + "grad_norm": 0.3424617501293415, + "learning_rate": 6.276035458709993e-05, + "loss": 1.155, + "step": 617 + }, + { + "epoch": 1.13, + "grad_norm": 0.3799189737987811, + "learning_rate": 6.269019628927067e-05, + "loss": 1.0701, + "step": 618 + }, + { + "epoch": 1.13, + "grad_norm": 0.3358898935253196, + "learning_rate": 6.261993491892791e-05, + "loss": 1.1649, + "step": 619 + }, + { + "epoch": 1.13, + "grad_norm": 0.31569979424117356, + "learning_rate": 6.254957079524099e-05, + "loss": 1.0633, + "step": 620 + }, + { + "epoch": 1.14, + "grad_norm": 0.3002168156888237, + "learning_rate": 6.247910423784609e-05, + "loss": 1.0846, + "step": 621 + }, + { + "epoch": 1.14, + "grad_norm": 0.3097238823450595, + "learning_rate": 6.24085355668447e-05, + "loss": 1.0808, + "step": 622 + }, + { + "epoch": 1.14, + "grad_norm": 0.3120312761417578, + "learning_rate": 6.233786510280212e-05, + "loss": 1.0142, + "step": 623 + }, + { + "epoch": 1.14, + "grad_norm": 0.3335343015064923, + "learning_rate": 6.22670931667461e-05, + "loss": 1.0674, + "step": 624 + }, + { + "epoch": 1.14, + "grad_norm": 0.3234062304634526, + "learning_rate": 6.219622008016533e-05, + "loss": 1.0981, + "step": 625 + }, + { + "epoch": 1.14, + "grad_norm": 0.32152678786547273, + "learning_rate": 6.212524616500798e-05, + "loss": 1.0244, + "step": 626 + }, + { + "epoch": 1.15, + "grad_norm": 0.39031977608147594, + "learning_rate": 6.205417174368023e-05, + "loss": 1.1205, + "step": 627 + }, + { + "epoch": 1.15, + "grad_norm": 0.3806189090017157, + "learning_rate": 6.198299713904485e-05, + "loss": 1.1134, + "step": 628 + }, + { + "epoch": 1.15, + "grad_norm": 0.2978349276971668, + "learning_rate": 6.191172267441967e-05, + "loss": 1.0088, + "step": 629 + }, + { + "epoch": 1.15, + "grad_norm": 0.3190354077382501, + "learning_rate": 6.184034867357617e-05, + "loss": 1.108, + "step": 630 + }, + { + "epoch": 1.15, + "grad_norm": 0.32633048665038994, + "learning_rate": 6.176887546073797e-05, + "loss": 1.0825, + "step": 631 + }, + { + "epoch": 1.16, + "grad_norm": 0.3428026413020903, + "learning_rate": 6.169730336057939e-05, + "loss": 1.0765, + "step": 632 + }, + { + "epoch": 1.16, + "grad_norm": 0.3475737151929015, + "learning_rate": 6.162563269822391e-05, + "loss": 1.0693, + "step": 633 + }, + { + "epoch": 1.16, + "grad_norm": 0.3870252154591392, + "learning_rate": 6.15538637992428e-05, + "loss": 1.1081, + "step": 634 + }, + { + "epoch": 1.16, + "grad_norm": 0.33597355193652834, + "learning_rate": 6.148199698965352e-05, + "loss": 1.0893, + "step": 635 + }, + { + "epoch": 1.16, + "grad_norm": 0.30805894179787247, + "learning_rate": 6.141003259591834e-05, + "loss": 1.0995, + "step": 636 + }, + { + "epoch": 1.17, + "grad_norm": 0.3025073882734066, + "learning_rate": 6.133797094494281e-05, + "loss": 1.0388, + "step": 637 + }, + { + "epoch": 1.17, + "grad_norm": 0.3524395196391662, + "learning_rate": 6.126581236407429e-05, + "loss": 1.1196, + "step": 638 + }, + { + "epoch": 1.17, + "grad_norm": 0.3377646188130345, + "learning_rate": 6.119355718110039e-05, + "loss": 1.0382, + "step": 639 + }, + { + "epoch": 1.17, + "grad_norm": 0.35508400659785483, + "learning_rate": 6.112120572424763e-05, + "loss": 1.1402, + "step": 640 + }, + { + "epoch": 1.17, + "grad_norm": 0.3454418793700457, + "learning_rate": 6.104875832217982e-05, + "loss": 1.1032, + "step": 641 + }, + { + "epoch": 1.17, + "grad_norm": 0.32629806837059866, + "learning_rate": 6.097621530399661e-05, + "loss": 1.0959, + "step": 642 + }, + { + "epoch": 1.18, + "grad_norm": 0.3329536837751315, + "learning_rate": 6.090357699923202e-05, + "loss": 1.0467, + "step": 643 + }, + { + "epoch": 1.18, + "grad_norm": 0.32302233828349475, + "learning_rate": 6.083084373785287e-05, + "loss": 1.0858, + "step": 644 + }, + { + "epoch": 1.18, + "grad_norm": 0.3310358826507611, + "learning_rate": 6.075801585025739e-05, + "loss": 1.0715, + "step": 645 + }, + { + "epoch": 1.18, + "grad_norm": 0.319322035854079, + "learning_rate": 6.068509366727362e-05, + "loss": 1.177, + "step": 646 + }, + { + "epoch": 1.18, + "grad_norm": 0.3065230667302707, + "learning_rate": 6.061207752015797e-05, + "loss": 1.0649, + "step": 647 + }, + { + "epoch": 1.19, + "grad_norm": 0.29926795565748227, + "learning_rate": 6.053896774059368e-05, + "loss": 1.1325, + "step": 648 + }, + { + "epoch": 1.19, + "grad_norm": 0.3556069634279046, + "learning_rate": 6.046576466068931e-05, + "loss": 1.1366, + "step": 649 + }, + { + "epoch": 1.19, + "grad_norm": 0.3189191131461966, + "learning_rate": 6.039246861297727e-05, + "loss": 1.0693, + "step": 650 + }, + { + "epoch": 1.19, + "grad_norm": 0.3347197156648834, + "learning_rate": 6.031907993041227e-05, + "loss": 1.1009, + "step": 651 + }, + { + "epoch": 1.19, + "grad_norm": 0.32274156348185445, + "learning_rate": 6.0245598946369826e-05, + "loss": 1.1675, + "step": 652 + }, + { + "epoch": 1.19, + "grad_norm": 0.35534089035455224, + "learning_rate": 6.017202599464476e-05, + "loss": 1.1723, + "step": 653 + }, + { + "epoch": 1.2, + "grad_norm": 0.3106026578570133, + "learning_rate": 6.009836140944965e-05, + "loss": 1.0954, + "step": 654 + }, + { + "epoch": 1.2, + "grad_norm": 0.3309144454564729, + "learning_rate": 6.002460552541331e-05, + "loss": 1.0209, + "step": 655 + }, + { + "epoch": 1.2, + "grad_norm": 0.3023619281400003, + "learning_rate": 5.9950758677579345e-05, + "loss": 1.0363, + "step": 656 + }, + { + "epoch": 1.2, + "grad_norm": 0.3311182880219704, + "learning_rate": 5.987682120140451e-05, + "loss": 1.0515, + "step": 657 + }, + { + "epoch": 1.2, + "grad_norm": 0.33396486010030413, + "learning_rate": 5.980279343275729e-05, + "loss": 1.1251, + "step": 658 + }, + { + "epoch": 1.21, + "grad_norm": 0.3465764556678002, + "learning_rate": 5.97286757079163e-05, + "loss": 1.165, + "step": 659 + }, + { + "epoch": 1.21, + "grad_norm": 0.304193441363374, + "learning_rate": 5.965446836356882e-05, + "loss": 1.0228, + "step": 660 + }, + { + "epoch": 1.21, + "grad_norm": 0.3415149030413082, + "learning_rate": 5.9580171736809224e-05, + "loss": 1.0742, + "step": 661 + }, + { + "epoch": 1.21, + "grad_norm": 0.33138658321132064, + "learning_rate": 5.950578616513746e-05, + "loss": 1.0843, + "step": 662 + }, + { + "epoch": 1.21, + "grad_norm": 0.30774403421162994, + "learning_rate": 5.943131198645752e-05, + "loss": 1.065, + "step": 663 + }, + { + "epoch": 1.21, + "grad_norm": 0.3428877492183819, + "learning_rate": 5.9356749539075885e-05, + "loss": 1.1101, + "step": 664 + }, + { + "epoch": 1.22, + "grad_norm": 0.3621290546130101, + "learning_rate": 5.928209916170003e-05, + "loss": 1.1372, + "step": 665 + }, + { + "epoch": 1.22, + "grad_norm": 0.3482375945469884, + "learning_rate": 5.9207361193436865e-05, + "loss": 1.132, + "step": 666 + }, + { + "epoch": 1.22, + "grad_norm": 0.31754384974068384, + "learning_rate": 5.9132535973791156e-05, + "loss": 1.148, + "step": 667 + }, + { + "epoch": 1.22, + "grad_norm": 0.36003834782050365, + "learning_rate": 5.9057623842664044e-05, + "loss": 1.1099, + "step": 668 + }, + { + "epoch": 1.22, + "grad_norm": 0.2963701622969662, + "learning_rate": 5.8982625140351464e-05, + "loss": 1.0755, + "step": 669 + }, + { + "epoch": 1.23, + "grad_norm": 0.32579569606066516, + "learning_rate": 5.8907540207542616e-05, + "loss": 1.0809, + "step": 670 + }, + { + "epoch": 1.23, + "grad_norm": 0.4247563451753457, + "learning_rate": 5.8832369385318416e-05, + "loss": 1.097, + "step": 671 + }, + { + "epoch": 1.23, + "grad_norm": 0.33076932102169776, + "learning_rate": 5.875711301514992e-05, + "loss": 1.1078, + "step": 672 + }, + { + "epoch": 1.23, + "grad_norm": 0.3609238032332309, + "learning_rate": 5.8681771438896815e-05, + "loss": 1.1031, + "step": 673 + }, + { + "epoch": 1.23, + "grad_norm": 0.325159585649425, + "learning_rate": 5.860634499880583e-05, + "loss": 1.0707, + "step": 674 + }, + { + "epoch": 1.23, + "grad_norm": 0.4620687271068983, + "learning_rate": 5.853083403750922e-05, + "loss": 1.1017, + "step": 675 + }, + { + "epoch": 1.24, + "grad_norm": 0.33485279064365936, + "learning_rate": 5.845523889802316e-05, + "loss": 1.0989, + "step": 676 + }, + { + "epoch": 1.24, + "grad_norm": 0.30952573170841513, + "learning_rate": 5.8379559923746214e-05, + "loss": 1.0393, + "step": 677 + }, + { + "epoch": 1.24, + "grad_norm": 0.33498605810588283, + "learning_rate": 5.830379745845781e-05, + "loss": 1.1259, + "step": 678 + }, + { + "epoch": 1.24, + "grad_norm": 0.35771921163037307, + "learning_rate": 5.822795184631659e-05, + "loss": 1.0815, + "step": 679 + }, + { + "epoch": 1.24, + "grad_norm": 0.3329650192347647, + "learning_rate": 5.815202343185894e-05, + "loss": 1.1344, + "step": 680 + }, + { + "epoch": 1.25, + "grad_norm": 0.3356634465845771, + "learning_rate": 5.807601255999736e-05, + "loss": 1.1297, + "step": 681 + }, + { + "epoch": 1.25, + "grad_norm": 0.3289442034151235, + "learning_rate": 5.7999919576018934e-05, + "loss": 1.022, + "step": 682 + }, + { + "epoch": 1.25, + "grad_norm": 0.3207007334784113, + "learning_rate": 5.7923744825583745e-05, + "loss": 1.0571, + "step": 683 + }, + { + "epoch": 1.25, + "grad_norm": 0.3582460325329284, + "learning_rate": 5.7847488654723304e-05, + "loss": 1.0778, + "step": 684 + }, + { + "epoch": 1.25, + "grad_norm": 0.3563317666176927, + "learning_rate": 5.777115140983899e-05, + "loss": 1.1003, + "step": 685 + }, + { + "epoch": 1.25, + "grad_norm": 3.4694912945702105, + "learning_rate": 5.769473343770047e-05, + "loss": 1.121, + "step": 686 + }, + { + "epoch": 1.26, + "grad_norm": 0.43002349520483113, + "learning_rate": 5.761823508544411e-05, + "loss": 1.0765, + "step": 687 + }, + { + "epoch": 1.26, + "grad_norm": 0.39467783104839754, + "learning_rate": 5.754165670057142e-05, + "loss": 1.0788, + "step": 688 + }, + { + "epoch": 1.26, + "grad_norm": 0.39629029674867916, + "learning_rate": 5.7464998630947464e-05, + "loss": 1.0812, + "step": 689 + }, + { + "epoch": 1.26, + "grad_norm": 0.3880152093965208, + "learning_rate": 5.738826122479929e-05, + "loss": 1.1228, + "step": 690 + }, + { + "epoch": 1.26, + "grad_norm": 0.3777874121959188, + "learning_rate": 5.7311444830714324e-05, + "loss": 1.0907, + "step": 691 + }, + { + "epoch": 1.27, + "grad_norm": 0.38004041653523696, + "learning_rate": 5.723454979763882e-05, + "loss": 1.1263, + "step": 692 + }, + { + "epoch": 1.27, + "grad_norm": 0.37049672627797636, + "learning_rate": 5.7157576474876246e-05, + "loss": 1.1438, + "step": 693 + }, + { + "epoch": 1.27, + "grad_norm": 0.32973606103437614, + "learning_rate": 5.7080525212085725e-05, + "loss": 1.0553, + "step": 694 + }, + { + "epoch": 1.27, + "grad_norm": 0.31674639252070325, + "learning_rate": 5.700339635928038e-05, + "loss": 1.06, + "step": 695 + }, + { + "epoch": 1.27, + "grad_norm": 0.32282199426553837, + "learning_rate": 5.692619026682588e-05, + "loss": 1.0841, + "step": 696 + }, + { + "epoch": 1.27, + "grad_norm": 0.4810882958061859, + "learning_rate": 5.684890728543869e-05, + "loss": 1.0803, + "step": 697 + }, + { + "epoch": 1.28, + "grad_norm": 0.3995638550178378, + "learning_rate": 5.6771547766184566e-05, + "loss": 1.1187, + "step": 698 + }, + { + "epoch": 1.28, + "grad_norm": 0.35264932960583484, + "learning_rate": 5.669411206047699e-05, + "loss": 1.0641, + "step": 699 + }, + { + "epoch": 1.28, + "grad_norm": 0.35240640524733, + "learning_rate": 5.661660052007547e-05, + "loss": 1.076, + "step": 700 + }, + { + "epoch": 1.28, + "grad_norm": 0.3540694609860389, + "learning_rate": 5.653901349708401e-05, + "loss": 1.1369, + "step": 701 + }, + { + "epoch": 1.28, + "grad_norm": 0.3196055112925304, + "learning_rate": 5.646135134394955e-05, + "loss": 1.0677, + "step": 702 + }, + { + "epoch": 1.29, + "grad_norm": 0.4214141007955914, + "learning_rate": 5.6383614413460266e-05, + "loss": 1.1139, + "step": 703 + }, + { + "epoch": 1.29, + "grad_norm": 0.3625611311798579, + "learning_rate": 5.630580305874402e-05, + "loss": 1.1845, + "step": 704 + }, + { + "epoch": 1.29, + "grad_norm": 0.3425208672181188, + "learning_rate": 5.62279176332668e-05, + "loss": 1.174, + "step": 705 + }, + { + "epoch": 1.29, + "grad_norm": 0.3108419862818321, + "learning_rate": 5.6149958490830996e-05, + "loss": 1.0331, + "step": 706 + }, + { + "epoch": 1.29, + "grad_norm": 0.3274644181571904, + "learning_rate": 5.607192598557394e-05, + "loss": 1.0664, + "step": 707 + }, + { + "epoch": 1.29, + "grad_norm": 0.346218197215145, + "learning_rate": 5.599382047196617e-05, + "loss": 1.2088, + "step": 708 + }, + { + "epoch": 1.3, + "grad_norm": 0.328497632267458, + "learning_rate": 5.591564230480989e-05, + "loss": 1.0287, + "step": 709 + }, + { + "epoch": 1.3, + "grad_norm": 0.3708173720611468, + "learning_rate": 5.583739183923732e-05, + "loss": 1.0883, + "step": 710 + }, + { + "epoch": 1.3, + "grad_norm": 0.3631427403535479, + "learning_rate": 5.575906943070915e-05, + "loss": 1.1155, + "step": 711 + }, + { + "epoch": 1.3, + "grad_norm": 0.3305201458598695, + "learning_rate": 5.5680675435012834e-05, + "loss": 1.0958, + "step": 712 + }, + { + "epoch": 1.3, + "grad_norm": 0.34978833532083714, + "learning_rate": 5.5602210208261036e-05, + "loss": 1.1437, + "step": 713 + }, + { + "epoch": 1.31, + "grad_norm": 0.3510553882510229, + "learning_rate": 5.552367410688999e-05, + "loss": 1.0941, + "step": 714 + }, + { + "epoch": 1.31, + "grad_norm": 0.3523747462465078, + "learning_rate": 5.544506748765789e-05, + "loss": 1.1289, + "step": 715 + }, + { + "epoch": 1.31, + "grad_norm": 0.38262637783927445, + "learning_rate": 5.5366390707643266e-05, + "loss": 1.099, + "step": 716 + }, + { + "epoch": 1.31, + "grad_norm": 0.38620065989073454, + "learning_rate": 5.528764412424334e-05, + "loss": 1.083, + "step": 717 + }, + { + "epoch": 1.31, + "grad_norm": 0.3401355276121096, + "learning_rate": 5.520882809517245e-05, + "loss": 1.028, + "step": 718 + }, + { + "epoch": 1.32, + "grad_norm": 0.3392061008943934, + "learning_rate": 5.512994297846039e-05, + "loss": 1.1083, + "step": 719 + }, + { + "epoch": 1.32, + "grad_norm": 0.34219480421015414, + "learning_rate": 5.505098913245077e-05, + "loss": 1.1108, + "step": 720 + }, + { + "epoch": 1.32, + "grad_norm": 0.3275058061553761, + "learning_rate": 5.497196691579945e-05, + "loss": 1.111, + "step": 721 + }, + { + "epoch": 1.32, + "grad_norm": 0.36800249746509384, + "learning_rate": 5.489287668747283e-05, + "loss": 1.1221, + "step": 722 + }, + { + "epoch": 1.32, + "grad_norm": 0.4129005533101575, + "learning_rate": 5.481371880674628e-05, + "loss": 1.0966, + "step": 723 + }, + { + "epoch": 1.32, + "grad_norm": 0.36563906596251655, + "learning_rate": 5.4734493633202505e-05, + "loss": 1.0927, + "step": 724 + }, + { + "epoch": 1.33, + "grad_norm": 0.3614650536839971, + "learning_rate": 5.465520152672986e-05, + "loss": 1.13, + "step": 725 + }, + { + "epoch": 1.33, + "grad_norm": 0.36419665098633497, + "learning_rate": 5.4575842847520765e-05, + "loss": 1.1183, + "step": 726 + }, + { + "epoch": 1.33, + "grad_norm": 0.34490689807258995, + "learning_rate": 5.449641795607005e-05, + "loss": 1.0919, + "step": 727 + }, + { + "epoch": 1.33, + "grad_norm": 0.3627643746876298, + "learning_rate": 5.441692721317334e-05, + "loss": 1.0411, + "step": 728 + }, + { + "epoch": 1.33, + "grad_norm": 0.323620411949565, + "learning_rate": 5.433737097992537e-05, + "loss": 1.0725, + "step": 729 + }, + { + "epoch": 1.34, + "grad_norm": 0.3521599501824965, + "learning_rate": 5.425774961771838e-05, + "loss": 1.0926, + "step": 730 + }, + { + "epoch": 1.34, + "grad_norm": 0.3302390546764222, + "learning_rate": 5.417806348824047e-05, + "loss": 1.0468, + "step": 731 + }, + { + "epoch": 1.34, + "grad_norm": 0.3833325802616019, + "learning_rate": 5.4098312953473956e-05, + "loss": 1.1291, + "step": 732 + }, + { + "epoch": 1.34, + "grad_norm": 0.3708621126835512, + "learning_rate": 5.401849837569372e-05, + "loss": 1.0887, + "step": 733 + }, + { + "epoch": 1.34, + "grad_norm": 0.3625834373416278, + "learning_rate": 5.393862011746555e-05, + "loss": 1.0981, + "step": 734 + }, + { + "epoch": 1.34, + "grad_norm": 0.3583343965080617, + "learning_rate": 5.385867854164451e-05, + "loss": 1.1021, + "step": 735 + }, + { + "epoch": 1.35, + "grad_norm": 0.34598320594096066, + "learning_rate": 5.377867401137332e-05, + "loss": 1.1376, + "step": 736 + }, + { + "epoch": 1.35, + "grad_norm": 0.3046382791315433, + "learning_rate": 5.369860689008066e-05, + "loss": 1.0206, + "step": 737 + }, + { + "epoch": 1.35, + "grad_norm": 0.34464948380043725, + "learning_rate": 5.3618477541479505e-05, + "loss": 1.1084, + "step": 738 + }, + { + "epoch": 1.35, + "grad_norm": 0.3203242519627101, + "learning_rate": 5.353828632956557e-05, + "loss": 1.0731, + "step": 739 + }, + { + "epoch": 1.35, + "grad_norm": 0.3431169960355163, + "learning_rate": 5.3458033618615516e-05, + "loss": 1.091, + "step": 740 + }, + { + "epoch": 1.36, + "grad_norm": 0.33492074521678705, + "learning_rate": 5.337771977318543e-05, + "loss": 1.1112, + "step": 741 + }, + { + "epoch": 1.36, + "grad_norm": 0.32576546585541344, + "learning_rate": 5.3297345158109086e-05, + "loss": 1.0993, + "step": 742 + }, + { + "epoch": 1.36, + "grad_norm": 0.3410007245037574, + "learning_rate": 5.3216910138496286e-05, + "loss": 1.094, + "step": 743 + }, + { + "epoch": 1.36, + "grad_norm": 0.34891180680896833, + "learning_rate": 5.313641507973128e-05, + "loss": 1.1331, + "step": 744 + }, + { + "epoch": 1.36, + "grad_norm": 0.37135766946717214, + "learning_rate": 5.3055860347471006e-05, + "loss": 1.1, + "step": 745 + }, + { + "epoch": 1.36, + "grad_norm": 0.3465019415478411, + "learning_rate": 5.297524630764349e-05, + "loss": 1.1256, + "step": 746 + }, + { + "epoch": 1.37, + "grad_norm": 0.37035388481626563, + "learning_rate": 5.289457332644615e-05, + "loss": 1.0366, + "step": 747 + }, + { + "epoch": 1.37, + "grad_norm": 0.33853883270759155, + "learning_rate": 5.281384177034421e-05, + "loss": 1.0547, + "step": 748 + }, + { + "epoch": 1.37, + "grad_norm": 0.364306618627317, + "learning_rate": 5.2733052006068897e-05, + "loss": 1.0768, + "step": 749 + }, + { + "epoch": 1.37, + "grad_norm": 0.4021754315731627, + "learning_rate": 5.2652204400615916e-05, + "loss": 1.1382, + "step": 750 + }, + { + "epoch": 1.37, + "grad_norm": 0.3332185389039008, + "learning_rate": 5.257129932124368e-05, + "loss": 1.0815, + "step": 751 + }, + { + "epoch": 1.38, + "grad_norm": 0.3453105709879854, + "learning_rate": 5.249033713547173e-05, + "loss": 1.1109, + "step": 752 + }, + { + "epoch": 1.38, + "grad_norm": 0.3385397539717797, + "learning_rate": 5.2409318211078966e-05, + "loss": 1.0529, + "step": 753 + }, + { + "epoch": 1.38, + "grad_norm": 0.33197994450130447, + "learning_rate": 5.232824291610206e-05, + "loss": 1.0721, + "step": 754 + }, + { + "epoch": 1.38, + "grad_norm": 0.32836289576124167, + "learning_rate": 5.224711161883375e-05, + "loss": 1.0459, + "step": 755 + }, + { + "epoch": 1.38, + "grad_norm": 0.32491620058831744, + "learning_rate": 5.216592468782117e-05, + "loss": 1.0897, + "step": 756 + }, + { + "epoch": 1.38, + "grad_norm": 0.3137879047811153, + "learning_rate": 5.2084682491864155e-05, + "loss": 1.096, + "step": 757 + }, + { + "epoch": 1.39, + "grad_norm": 0.3356938043023012, + "learning_rate": 5.200338540001364e-05, + "loss": 1.0827, + "step": 758 + }, + { + "epoch": 1.39, + "grad_norm": 0.36044340490819055, + "learning_rate": 5.192203378156984e-05, + "loss": 1.0617, + "step": 759 + }, + { + "epoch": 1.39, + "grad_norm": 0.34674262047888293, + "learning_rate": 5.184062800608077e-05, + "loss": 1.1267, + "step": 760 + }, + { + "epoch": 1.39, + "grad_norm": 0.32469442322149333, + "learning_rate": 5.1759168443340375e-05, + "loss": 1.1483, + "step": 761 + }, + { + "epoch": 1.39, + "grad_norm": 0.3290384307774216, + "learning_rate": 5.167765546338698e-05, + "loss": 1.047, + "step": 762 + }, + { + "epoch": 1.4, + "grad_norm": 0.31637612188770403, + "learning_rate": 5.1596089436501525e-05, + "loss": 1.0311, + "step": 763 + }, + { + "epoch": 1.4, + "grad_norm": 0.3168693829641207, + "learning_rate": 5.151447073320597e-05, + "loss": 1.1405, + "step": 764 + }, + { + "epoch": 1.4, + "grad_norm": 0.34322421571238926, + "learning_rate": 5.143279972426153e-05, + "loss": 1.1428, + "step": 765 + }, + { + "epoch": 1.4, + "grad_norm": 0.3291030435830325, + "learning_rate": 5.1351076780667026e-05, + "loss": 1.0473, + "step": 766 + }, + { + "epoch": 1.4, + "grad_norm": 0.33772039158758044, + "learning_rate": 5.1269302273657195e-05, + "loss": 1.0909, + "step": 767 + }, + { + "epoch": 1.4, + "grad_norm": 0.3802031736890876, + "learning_rate": 5.118747657470102e-05, + "loss": 1.1482, + "step": 768 + }, + { + "epoch": 1.41, + "grad_norm": 0.3296067628997962, + "learning_rate": 5.1105600055500025e-05, + "loss": 1.0085, + "step": 769 + }, + { + "epoch": 1.41, + "grad_norm": 0.3707139982828035, + "learning_rate": 5.102367308798658e-05, + "loss": 1.0746, + "step": 770 + }, + { + "epoch": 1.41, + "grad_norm": 0.3378537316757011, + "learning_rate": 5.094169604432225e-05, + "loss": 1.0482, + "step": 771 + }, + { + "epoch": 1.41, + "grad_norm": 0.4008417246255145, + "learning_rate": 5.085966929689601e-05, + "loss": 1.1065, + "step": 772 + }, + { + "epoch": 1.41, + "grad_norm": 0.3244385106988064, + "learning_rate": 5.077759321832271e-05, + "loss": 1.0827, + "step": 773 + }, + { + "epoch": 1.42, + "grad_norm": 0.37228575732812336, + "learning_rate": 5.0695468181441215e-05, + "loss": 1.1146, + "step": 774 + }, + { + "epoch": 1.42, + "grad_norm": 0.33761714797540276, + "learning_rate": 5.061329455931283e-05, + "loss": 1.092, + "step": 775 + }, + { + "epoch": 1.42, + "grad_norm": 0.3158158390913494, + "learning_rate": 5.053107272521955e-05, + "loss": 1.1058, + "step": 776 + }, + { + "epoch": 1.42, + "grad_norm": 0.3691501929738938, + "learning_rate": 5.044880305266239e-05, + "loss": 1.1599, + "step": 777 + }, + { + "epoch": 1.42, + "grad_norm": 0.33730914019805525, + "learning_rate": 5.0366485915359645e-05, + "loss": 1.0615, + "step": 778 + }, + { + "epoch": 1.42, + "grad_norm": 0.34970059240017, + "learning_rate": 5.0284121687245257e-05, + "loss": 1.1475, + "step": 779 + }, + { + "epoch": 1.43, + "grad_norm": 0.3374028029407197, + "learning_rate": 5.020171074246707e-05, + "loss": 1.0926, + "step": 780 + }, + { + "epoch": 1.43, + "grad_norm": 0.3350020681123992, + "learning_rate": 5.011925345538514e-05, + "loss": 1.1276, + "step": 781 + }, + { + "epoch": 1.43, + "grad_norm": 0.3224228965786606, + "learning_rate": 5.003675020057003e-05, + "loss": 1.0183, + "step": 782 + }, + { + "epoch": 1.43, + "grad_norm": 0.3357310714740298, + "learning_rate": 4.995420135280114e-05, + "loss": 1.1114, + "step": 783 + }, + { + "epoch": 1.43, + "grad_norm": 0.3590203255363759, + "learning_rate": 4.9871607287064966e-05, + "loss": 1.1504, + "step": 784 + }, + { + "epoch": 1.44, + "grad_norm": 0.33011195419611655, + "learning_rate": 4.9788968378553396e-05, + "loss": 1.0826, + "step": 785 + }, + { + "epoch": 1.44, + "grad_norm": 0.31088868195439445, + "learning_rate": 4.970628500266207e-05, + "loss": 1.0704, + "step": 786 + }, + { + "epoch": 1.44, + "grad_norm": 0.3144996103179409, + "learning_rate": 4.962355753498858e-05, + "loss": 1.1403, + "step": 787 + }, + { + "epoch": 1.44, + "grad_norm": 0.3147269555419068, + "learning_rate": 4.954078635133081e-05, + "loss": 1.0898, + "step": 788 + }, + { + "epoch": 1.44, + "grad_norm": 0.3280151747783868, + "learning_rate": 4.945797182768524e-05, + "loss": 1.1115, + "step": 789 + }, + { + "epoch": 1.44, + "grad_norm": 0.3551996569232493, + "learning_rate": 4.937511434024524e-05, + "loss": 1.1731, + "step": 790 + }, + { + "epoch": 1.45, + "grad_norm": 0.343863208057807, + "learning_rate": 4.9292214265399336e-05, + "loss": 1.0866, + "step": 791 + }, + { + "epoch": 1.45, + "grad_norm": 0.37316699385322466, + "learning_rate": 4.920927197972949e-05, + "loss": 1.1083, + "step": 792 + }, + { + "epoch": 1.45, + "grad_norm": 0.3635739774067832, + "learning_rate": 4.9126287860009453e-05, + "loss": 1.1393, + "step": 793 + }, + { + "epoch": 1.45, + "grad_norm": 0.3755910554972886, + "learning_rate": 4.9043262283202974e-05, + "loss": 1.1624, + "step": 794 + }, + { + "epoch": 1.45, + "grad_norm": 0.3635899120146823, + "learning_rate": 4.8960195626462145e-05, + "loss": 1.2095, + "step": 795 + }, + { + "epoch": 1.46, + "grad_norm": 0.3642202684342816, + "learning_rate": 4.8877088267125664e-05, + "loss": 1.1099, + "step": 796 + }, + { + "epoch": 1.46, + "grad_norm": 0.3339946548799316, + "learning_rate": 4.879394058271712e-05, + "loss": 1.1157, + "step": 797 + }, + { + "epoch": 1.46, + "grad_norm": 0.3457189703100475, + "learning_rate": 4.871075295094329e-05, + "loss": 1.129, + "step": 798 + }, + { + "epoch": 1.46, + "grad_norm": 0.3550931839691424, + "learning_rate": 4.862752574969241e-05, + "loss": 1.076, + "step": 799 + }, + { + "epoch": 1.46, + "grad_norm": 0.36139108917966734, + "learning_rate": 4.8544259357032475e-05, + "loss": 1.1577, + "step": 800 + }, + { + "epoch": 1.0, + "grad_norm": 0.39569703665247874, + "learning_rate": 4.8460954151209486e-05, + "loss": 1.0543, + "step": 801 + }, + { + "epoch": 1.0, + "grad_norm": 0.3879033670170866, + "learning_rate": 4.837761051064579e-05, + "loss": 1.0688, + "step": 802 + }, + { + "epoch": 1.01, + "grad_norm": 0.3796846713967255, + "learning_rate": 4.8294228813938285e-05, + "loss": 0.9911, + "step": 803 + }, + { + "epoch": 1.01, + "grad_norm": 0.4007831430409375, + "learning_rate": 4.8210809439856804e-05, + "loss": 1.0126, + "step": 804 + }, + { + "epoch": 1.01, + "grad_norm": 0.37588078665500885, + "learning_rate": 4.8127352767342276e-05, + "loss": 0.9302, + "step": 805 + }, + { + "epoch": 1.01, + "grad_norm": 0.4078509175013281, + "learning_rate": 4.8043859175505095e-05, + "loss": 0.9982, + "step": 806 + }, + { + "epoch": 1.01, + "grad_norm": 0.379096046185539, + "learning_rate": 4.7960329043623344e-05, + "loss": 1.0035, + "step": 807 + }, + { + "epoch": 1.01, + "grad_norm": 0.3813938568133554, + "learning_rate": 4.787676275114111e-05, + "loss": 0.9579, + "step": 808 + }, + { + "epoch": 1.02, + "grad_norm": 0.3686863564511168, + "learning_rate": 4.779316067766673e-05, + "loss": 1.0105, + "step": 809 + }, + { + "epoch": 1.02, + "grad_norm": 0.4263940878847523, + "learning_rate": 4.770952320297109e-05, + "loss": 1.0677, + "step": 810 + }, + { + "epoch": 1.02, + "grad_norm": 0.37178778374665006, + "learning_rate": 4.7625850706985886e-05, + "loss": 1.0019, + "step": 811 + }, + { + "epoch": 1.02, + "grad_norm": 0.36803355429187945, + "learning_rate": 4.7542143569801894e-05, + "loss": 0.9937, + "step": 812 + }, + { + "epoch": 1.02, + "grad_norm": 0.3897072472941179, + "learning_rate": 4.745840217166725e-05, + "loss": 1.0877, + "step": 813 + }, + { + "epoch": 1.03, + "grad_norm": 0.35571833841716255, + "learning_rate": 4.737462689298577e-05, + "loss": 1.0015, + "step": 814 + }, + { + "epoch": 1.03, + "grad_norm": 0.38930229991094323, + "learning_rate": 4.7290818114315086e-05, + "loss": 1.028, + "step": 815 + }, + { + "epoch": 1.03, + "grad_norm": 0.411005007105147, + "learning_rate": 4.72069762163651e-05, + "loss": 1.0068, + "step": 816 + }, + { + "epoch": 1.03, + "grad_norm": 0.3980240190337736, + "learning_rate": 4.7123101579996106e-05, + "loss": 0.9919, + "step": 817 + }, + { + "epoch": 1.03, + "grad_norm": 0.36369517703115467, + "learning_rate": 4.7039194586217136e-05, + "loss": 0.967, + "step": 818 + }, + { + "epoch": 1.03, + "grad_norm": 0.38591148840458894, + "learning_rate": 4.695525561618418e-05, + "loss": 0.9743, + "step": 819 + }, + { + "epoch": 1.04, + "grad_norm": 0.45873135108949337, + "learning_rate": 4.687128505119853e-05, + "loss": 1.0516, + "step": 820 + }, + { + "epoch": 1.04, + "grad_norm": 0.3866330351411308, + "learning_rate": 4.6787283272704966e-05, + "loss": 0.9939, + "step": 821 + }, + { + "epoch": 1.04, + "grad_norm": 0.4620340173291326, + "learning_rate": 4.670325066229009e-05, + "loss": 1.0526, + "step": 822 + }, + { + "epoch": 1.04, + "grad_norm": 0.38877454299870284, + "learning_rate": 4.661918760168052e-05, + "loss": 0.9904, + "step": 823 + }, + { + "epoch": 1.04, + "grad_norm": 0.3880489386116793, + "learning_rate": 4.653509447274121e-05, + "loss": 0.9623, + "step": 824 + }, + { + "epoch": 1.05, + "grad_norm": 0.3827392356186151, + "learning_rate": 4.6450971657473743e-05, + "loss": 1.0772, + "step": 825 + }, + { + "epoch": 1.05, + "grad_norm": 0.4132814641854327, + "learning_rate": 4.63668195380145e-05, + "loss": 1.0533, + "step": 826 + }, + { + "epoch": 1.05, + "grad_norm": 0.3703610182402835, + "learning_rate": 4.628263849663301e-05, + "loss": 0.9336, + "step": 827 + }, + { + "epoch": 1.05, + "grad_norm": 0.4152053683299823, + "learning_rate": 4.619842891573016e-05, + "loss": 0.9801, + "step": 828 + }, + { + "epoch": 1.05, + "grad_norm": 0.41791059043554274, + "learning_rate": 4.6114191177836514e-05, + "loss": 1.0617, + "step": 829 + }, + { + "epoch": 1.05, + "grad_norm": 0.46363896517299136, + "learning_rate": 4.6029925665610524e-05, + "loss": 0.9687, + "step": 830 + }, + { + "epoch": 1.06, + "grad_norm": 0.41141959057512445, + "learning_rate": 4.59456327618368e-05, + "loss": 1.0965, + "step": 831 + }, + { + "epoch": 1.06, + "grad_norm": 0.3789192764519836, + "learning_rate": 4.5861312849424386e-05, + "loss": 0.9793, + "step": 832 + }, + { + "epoch": 1.06, + "grad_norm": 0.4047291581107866, + "learning_rate": 4.5776966311405035e-05, + "loss": 1.0342, + "step": 833 + }, + { + "epoch": 1.06, + "grad_norm": 0.4425157400959256, + "learning_rate": 4.5692593530931416e-05, + "loss": 1.0892, + "step": 834 + }, + { + "epoch": 1.06, + "grad_norm": 0.3707332144806616, + "learning_rate": 4.560819489127545e-05, + "loss": 0.9815, + "step": 835 + }, + { + "epoch": 1.07, + "grad_norm": 0.3897444102572823, + "learning_rate": 4.552377077582646e-05, + "loss": 0.9884, + "step": 836 + }, + { + "epoch": 1.07, + "grad_norm": 0.42725787957019346, + "learning_rate": 4.543932156808959e-05, + "loss": 0.9972, + "step": 837 + }, + { + "epoch": 1.07, + "grad_norm": 0.40615269781820007, + "learning_rate": 4.535484765168386e-05, + "loss": 0.9529, + "step": 838 + }, + { + "epoch": 1.07, + "grad_norm": 0.3505829736050887, + "learning_rate": 4.527034941034063e-05, + "loss": 0.9492, + "step": 839 + }, + { + "epoch": 1.07, + "grad_norm": 0.36688064686440497, + "learning_rate": 4.51858272279017e-05, + "loss": 0.9592, + "step": 840 + }, + { + "epoch": 1.07, + "grad_norm": 0.4043468777955929, + "learning_rate": 4.5101281488317634e-05, + "loss": 1.048, + "step": 841 + }, + { + "epoch": 1.08, + "grad_norm": 0.3811489793242706, + "learning_rate": 4.501671257564602e-05, + "loss": 1.0138, + "step": 842 + }, + { + "epoch": 1.08, + "grad_norm": 0.39813004142325986, + "learning_rate": 4.49321208740497e-05, + "loss": 1.071, + "step": 843 + }, + { + "epoch": 1.08, + "grad_norm": 0.3809751022095503, + "learning_rate": 4.484750676779504e-05, + "loss": 1.0351, + "step": 844 + }, + { + "epoch": 1.08, + "grad_norm": 0.384312178013823, + "learning_rate": 4.4762870641250185e-05, + "loss": 0.9737, + "step": 845 + }, + { + "epoch": 1.08, + "grad_norm": 0.40769404907923557, + "learning_rate": 4.467821287888331e-05, + "loss": 0.9659, + "step": 846 + }, + { + "epoch": 1.09, + "grad_norm": 0.39594136851937817, + "learning_rate": 4.459353386526086e-05, + "loss": 0.9405, + "step": 847 + }, + { + "epoch": 1.09, + "grad_norm": 0.37180161011562185, + "learning_rate": 4.450883398504584e-05, + "loss": 1.0732, + "step": 848 + }, + { + "epoch": 1.09, + "grad_norm": 0.3772603623154663, + "learning_rate": 4.442411362299602e-05, + "loss": 0.9646, + "step": 849 + }, + { + "epoch": 1.09, + "grad_norm": 0.4346142368506476, + "learning_rate": 4.433937316396224e-05, + "loss": 0.9572, + "step": 850 + }, + { + "epoch": 1.09, + "grad_norm": 0.3997258084612474, + "learning_rate": 4.425461299288659e-05, + "loss": 0.9492, + "step": 851 + }, + { + "epoch": 1.1, + "grad_norm": 0.41245476865247155, + "learning_rate": 4.416983349480073e-05, + "loss": 0.8732, + "step": 852 + }, + { + "epoch": 1.1, + "grad_norm": 0.6761499297939195, + "learning_rate": 4.408503505482412e-05, + "loss": 1.0425, + "step": 853 + }, + { + "epoch": 1.1, + "grad_norm": 0.40340979486858985, + "learning_rate": 4.400021805816225e-05, + "loss": 0.9596, + "step": 854 + }, + { + "epoch": 1.1, + "grad_norm": 0.43290732392699666, + "learning_rate": 4.391538289010493e-05, + "loss": 1.0123, + "step": 855 + }, + { + "epoch": 1.1, + "grad_norm": 0.36878054442190156, + "learning_rate": 4.383052993602448e-05, + "loss": 0.9448, + "step": 856 + }, + { + "epoch": 1.1, + "grad_norm": 0.7146145128961262, + "learning_rate": 4.374565958137404e-05, + "loss": 1.0342, + "step": 857 + }, + { + "epoch": 1.11, + "grad_norm": 0.44429357586145607, + "learning_rate": 4.3660772211685775e-05, + "loss": 1.0436, + "step": 858 + }, + { + "epoch": 1.11, + "grad_norm": 0.4565751973640598, + "learning_rate": 4.357586821256918e-05, + "loss": 1.0311, + "step": 859 + }, + { + "epoch": 1.11, + "grad_norm": 0.3919991236654277, + "learning_rate": 4.349094796970925e-05, + "loss": 1.1401, + "step": 860 + }, + { + "epoch": 1.11, + "grad_norm": 0.4347441949284011, + "learning_rate": 4.3406011868864795e-05, + "loss": 1.0252, + "step": 861 + }, + { + "epoch": 1.11, + "grad_norm": 0.38339976027415407, + "learning_rate": 4.3321060295866635e-05, + "loss": 1.0536, + "step": 862 + }, + { + "epoch": 1.12, + "grad_norm": 0.37688790408195166, + "learning_rate": 4.32360936366159e-05, + "loss": 1.012, + "step": 863 + }, + { + "epoch": 1.12, + "grad_norm": 0.4317538207582504, + "learning_rate": 4.315111227708224e-05, + "loss": 1.0505, + "step": 864 + }, + { + "epoch": 1.12, + "grad_norm": 0.4145324872228796, + "learning_rate": 4.306611660330208e-05, + "loss": 1.0496, + "step": 865 + }, + { + "epoch": 1.12, + "grad_norm": 0.416535227064448, + "learning_rate": 4.298110700137687e-05, + "loss": 0.9628, + "step": 866 + }, + { + "epoch": 1.12, + "grad_norm": 0.46564356187492717, + "learning_rate": 4.2896083857471345e-05, + "loss": 1.0016, + "step": 867 + }, + { + "epoch": 1.12, + "grad_norm": 0.4228980941889828, + "learning_rate": 4.281104755781172e-05, + "loss": 1.0904, + "step": 868 + }, + { + "epoch": 1.13, + "grad_norm": 0.4267821214430208, + "learning_rate": 4.272599848868402e-05, + "loss": 1.0544, + "step": 869 + }, + { + "epoch": 1.13, + "grad_norm": 0.45763332095792075, + "learning_rate": 4.264093703643223e-05, + "loss": 1.0686, + "step": 870 + }, + { + "epoch": 1.13, + "grad_norm": 0.4347555516548761, + "learning_rate": 4.255586358745662e-05, + "loss": 1.0264, + "step": 871 + }, + { + "epoch": 1.13, + "grad_norm": 0.3817726381103066, + "learning_rate": 4.247077852821194e-05, + "loss": 1.0045, + "step": 872 + }, + { + "epoch": 1.13, + "grad_norm": 0.3882808845457995, + "learning_rate": 4.2385682245205685e-05, + "loss": 1.0193, + "step": 873 + }, + { + "epoch": 1.14, + "grad_norm": 0.39410930252966775, + "learning_rate": 4.230057512499634e-05, + "loss": 0.9832, + "step": 874 + }, + { + "epoch": 1.14, + "grad_norm": 0.4373094593907156, + "learning_rate": 4.221545755419159e-05, + "loss": 1.0343, + "step": 875 + }, + { + "epoch": 1.14, + "grad_norm": 0.4462843721698891, + "learning_rate": 4.2130329919446646e-05, + "loss": 1.0324, + "step": 876 + }, + { + "epoch": 1.14, + "grad_norm": 0.4747274247448112, + "learning_rate": 4.20451926074624e-05, + "loss": 0.9903, + "step": 877 + }, + { + "epoch": 1.14, + "grad_norm": 0.4157472897596409, + "learning_rate": 4.196004600498369e-05, + "loss": 0.9266, + "step": 878 + }, + { + "epoch": 1.14, + "grad_norm": 0.41625958088960685, + "learning_rate": 4.1874890498797605e-05, + "loss": 0.9658, + "step": 879 + }, + { + "epoch": 1.15, + "grad_norm": 0.44784944130574333, + "learning_rate": 4.178972647573163e-05, + "loss": 0.9671, + "step": 880 + }, + { + "epoch": 1.15, + "grad_norm": 0.4116839177956385, + "learning_rate": 4.1704554322651975e-05, + "loss": 0.9591, + "step": 881 + }, + { + "epoch": 1.15, + "grad_norm": 0.4025569857639452, + "learning_rate": 4.161937442646176e-05, + "loss": 1.0072, + "step": 882 + }, + { + "epoch": 1.15, + "grad_norm": 0.41518478124763597, + "learning_rate": 4.1534187174099285e-05, + "loss": 1.0275, + "step": 883 + }, + { + "epoch": 1.15, + "grad_norm": 0.3987815564664466, + "learning_rate": 4.1448992952536275e-05, + "loss": 1.0039, + "step": 884 + }, + { + "epoch": 1.16, + "grad_norm": 0.4270378155679982, + "learning_rate": 4.136379214877609e-05, + "loss": 1.0369, + "step": 885 + }, + { + "epoch": 1.16, + "grad_norm": 0.42144733922972777, + "learning_rate": 4.127858514985203e-05, + "loss": 1.0269, + "step": 886 + }, + { + "epoch": 1.16, + "grad_norm": 0.4198664438272548, + "learning_rate": 4.1193372342825494e-05, + "loss": 1.0427, + "step": 887 + }, + { + "epoch": 1.16, + "grad_norm": 0.3985048256281719, + "learning_rate": 4.1108154114784275e-05, + "loss": 1.0702, + "step": 888 + }, + { + "epoch": 1.16, + "grad_norm": 0.605520808292362, + "learning_rate": 4.102293085284083e-05, + "loss": 0.9749, + "step": 889 + }, + { + "epoch": 1.16, + "grad_norm": 0.4150515863924052, + "learning_rate": 4.0937702944130426e-05, + "loss": 1.0231, + "step": 890 + }, + { + "epoch": 1.17, + "grad_norm": 0.3935997576565283, + "learning_rate": 4.085247077580948e-05, + "loss": 1.0014, + "step": 891 + }, + { + "epoch": 1.17, + "grad_norm": 0.399446131403209, + "learning_rate": 4.076723473505374e-05, + "loss": 0.9602, + "step": 892 + }, + { + "epoch": 1.17, + "grad_norm": 0.4406024397129952, + "learning_rate": 4.068199520905655e-05, + "loss": 1.0425, + "step": 893 + }, + { + "epoch": 1.17, + "grad_norm": 0.4036917571496492, + "learning_rate": 4.059675258502709e-05, + "loss": 0.973, + "step": 894 + }, + { + "epoch": 1.17, + "grad_norm": 0.4057196459433299, + "learning_rate": 4.05115072501886e-05, + "loss": 0.9997, + "step": 895 + }, + { + "epoch": 1.18, + "grad_norm": 0.4374124954708759, + "learning_rate": 4.0426259591776645e-05, + "loss": 0.9826, + "step": 896 + }, + { + "epoch": 1.18, + "grad_norm": 0.4545699371285546, + "learning_rate": 4.0341009997037356e-05, + "loss": 1.0554, + "step": 897 + }, + { + "epoch": 1.18, + "grad_norm": 0.4251917031237376, + "learning_rate": 4.025575885322563e-05, + "loss": 1.0217, + "step": 898 + }, + { + "epoch": 1.18, + "grad_norm": 0.3857651901893941, + "learning_rate": 4.0170506547603427e-05, + "loss": 1.0317, + "step": 899 + }, + { + "epoch": 1.18, + "grad_norm": 0.46323573798490897, + "learning_rate": 4.008525346743797e-05, + "loss": 1.0398, + "step": 900 + }, + { + "epoch": 1.18, + "grad_norm": 0.4011541121460918, + "learning_rate": 4e-05, + "loss": 1.0706, + "step": 901 + }, + { + "epoch": 1.19, + "grad_norm": 0.46493281221028004, + "learning_rate": 3.991474653256204e-05, + "loss": 1.0525, + "step": 902 + }, + { + "epoch": 1.19, + "grad_norm": 0.41683080924539023, + "learning_rate": 3.982949345239658e-05, + "loss": 1.0905, + "step": 903 + }, + { + "epoch": 1.19, + "grad_norm": 0.4750350025014512, + "learning_rate": 3.974424114677437e-05, + "loss": 1.049, + "step": 904 + }, + { + "epoch": 1.19, + "grad_norm": 0.3867445073614702, + "learning_rate": 3.965899000296266e-05, + "loss": 0.9624, + "step": 905 + }, + { + "epoch": 1.19, + "grad_norm": 0.378387661131469, + "learning_rate": 3.957374040822335e-05, + "loss": 1.0223, + "step": 906 + }, + { + "epoch": 1.2, + "grad_norm": 0.3905996390559077, + "learning_rate": 3.948849274981141e-05, + "loss": 1.0315, + "step": 907 + }, + { + "epoch": 1.2, + "grad_norm": 0.4139717689498189, + "learning_rate": 3.940324741497291e-05, + "loss": 0.9297, + "step": 908 + }, + { + "epoch": 1.2, + "grad_norm": 0.39086355684921514, + "learning_rate": 3.9318004790943465e-05, + "loss": 0.9684, + "step": 909 + }, + { + "epoch": 1.2, + "grad_norm": 0.4334915643736419, + "learning_rate": 3.923276526494627e-05, + "loss": 0.996, + "step": 910 + }, + { + "epoch": 1.2, + "grad_norm": 0.40782018986229496, + "learning_rate": 3.9147529224190536e-05, + "loss": 1.0875, + "step": 911 + }, + { + "epoch": 1.2, + "grad_norm": 0.43578702386625723, + "learning_rate": 3.906229705586959e-05, + "loss": 1.1214, + "step": 912 + }, + { + "epoch": 1.21, + "grad_norm": 0.414945683409524, + "learning_rate": 3.89770691471592e-05, + "loss": 1.1037, + "step": 913 + }, + { + "epoch": 1.21, + "grad_norm": 0.40665801579679106, + "learning_rate": 3.889184588521573e-05, + "loss": 0.9743, + "step": 914 + }, + { + "epoch": 1.21, + "grad_norm": 0.4064250611574517, + "learning_rate": 3.880662765717453e-05, + "loss": 0.8814, + "step": 915 + }, + { + "epoch": 1.21, + "grad_norm": 0.48023046298843347, + "learning_rate": 3.8721414850147985e-05, + "loss": 0.9663, + "step": 916 + }, + { + "epoch": 1.21, + "grad_norm": 0.42358024833566227, + "learning_rate": 3.8636207851223924e-05, + "loss": 1.0491, + "step": 917 + }, + { + "epoch": 1.22, + "grad_norm": 0.41522494786195835, + "learning_rate": 3.855100704746374e-05, + "loss": 1.033, + "step": 918 + }, + { + "epoch": 1.22, + "grad_norm": 0.40890517696706496, + "learning_rate": 3.8465812825900715e-05, + "loss": 1.0369, + "step": 919 + }, + { + "epoch": 1.22, + "grad_norm": 0.4325851866408538, + "learning_rate": 3.838062557353825e-05, + "loss": 0.9362, + "step": 920 + }, + { + "epoch": 1.22, + "grad_norm": 0.4185860919050069, + "learning_rate": 3.8295445677348025e-05, + "loss": 1.026, + "step": 921 + }, + { + "epoch": 1.22, + "grad_norm": 0.3975762375934804, + "learning_rate": 3.8210273524268375e-05, + "loss": 1.0412, + "step": 922 + }, + { + "epoch": 1.22, + "grad_norm": 0.41725298241987474, + "learning_rate": 3.8125109501202395e-05, + "loss": 1.0004, + "step": 923 + }, + { + "epoch": 1.23, + "grad_norm": 0.455183913149126, + "learning_rate": 3.803995399501632e-05, + "loss": 1.0594, + "step": 924 + }, + { + "epoch": 1.23, + "grad_norm": 0.3993993856483797, + "learning_rate": 3.795480739253761e-05, + "loss": 0.9761, + "step": 925 + }, + { + "epoch": 1.23, + "grad_norm": 0.41638796815161494, + "learning_rate": 3.786967008055337e-05, + "loss": 1.0369, + "step": 926 + }, + { + "epoch": 1.23, + "grad_norm": 0.40015112695810534, + "learning_rate": 3.7784542445808414e-05, + "loss": 1.0271, + "step": 927 + }, + { + "epoch": 1.23, + "grad_norm": 0.3995749494729548, + "learning_rate": 3.769942487500368e-05, + "loss": 1.0613, + "step": 928 + }, + { + "epoch": 1.24, + "grad_norm": 0.4073556267037492, + "learning_rate": 3.761431775479432e-05, + "loss": 1.0528, + "step": 929 + }, + { + "epoch": 1.24, + "grad_norm": 0.44218148822636044, + "learning_rate": 3.752922147178807e-05, + "loss": 1.0742, + "step": 930 + }, + { + "epoch": 1.24, + "grad_norm": 0.4435063485893757, + "learning_rate": 3.744413641254339e-05, + "loss": 1.0825, + "step": 931 + }, + { + "epoch": 1.24, + "grad_norm": 0.46841574994107515, + "learning_rate": 3.735906296356778e-05, + "loss": 1.0471, + "step": 932 + }, + { + "epoch": 1.24, + "grad_norm": 0.40093716627657294, + "learning_rate": 3.727400151131599e-05, + "loss": 1.0474, + "step": 933 + }, + { + "epoch": 1.25, + "grad_norm": 0.3866415067997244, + "learning_rate": 3.71889524421883e-05, + "loss": 1.0209, + "step": 934 + }, + { + "epoch": 1.25, + "grad_norm": 0.4881546110706673, + "learning_rate": 3.710391614252867e-05, + "loss": 1.0768, + "step": 935 + }, + { + "epoch": 1.25, + "grad_norm": 0.4133084639324523, + "learning_rate": 3.701889299862314e-05, + "loss": 1.0423, + "step": 936 + }, + { + "epoch": 1.25, + "grad_norm": 0.40523563084001196, + "learning_rate": 3.6933883396697936e-05, + "loss": 1.005, + "step": 937 + }, + { + "epoch": 1.25, + "grad_norm": 0.38757352418642405, + "learning_rate": 3.684888772291777e-05, + "loss": 0.9659, + "step": 938 + }, + { + "epoch": 1.25, + "grad_norm": 0.421394551890689, + "learning_rate": 3.676390636338411e-05, + "loss": 1.0454, + "step": 939 + }, + { + "epoch": 1.26, + "grad_norm": 0.45693070958342186, + "learning_rate": 3.667893970413337e-05, + "loss": 1.1459, + "step": 940 + }, + { + "epoch": 1.26, + "grad_norm": 0.4172025376377795, + "learning_rate": 3.659398813113522e-05, + "loss": 0.9954, + "step": 941 + }, + { + "epoch": 1.26, + "grad_norm": 0.3871624019510191, + "learning_rate": 3.650905203029075e-05, + "loss": 1.0441, + "step": 942 + }, + { + "epoch": 1.26, + "grad_norm": 0.38541342610032325, + "learning_rate": 3.642413178743083e-05, + "loss": 0.9465, + "step": 943 + }, + { + "epoch": 1.26, + "grad_norm": 0.4208031670525743, + "learning_rate": 3.633922778831423e-05, + "loss": 1.0367, + "step": 944 + }, + { + "epoch": 1.27, + "grad_norm": 0.41867209013040035, + "learning_rate": 3.6254340418625975e-05, + "loss": 1.0868, + "step": 945 + }, + { + "epoch": 1.27, + "grad_norm": 0.431758149074127, + "learning_rate": 3.6169470063975536e-05, + "loss": 1.0689, + "step": 946 + }, + { + "epoch": 1.27, + "grad_norm": 0.4988803338819952, + "learning_rate": 3.608461710989509e-05, + "loss": 1.0879, + "step": 947 + }, + { + "epoch": 1.27, + "grad_norm": 0.4094858411191625, + "learning_rate": 3.5999781941837755e-05, + "loss": 1.0332, + "step": 948 + }, + { + "epoch": 1.27, + "grad_norm": 0.3831847195845155, + "learning_rate": 3.591496494517589e-05, + "loss": 0.9751, + "step": 949 + }, + { + "epoch": 1.27, + "grad_norm": 0.40535692821947267, + "learning_rate": 3.5830166505199284e-05, + "loss": 1.0594, + "step": 950 + }, + { + "epoch": 1.28, + "grad_norm": 0.4875663789389966, + "learning_rate": 3.574538700711343e-05, + "loss": 0.9749, + "step": 951 + }, + { + "epoch": 1.28, + "grad_norm": 0.5155923998285772, + "learning_rate": 3.566062683603778e-05, + "loss": 0.9999, + "step": 952 + }, + { + "epoch": 1.28, + "grad_norm": 0.5280285947816189, + "learning_rate": 3.557588637700399e-05, + "loss": 1.1061, + "step": 953 + }, + { + "epoch": 1.28, + "grad_norm": 0.46573407357796753, + "learning_rate": 3.5491166014954174e-05, + "loss": 1.102, + "step": 954 + }, + { + "epoch": 1.28, + "grad_norm": 0.4122542582865379, + "learning_rate": 3.540646613473915e-05, + "loss": 1.0469, + "step": 955 + }, + { + "epoch": 1.29, + "grad_norm": 0.41414476980823367, + "learning_rate": 3.53217871211167e-05, + "loss": 0.9973, + "step": 956 + }, + { + "epoch": 1.29, + "grad_norm": 0.4030707611608045, + "learning_rate": 3.523712935874983e-05, + "loss": 0.9796, + "step": 957 + }, + { + "epoch": 1.29, + "grad_norm": 0.4235313349747291, + "learning_rate": 3.5152493232204975e-05, + "loss": 1.0601, + "step": 958 + }, + { + "epoch": 1.29, + "grad_norm": 0.4165235178302652, + "learning_rate": 3.5067879125950316e-05, + "loss": 1.0358, + "step": 959 + }, + { + "epoch": 1.29, + "grad_norm": 0.44083984701952955, + "learning_rate": 3.4983287424354e-05, + "loss": 1.0957, + "step": 960 + }, + { + "epoch": 1.29, + "grad_norm": 0.3781161039063518, + "learning_rate": 3.489871851168238e-05, + "loss": 0.9838, + "step": 961 + }, + { + "epoch": 1.3, + "grad_norm": 0.4095747724038915, + "learning_rate": 3.4814172772098314e-05, + "loss": 1.014, + "step": 962 + }, + { + "epoch": 1.3, + "grad_norm": 0.42197119558898466, + "learning_rate": 3.472965058965938e-05, + "loss": 1.0096, + "step": 963 + }, + { + "epoch": 1.3, + "grad_norm": 0.4339963388152155, + "learning_rate": 3.464515234831615e-05, + "loss": 1.0158, + "step": 964 + }, + { + "epoch": 1.3, + "grad_norm": 0.4284638765548976, + "learning_rate": 3.4560678431910424e-05, + "loss": 1.1047, + "step": 965 + }, + { + "epoch": 1.3, + "grad_norm": 0.3935144535755794, + "learning_rate": 3.447622922417355e-05, + "loss": 0.9925, + "step": 966 + }, + { + "epoch": 1.31, + "grad_norm": 0.45884343961025, + "learning_rate": 3.439180510872457e-05, + "loss": 1.0583, + "step": 967 + }, + { + "epoch": 1.31, + "grad_norm": 0.42439320759788374, + "learning_rate": 3.4307406469068604e-05, + "loss": 0.9305, + "step": 968 + }, + { + "epoch": 1.31, + "grad_norm": 0.45770082390324845, + "learning_rate": 3.4223033688594985e-05, + "loss": 1.054, + "step": 969 + }, + { + "epoch": 1.31, + "grad_norm": 0.4284786643981094, + "learning_rate": 3.4138687150575634e-05, + "loss": 0.9409, + "step": 970 + }, + { + "epoch": 1.31, + "grad_norm": 0.41356124058383237, + "learning_rate": 3.4054367238163215e-05, + "loss": 1.0739, + "step": 971 + }, + { + "epoch": 1.31, + "grad_norm": 0.4255832249412624, + "learning_rate": 3.3970074334389496e-05, + "loss": 1.0764, + "step": 972 + }, + { + "epoch": 1.32, + "grad_norm": 0.4337695536142702, + "learning_rate": 3.388580882216349e-05, + "loss": 1.0195, + "step": 973 + }, + { + "epoch": 1.32, + "grad_norm": 0.41363495650922455, + "learning_rate": 3.380157108426985e-05, + "loss": 1.0615, + "step": 974 + }, + { + "epoch": 1.32, + "grad_norm": 0.3950691247686479, + "learning_rate": 3.371736150336701e-05, + "loss": 1.0283, + "step": 975 + }, + { + "epoch": 1.32, + "grad_norm": 0.4042823691555822, + "learning_rate": 3.3633180461985505e-05, + "loss": 1.0309, + "step": 976 + }, + { + "epoch": 1.32, + "grad_norm": 0.3921158850479399, + "learning_rate": 3.354902834252627e-05, + "loss": 1.068, + "step": 977 + }, + { + "epoch": 1.33, + "grad_norm": 0.38349545732725654, + "learning_rate": 3.346490552725879e-05, + "loss": 1.0886, + "step": 978 + }, + { + "epoch": 1.33, + "grad_norm": 0.38689221457248724, + "learning_rate": 3.33808123983195e-05, + "loss": 0.987, + "step": 979 + }, + { + "epoch": 1.33, + "grad_norm": 0.38660550867425647, + "learning_rate": 3.329674933770992e-05, + "loss": 1.069, + "step": 980 + }, + { + "epoch": 1.33, + "grad_norm": 0.3917593746353493, + "learning_rate": 3.321271672729504e-05, + "loss": 0.9858, + "step": 981 + }, + { + "epoch": 1.33, + "grad_norm": 0.4292314072827653, + "learning_rate": 3.3128714948801474e-05, + "loss": 1.0477, + "step": 982 + }, + { + "epoch": 1.33, + "grad_norm": 0.479414638418211, + "learning_rate": 3.3044744383815835e-05, + "loss": 1.0763, + "step": 983 + }, + { + "epoch": 1.34, + "grad_norm": 0.380831894995463, + "learning_rate": 3.2960805413782884e-05, + "loss": 1.0393, + "step": 984 + }, + { + "epoch": 1.34, + "grad_norm": 0.42402274703362114, + "learning_rate": 3.2876898420003914e-05, + "loss": 1.0837, + "step": 985 + }, + { + "epoch": 1.34, + "grad_norm": 0.4571447203722258, + "learning_rate": 3.279302378363491e-05, + "loss": 1.0594, + "step": 986 + }, + { + "epoch": 1.34, + "grad_norm": 0.3776673281658531, + "learning_rate": 3.270918188568493e-05, + "loss": 1.0121, + "step": 987 + }, + { + "epoch": 1.34, + "grad_norm": 0.4367173448132159, + "learning_rate": 3.262537310701425e-05, + "loss": 0.9612, + "step": 988 + }, + { + "epoch": 1.35, + "grad_norm": 0.43679765208840926, + "learning_rate": 3.254159782833276e-05, + "loss": 1.0565, + "step": 989 + }, + { + "epoch": 1.35, + "grad_norm": 0.4018151260013493, + "learning_rate": 3.2457856430198126e-05, + "loss": 0.9975, + "step": 990 + }, + { + "epoch": 1.35, + "grad_norm": 0.40461959940721076, + "learning_rate": 3.237414929301412e-05, + "loss": 1.0255, + "step": 991 + }, + { + "epoch": 1.35, + "grad_norm": 0.41342378541540653, + "learning_rate": 3.2290476797028926e-05, + "loss": 1.024, + "step": 992 + }, + { + "epoch": 1.35, + "grad_norm": 0.3926173909201105, + "learning_rate": 3.220683932233328e-05, + "loss": 1.0877, + "step": 993 + }, + { + "epoch": 1.35, + "grad_norm": 0.3835623199834992, + "learning_rate": 3.21232372488589e-05, + "loss": 1.0992, + "step": 994 + }, + { + "epoch": 1.36, + "grad_norm": 0.39901809497083496, + "learning_rate": 3.2039670956376656e-05, + "loss": 1.0723, + "step": 995 + }, + { + "epoch": 1.36, + "grad_norm": 0.3979604537466272, + "learning_rate": 3.195614082449492e-05, + "loss": 1.0201, + "step": 996 + }, + { + "epoch": 1.36, + "grad_norm": 0.4057122427176845, + "learning_rate": 3.1872647232657723e-05, + "loss": 1.0885, + "step": 997 + }, + { + "epoch": 1.36, + "grad_norm": 0.39747060350754754, + "learning_rate": 3.17891905601432e-05, + "loss": 1.0544, + "step": 998 + }, + { + "epoch": 1.36, + "grad_norm": 0.4397658078291558, + "learning_rate": 3.1705771186061715e-05, + "loss": 1.0998, + "step": 999 + }, + { + "epoch": 1.37, + "grad_norm": 0.37373547663810053, + "learning_rate": 3.162238948935423e-05, + "loss": 1.0465, + "step": 1000 + } + ], + "logging_steps": 1.0, + "max_steps": 1638, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 50, + "total_flos": 1036715984683008.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-1000/training_args.bin b/checkpoint-1000/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..8c2dfa20e1da5754719c3d7e300b9b86407f077f --- /dev/null +++ b/checkpoint-1000/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3f2f7bd873b9dca108c5ca2e32ea140480fabeed2dec60f702daabd0a44d071e +size 6776 diff --git a/checkpoint-1000/zero_to_fp32.py b/checkpoint-1000/zero_to_fp32.py new file mode 100755 index 0000000000000000000000000000000000000000..24cc342e78d1a006c782b3a4cd68d9ce786d8fd8 --- /dev/null +++ b/checkpoint-1000/zero_to_fp32.py @@ -0,0 +1,604 @@ +#!/usr/bin/env python + +# Copyright (c) Microsoft Corporation. +# SPDX-License-Identifier: Apache-2.0 + +# DeepSpeed Team + +# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets +# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in +# the future. Once extracted, the weights don't require DeepSpeed and can be used in any +# application. +# +# example: python zero_to_fp32.py . pytorch_model.bin + +import argparse +import torch +import glob +import math +import os +import re +from collections import OrderedDict +from dataclasses import dataclass + +# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with +# DeepSpeed data structures it has to be available in the current python environment. +from deepspeed.utils import logger +from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS, + FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES, + FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS) + + +@dataclass +class zero_model_state: + buffers: dict() + param_shapes: dict() + shared_params: list + ds_version: int + frozen_param_shapes: dict() + frozen_param_fragments: dict() + + +debug = 0 + +# load to cpu +device = torch.device('cpu') + + +def atoi(text): + return int(text) if text.isdigit() else text + + +def natural_keys(text): + ''' + alist.sort(key=natural_keys) sorts in human order + http://nedbatchelder.com/blog/200712/human_sorting.html + (See Toothy's implementation in the comments) + ''' + return [atoi(c) for c in re.split(r'(\d+)', text)] + + +def get_model_state_file(checkpoint_dir, zero_stage): + if not os.path.isdir(checkpoint_dir): + raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist") + + # there should be only one file + if zero_stage <= 2: + file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt") + elif zero_stage == 3: + file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt") + + if not os.path.exists(file): + raise FileNotFoundError(f"can't find model states file at '{file}'") + + return file + + +def get_checkpoint_files(checkpoint_dir, glob_pattern): + # XXX: need to test that this simple glob rule works for multi-node setup too + ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys) + + if len(ckpt_files) == 0: + raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'") + + return ckpt_files + + +def get_optim_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt") + + +def get_model_state_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_model_states.pt") + + +def parse_model_states(files): + zero_model_states = [] + for file in files: + state_dict = torch.load(file, map_location=device) + + if BUFFER_NAMES not in state_dict: + raise ValueError(f"{file} is not a model state checkpoint") + buffer_names = state_dict[BUFFER_NAMES] + if debug: + print("Found buffers:", buffer_names) + + # recover just the buffers while restoring them to fp32 if they were saved in fp16 + buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names} + param_shapes = state_dict[PARAM_SHAPES] + + # collect parameters that are included in param_shapes + param_names = [] + for s in param_shapes: + for name in s.keys(): + param_names.append(name) + + # update with frozen parameters + frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None) + if frozen_param_shapes is not None: + if debug: + print(f"Found frozen_param_shapes: {frozen_param_shapes}") + param_names += list(frozen_param_shapes.keys()) + + # handle shared params + shared_params = [[k, v] for k, v in state_dict["shared_params"].items()] + + ds_version = state_dict.get(DS_VERSION, None) + + frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None) + + z_model_state = zero_model_state(buffers=buffers, + param_shapes=param_shapes, + shared_params=shared_params, + ds_version=ds_version, + frozen_param_shapes=frozen_param_shapes, + frozen_param_fragments=frozen_param_fragments) + zero_model_states.append(z_model_state) + + return zero_model_states + + +def parse_optim_states(files, ds_checkpoint_dir): + + total_files = len(files) + state_dicts = [] + for f in files: + state_dict = torch.load(f, map_location=device) + # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights + # and also handle the case where it was already removed by another helper script + state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None) + state_dicts.append(state_dict) + + if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]: + raise ValueError(f"{files[0]} is not a zero checkpoint") + zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE] + world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT] + + # For ZeRO-2 each param group can have different partition_count as data parallelism for expert + # parameters can be different from data parallelism for non-expert parameters. So we can just + # use the max of the partition_count to get the dp world_size. + + if type(world_size) is list: + world_size = max(world_size) + + if world_size != total_files: + raise ValueError( + f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. " + "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes." + ) + + # the groups are named differently in each stage + if zero_stage <= 2: + fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS + elif zero_stage == 3: + fp32_groups_key = FP32_FLAT_GROUPS + else: + raise ValueError(f"unknown zero stage {zero_stage}") + + if zero_stage <= 2: + fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))] + elif zero_stage == 3: + # if there is more than one param group, there will be multiple flattened tensors - one + # flattened tensor per group - for simplicity merge them into a single tensor + # + # XXX: could make the script more memory efficient for when there are multiple groups - it + # will require matching the sub-lists of param_shapes for each param group flattened tensor + + fp32_flat_groups = [ + torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts)) + ] + + return zero_stage, world_size, fp32_flat_groups + + +def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters): + """ + Returns fp32 state_dict reconstructed from ds checkpoint + + Args: + - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are) + + """ + print(f"Processing zero checkpoint '{ds_checkpoint_dir}'") + + optim_files = get_optim_files(ds_checkpoint_dir) + zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir) + print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}") + + model_files = get_model_state_files(ds_checkpoint_dir) + + zero_model_states = parse_model_states(model_files) + print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}') + + if zero_stage <= 2: + return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + elif zero_stage == 3: + return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + + +def _zero2_merge_frozen_params(state_dict, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + frozen_param_fragments = zero_model_states[0].frozen_param_fragments + + if debug: + num_elem = sum(s.numel() for s in frozen_param_shapes.values()) + print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in frozen_param_fragments.values()]) + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + state_dict[name] = frozen_param_fragments[name] + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _has_callable(obj, fn): + attr = getattr(obj, fn, None) + return callable(attr) + + +def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + + # Reconstruction protocol: + # + # XXX: document this + + if debug: + for i in range(world_size): + for j in range(len(fp32_flat_groups[0])): + print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}") + + # XXX: memory usage doubles here (zero2) + num_param_groups = len(fp32_flat_groups[0]) + merged_single_partition_of_fp32_groups = [] + for i in range(num_param_groups): + merged_partitions = [sd[i] for sd in fp32_flat_groups] + full_single_fp32_vector = torch.cat(merged_partitions, 0) + merged_single_partition_of_fp32_groups.append(full_single_fp32_vector) + avail_numel = sum( + [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups]) + + if debug: + wanted_params = sum([len(shapes) for shapes in param_shapes]) + wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes]) + # not asserting if there is a mismatch due to possible padding + print(f"Have {avail_numel} numels to process.") + print(f"Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + total_numel = 0 + total_params = 0 + for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups): + offset = 0 + avail_numel = full_single_fp32_vector.numel() + for name, shape in shapes.items(): + + unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape) + total_numel += unpartitioned_numel + total_params += 1 + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape) + offset += unpartitioned_numel + + # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and + # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex + # paddings performed in the code it's almost impossible to predict the exact numbers w/o the + # live optimizer object, so we are checking that the numbers are within the right range + align_to = 2 * world_size + + def zero2_align(x): + return align_to * math.ceil(x / align_to) + + if debug: + print(f"original offset={offset}, avail_numel={avail_numel}") + + offset = zero2_align(offset) + avail_numel = zero2_align(avail_numel) + + if debug: + print(f"aligned offset={offset}, avail_numel={avail_numel}") + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero2_merge_frozen_params(state_dict, zero_model_states) + + _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def zero3_partitioned_param_info(unpartitioned_numel, world_size): + remainder = unpartitioned_numel % world_size + padding_numel = (world_size - remainder) if remainder else 0 + partitioned_numel = math.ceil(unpartitioned_numel / world_size) + return partitioned_numel, padding_numel + + +def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + if debug: + for i in range(world_size): + num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values()) + print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in zero_model_states[0].frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states) + state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape) + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + avail_numel = fp32_flat_groups[0].numel() * world_size + # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each + # param, re-consolidating each param, while dealing with padding if any + + # merge list of dicts, preserving order + param_shapes = {k: v for d in param_shapes for k, v in d.items()} + + if debug: + for i in range(world_size): + print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}") + + wanted_params = len(param_shapes) + wanted_numel = sum(shape.numel() for shape in param_shapes.values()) + # not asserting if there is a mismatch due to possible padding + avail_numel = fp32_flat_groups[0].numel() * world_size + print(f"Trainable params: Have {avail_numel} numels to process.") + print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + offset = 0 + total_numel = 0 + total_params = 0 + for name, shape in param_shapes.items(): + + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + total_params += 1 + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + # XXX: memory usage doubles here + state_dict[name] = torch.cat( + tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)), + 0).narrow(0, 0, unpartitioned_numel).view(shape) + offset += partitioned_numel + + offset *= world_size + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero3_merge_frozen_params(state_dict, world_size, zero_model_states) + + _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with + ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example + via a model hub. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + + Returns: + - pytorch ``state_dict`` + + Note: this approach may not work if your application doesn't have sufficient free CPU memory and + you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with + the checkpoint. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + # do the training and checkpoint saving + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu + model = model.cpu() # move to cpu + model.load_state_dict(state_dict) + # submit to model hub or save the model to share with others + + In this example the ``model`` will no longer be usable in the deepspeed context of the same + application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead. + + """ + if tag is None: + latest_path = os.path.join(checkpoint_dir, 'latest') + if os.path.isfile(latest_path): + with open(latest_path, 'r') as fd: + tag = fd.read().strip() + else: + raise ValueError(f"Unable to find 'latest' file at {latest_path}") + + ds_checkpoint_dir = os.path.join(checkpoint_dir, tag) + + if not os.path.isdir(ds_checkpoint_dir): + raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist") + + return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters) + + +def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be + loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + """ + + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters) + print(f"Saving fp32 state dict to {output_file}") + torch.save(state_dict, output_file) + + +def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None): + """ + 1. Put the provided model to cpu + 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` + 3. Load it into the provided model + + Args: + - ``model``: the model object to update + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + + Returns: + - ``model`: modified model + + Make sure you have plenty of CPU memory available before you call this function. If you don't + have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it + conveniently placed for you in the checkpoint folder. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint + model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir) + # submit to model hub or save the model to share with others + + Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context + of the same application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + """ + logger.info(f"Extracting fp32 weights") + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag) + + logger.info(f"Overwriting model with fp32 weights") + model = model.cpu() + model.load_state_dict(state_dict, strict=False) + + return model + + +if __name__ == "__main__": + + parser = argparse.ArgumentParser() + parser.add_argument("checkpoint_dir", + type=str, + help="path to the desired checkpoint folder, e.g., path/checkpoint-12") + parser.add_argument( + "output_file", + type=str, + help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)") + parser.add_argument("-t", + "--tag", + type=str, + default=None, + help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1") + parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters") + parser.add_argument("-d", "--debug", action='store_true', help="enable debug") + args = parser.parse_args() + + debug = args.debug + + convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, + args.output_file, + tag=args.tag, + exclude_frozen_parameters=args.exclude_frozen_parameters) diff --git a/checkpoint-1050/README.md b/checkpoint-1050/README.md new file mode 100644 index 0000000000000000000000000000000000000000..16b1eacdd9353dec380a08ee77ce6ed5ab50f12e --- /dev/null +++ b/checkpoint-1050/README.md @@ -0,0 +1,202 @@ +--- +library_name: peft +base_model: gotzmann/uni +--- + +# Model Card for Model ID + + + + + +## Model Details + +### Model Description + + + + + +- **Developed by:** [More Information Needed] +- **Funded by [optional]:** [More Information Needed] +- **Shared by [optional]:** [More Information Needed] +- **Model type:** [More Information Needed] +- **Language(s) (NLP):** [More Information Needed] +- **License:** [More Information Needed] +- **Finetuned from model [optional]:** [More Information Needed] + +### Model Sources [optional] + + + +- **Repository:** [More Information Needed] +- **Paper [optional]:** [More Information Needed] +- **Demo [optional]:** [More Information Needed] + +## Uses + + + +### Direct Use + + + +[More Information Needed] + +### Downstream Use [optional] + + + +[More Information Needed] + +### Out-of-Scope Use + + + +[More Information Needed] + +## Bias, Risks, and Limitations + + + +[More Information Needed] + +### Recommendations + + + +Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations. + +## How to Get Started with the Model + +Use the code below to get started with the model. + +[More Information Needed] + +## Training Details + +### Training Data + + + +[More Information Needed] + +### Training Procedure + + + +#### Preprocessing [optional] + +[More Information Needed] + + +#### Training Hyperparameters + +- **Training regime:** [More Information Needed] + +#### Speeds, Sizes, Times [optional] + + + +[More Information Needed] + +## Evaluation + + + +### Testing Data, Factors & Metrics + +#### Testing Data + + + +[More Information Needed] + +#### Factors + + + +[More Information Needed] + +#### Metrics + + + +[More Information Needed] + +### Results + +[More Information Needed] + +#### Summary + + + +## Model Examination [optional] + + + +[More Information Needed] + +## Environmental Impact + + + +Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700). + +- **Hardware Type:** [More Information Needed] +- **Hours used:** [More Information Needed] +- **Cloud Provider:** [More Information Needed] +- **Compute Region:** [More Information Needed] +- **Carbon Emitted:** [More Information Needed] + +## Technical Specifications [optional] + +### Model Architecture and Objective + +[More Information Needed] + +### Compute Infrastructure + +[More Information Needed] + +#### Hardware + +[More Information Needed] + +#### Software + +[More Information Needed] + +## Citation [optional] + + + +**BibTeX:** + +[More Information Needed] + +**APA:** + +[More Information Needed] + +## Glossary [optional] + + + +[More Information Needed] + +## More Information [optional] + +[More Information Needed] + +## Model Card Authors [optional] + +[More Information Needed] + +## Model Card Contact + +[More Information Needed] +### Framework versions + +- PEFT 0.10.0 \ No newline at end of file diff --git a/checkpoint-1050/adapter_config.json b/checkpoint-1050/adapter_config.json new file mode 100644 index 0000000000000000000000000000000000000000..e9287964d2e2ba3ecf1c52d1ac5d36e39722c816 --- /dev/null +++ b/checkpoint-1050/adapter_config.json @@ -0,0 +1,31 @@ +{ + "alpha_pattern": {}, + "auto_mapping": null, + "base_model_name_or_path": "gotzmann/uni", + "bias": "none", + "fan_in_fan_out": false, + "inference_mode": true, + "init_lora_weights": true, + "layer_replication": null, + "layers_pattern": null, + "layers_to_transform": null, + "loftq_config": {}, + "lora_alpha": 128, + "lora_dropout": 0.0, + "megatron_config": null, + "megatron_core": "megatron.core", + "modules_to_save": null, + "peft_type": "LORA", + "r": 128, + "rank_pattern": {}, + "revision": null, + "target_modules": [ + "q_proj", + "v_proj", + "o_proj", + "k_proj" + ], + "task_type": "CAUSAL_LM", + "use_dora": false, + "use_rslora": true +} \ No newline at end of file diff --git a/checkpoint-1050/adapter_model.safetensors b/checkpoint-1050/adapter_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..52a256b8635154e3e1c50914fb9731c141fd6e97 --- /dev/null +++ b/checkpoint-1050/adapter_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:74fa537a65ece7afb48497d0fc7176394772e474c2176bdf26eb2a12c5eb6133 +size 1048664848 diff --git a/checkpoint-1050/global_step1050/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt b/checkpoint-1050/global_step1050/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..f072516904992e6467b803eac8b74bf51da2113d --- /dev/null +++ b/checkpoint-1050/global_step1050/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:a5d9fa14d2e94404b03a22c39c78e592828c9a5097f79e10ba4542e903b7a1c5 +size 787270042 diff --git a/checkpoint-1050/global_step1050/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt b/checkpoint-1050/global_step1050/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..dc231d33088f1706c848d0925c9931e2af3ca505 --- /dev/null +++ b/checkpoint-1050/global_step1050/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:fba576d371c7e5e476dcd7036e9b13fccea09de427e33b695c9568a2be1f7cf5 +size 787270042 diff --git a/checkpoint-1050/global_step1050/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt b/checkpoint-1050/global_step1050/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..7af46958d6d2d4c9fab3c1a2844775c7d3e5fe91 --- /dev/null +++ b/checkpoint-1050/global_step1050/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:a1334c39b8090974ad0010c20339347ea540835715d396496ed7b1a12c65a9e1 +size 787270042 diff --git a/checkpoint-1050/global_step1050/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt b/checkpoint-1050/global_step1050/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..50b2f75e4a20bbedfd177ccb39c167299428e65b --- /dev/null +++ b/checkpoint-1050/global_step1050/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:90bf5319cb9f6c53b43afddef0851bd6b32baf2f58e3aa847dd87dc8f45d3376 +size 787270042 diff --git a/checkpoint-1050/global_step1050/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt b/checkpoint-1050/global_step1050/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..53356fdcb6b53724cd7e1df6126b087fbde6263c --- /dev/null +++ b/checkpoint-1050/global_step1050/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:199ef91ca33e98122ae5d9db72923ca01b0d645958c36e49a0a5cfda5db895e3 +size 787270042 diff --git a/checkpoint-1050/global_step1050/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt b/checkpoint-1050/global_step1050/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..27dc590d694566ef99411fdb57cfc4979250d40d --- /dev/null +++ b/checkpoint-1050/global_step1050/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:d994fb7db41608b2f3eced488c5cd4e9fb33d5a927664ab830e3bdb2ca39b711 +size 787270042 diff --git a/checkpoint-1050/global_step1050/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt b/checkpoint-1050/global_step1050/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..b1ec1ae28902cd9e9920ad9c4041b6db2ffb8a88 --- /dev/null +++ b/checkpoint-1050/global_step1050/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:99edcf3ebb349e86e46ec91ad1777bf3678b03b4761aff5c7aff143cb43c1a15 +size 787270042 diff --git a/checkpoint-1050/global_step1050/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt b/checkpoint-1050/global_step1050/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..c84a53f2a53e81bc3732e711354bd6edc3b22e72 --- /dev/null +++ b/checkpoint-1050/global_step1050/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:93429d616accac9d7fa2e84862008e4559616af4790d2826f1dd941869e8442b +size 787270042 diff --git a/checkpoint-1050/global_step1050/zero_pp_rank_0_mp_rank_00_model_states.pt b/checkpoint-1050/global_step1050/zero_pp_rank_0_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..9bb149842f0c52e742d84b64484866ee97d0db5b --- /dev/null +++ b/checkpoint-1050/global_step1050/zero_pp_rank_0_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4b3f51e27f57bf07ab75a22b3be7686c834716ac6bf1e42e58696786749e40eb +size 653742 diff --git a/checkpoint-1050/global_step1050/zero_pp_rank_1_mp_rank_00_model_states.pt b/checkpoint-1050/global_step1050/zero_pp_rank_1_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..1219ce5aa5f2d8540ecf3486a6ad722b4466a575 --- /dev/null +++ b/checkpoint-1050/global_step1050/zero_pp_rank_1_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4bc7d961b5a083a83f9206e13485480e5edd384db7a2eccd8f5de849a65b61f5 +size 653742 diff --git a/checkpoint-1050/global_step1050/zero_pp_rank_2_mp_rank_00_model_states.pt b/checkpoint-1050/global_step1050/zero_pp_rank_2_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..a1e506bf194d5bd563e7858c695a0dca57eb5b1a --- /dev/null +++ b/checkpoint-1050/global_step1050/zero_pp_rank_2_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:414a5f5d4247b734ac16b08d4c1e5659697300fde198d5056b0505887c25a2b8 +size 653742 diff --git a/checkpoint-1050/global_step1050/zero_pp_rank_3_mp_rank_00_model_states.pt b/checkpoint-1050/global_step1050/zero_pp_rank_3_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..1ecc7612d7f59e3d15454c22604efc92a13b066b --- /dev/null +++ b/checkpoint-1050/global_step1050/zero_pp_rank_3_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:be1b2039a29e8f23b7cad84b8eaf0024e18a2fbdcbc41fdb9a652e4b7e95bbba +size 653742 diff --git a/checkpoint-1050/global_step1050/zero_pp_rank_4_mp_rank_00_model_states.pt b/checkpoint-1050/global_step1050/zero_pp_rank_4_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..112eaa5d108be50233d04f7818f3ff710d29f49b --- /dev/null +++ b/checkpoint-1050/global_step1050/zero_pp_rank_4_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:58a942392d335b9d82ec508fea9e99a76ccf8a2ff510cb38c92c0b167fcf67fe +size 653742 diff --git a/checkpoint-1050/global_step1050/zero_pp_rank_5_mp_rank_00_model_states.pt b/checkpoint-1050/global_step1050/zero_pp_rank_5_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..714f18ffa311c637f8bdce3df1110cbf6384db5e --- /dev/null +++ b/checkpoint-1050/global_step1050/zero_pp_rank_5_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0a11536ae6ed3f56eb0983e22371db0076f2a0540b433142639548e46348cf36 +size 653742 diff --git a/checkpoint-1050/global_step1050/zero_pp_rank_6_mp_rank_00_model_states.pt b/checkpoint-1050/global_step1050/zero_pp_rank_6_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..982c2af2865dc7330afdf3a30411d22b48d449ec --- /dev/null +++ b/checkpoint-1050/global_step1050/zero_pp_rank_6_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c76a9d81f76ff90f486b98ee2f90924bfa83ac298d3b1a865d2945e241717a6b +size 653742 diff --git a/checkpoint-1050/global_step1050/zero_pp_rank_7_mp_rank_00_model_states.pt b/checkpoint-1050/global_step1050/zero_pp_rank_7_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..4660be82da4673bca83817646ebc658ec434cc74 --- /dev/null +++ b/checkpoint-1050/global_step1050/zero_pp_rank_7_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:31c89a7f8a58901e1ef71eb63cc914bb06f34c661e4657336851817a1cfa1ccb +size 653742 diff --git a/checkpoint-1050/latest b/checkpoint-1050/latest new file mode 100644 index 0000000000000000000000000000000000000000..9003e5f7e95704409b5d8f438ed7572043c8b9ad --- /dev/null +++ b/checkpoint-1050/latest @@ -0,0 +1 @@ +global_step1050 \ No newline at end of file diff --git a/checkpoint-1050/rng_state_0.pth b/checkpoint-1050/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..6a74f25da28f01a2e6b66587824ee5f5cc9be737 --- /dev/null +++ b/checkpoint-1050/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3ee195ebde9bf012f945f068f133e7fe22fef5450c496607e3ef11cc2034a186 +size 15984 diff --git a/checkpoint-1050/rng_state_1.pth b/checkpoint-1050/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..f44ddc47315653477728c971b4ea191a3df8b92c --- /dev/null +++ b/checkpoint-1050/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bf0fe1a3315d60b197207c5cb249d0ce4f9ce6d7585e696276d9ffbcb5379893 +size 15984 diff --git a/checkpoint-1050/rng_state_2.pth b/checkpoint-1050/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..04636b9eca6484a4339eaa1e3acdf15d42d493b3 --- /dev/null +++ b/checkpoint-1050/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:01c5bd6eae04542162b3e94245555bd81312524066bc01d0ebbfc4fd8554240e +size 15984 diff --git a/checkpoint-1050/rng_state_3.pth b/checkpoint-1050/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..05435e407541728c3159054a4beb6705039a8ddf --- /dev/null +++ b/checkpoint-1050/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:45b74942c68b00d657cfce186b0eeb4aa8f52efa04b114803b605fee8de45972 +size 15984 diff --git a/checkpoint-1050/rng_state_4.pth b/checkpoint-1050/rng_state_4.pth new file mode 100644 index 0000000000000000000000000000000000000000..94fdf5f2c3e5df27424e6482bf52255531147a23 --- /dev/null +++ b/checkpoint-1050/rng_state_4.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0cd66dd2ba958fc9929441817d8154abbd929c0aa9cd66ff3171965bdaaf5d78 +size 15984 diff --git a/checkpoint-1050/rng_state_5.pth b/checkpoint-1050/rng_state_5.pth new file mode 100644 index 0000000000000000000000000000000000000000..da6e37fc011d97a1512e1e746bdd410a738c018a --- /dev/null +++ b/checkpoint-1050/rng_state_5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:89eeedefdd62514d0130acc330a5c08e9774c95d38c60997905cfd65fc54b710 +size 15984 diff --git a/checkpoint-1050/rng_state_6.pth b/checkpoint-1050/rng_state_6.pth new file mode 100644 index 0000000000000000000000000000000000000000..751fd85c617e15dee9713bc0f0c533af5bd18c8e --- /dev/null +++ b/checkpoint-1050/rng_state_6.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f43ced939100082608f57561a10e1888e69210c80675068db530c5815889910e +size 15984 diff --git a/checkpoint-1050/rng_state_7.pth b/checkpoint-1050/rng_state_7.pth new file mode 100644 index 0000000000000000000000000000000000000000..4aacf54fa8285b7e199a7cd62f1ee3d8b9beb5e5 --- /dev/null +++ b/checkpoint-1050/rng_state_7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0d8d6ee244d99525e7004ae3f02d44ae63082d81fbbab7306f641ac6aeeb736f +size 15984 diff --git a/checkpoint-1050/scheduler.pt b/checkpoint-1050/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..706e20a94a7e2991f07bfa5aa44e2d6c47b72da7 --- /dev/null +++ b/checkpoint-1050/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:97feef438f4431a4ad6f82ff4896fbb93af122cfbe57a1b829aa9b6b13da43a5 +size 1064 diff --git a/checkpoint-1050/special_tokens_map.json b/checkpoint-1050/special_tokens_map.json new file mode 100644 index 0000000000000000000000000000000000000000..72ecfeeb7e14d244c936169d2ed139eeae235ef1 --- /dev/null +++ b/checkpoint-1050/special_tokens_map.json @@ -0,0 +1,24 @@ +{ + "bos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "eos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "pad_token": "", + "unk_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + } +} diff --git a/checkpoint-1050/tokenizer.model b/checkpoint-1050/tokenizer.model new file mode 100644 index 0000000000000000000000000000000000000000..6c00c742ce03c627d6cd5b795984876fa49fa899 --- /dev/null +++ b/checkpoint-1050/tokenizer.model @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347 +size 499723 diff --git a/checkpoint-1050/tokenizer_config.json b/checkpoint-1050/tokenizer_config.json new file mode 100644 index 0000000000000000000000000000000000000000..bb5a9f09d8c0f3c32c66fc6118fe5c76c5c6fd90 --- /dev/null +++ b/checkpoint-1050/tokenizer_config.json @@ -0,0 +1,45 @@ +{ + "add_bos_token": true, + "add_eos_token": false, + "add_prefix_space": false, + "added_tokens_decoder": { + "0": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "1": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "2": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + } + }, + "bos_token": "", + "chat_template": "{% if messages[0]['role'] == 'system' %}{% set system_message = messages[0]['content'] %}{% endif %}{% if system_message is defined %}{{ '' + '### System:\\n\\n' + system_message }}{% endif %}{% for message in messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ '\\n\\n### Human:\\n\\n' + content }}{% elif message['role'] == 'assistant' %}{{ '\\n\\n### Assistant:\\n\\n' + content + '' }}{% endif %}{% endfor %}", + "clean_up_tokenization_spaces": false, + "eos_token": "", + "legacy": false, + "model_max_length": 1000000000000000019884624838656, + "pad_token": "", + "padding_side": "right", + "sp_model_kwargs": {}, + "spaces_between_special_tokens": false, + "split_special_tokens": false, + "tokenizer_class": "LlamaTokenizer", + "unk_token": "", + "use_default_system_prompt": false +} diff --git a/checkpoint-1050/trainer_state.json b/checkpoint-1050/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..b72a74def5711737305b1858d0876290508e5a13 --- /dev/null +++ b/checkpoint-1050/trainer_state.json @@ -0,0 +1,7371 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 1.0914494741655236, + "eval_steps": 500, + "global_step": 1050, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "grad_norm": 0.849355824164473, + "learning_rate": 4.878048780487805e-07, + "loss": 1.3655, + "step": 1 + }, + { + "epoch": 0.0, + "grad_norm": 10.01567518957158, + "learning_rate": 9.75609756097561e-07, + "loss": 1.5767, + "step": 2 + }, + { + "epoch": 0.01, + "grad_norm": 0.6466000875559635, + "learning_rate": 1.4634146341463414e-06, + "loss": 1.3913, + "step": 3 + }, + { + "epoch": 0.01, + "grad_norm": 0.6644565932010504, + "learning_rate": 1.951219512195122e-06, + "loss": 1.3218, + "step": 4 + }, + { + "epoch": 0.01, + "grad_norm": 0.571354207588475, + "learning_rate": 2.4390243902439027e-06, + "loss": 1.3597, + "step": 5 + }, + { + "epoch": 0.01, + "grad_norm": 0.31036262839244955, + "learning_rate": 2.926829268292683e-06, + "loss": 1.2832, + "step": 6 + }, + { + "epoch": 0.01, + "grad_norm": 0.2622135027188184, + "learning_rate": 3.414634146341464e-06, + "loss": 1.2161, + "step": 7 + }, + { + "epoch": 0.01, + "grad_norm": 0.296824630261661, + "learning_rate": 3.902439024390244e-06, + "loss": 1.2985, + "step": 8 + }, + { + "epoch": 0.02, + "grad_norm": 0.2557267467361569, + "learning_rate": 4.390243902439025e-06, + "loss": 1.3175, + "step": 9 + }, + { + "epoch": 0.02, + "grad_norm": 0.23418939513890769, + "learning_rate": 4.8780487804878055e-06, + "loss": 1.2617, + "step": 10 + }, + { + "epoch": 0.02, + "grad_norm": 0.2364760983285843, + "learning_rate": 5.365853658536586e-06, + "loss": 1.3103, + "step": 11 + }, + { + "epoch": 0.02, + "grad_norm": 0.23893034721889, + "learning_rate": 5.853658536585366e-06, + "loss": 1.2405, + "step": 12 + }, + { + "epoch": 0.02, + "grad_norm": 0.25563593295485887, + "learning_rate": 6.341463414634147e-06, + "loss": 1.2831, + "step": 13 + }, + { + "epoch": 0.03, + "grad_norm": 0.23239975352661665, + "learning_rate": 6.829268292682928e-06, + "loss": 1.3125, + "step": 14 + }, + { + "epoch": 0.03, + "grad_norm": 0.3092813858209507, + "learning_rate": 7.317073170731707e-06, + "loss": 1.2422, + "step": 15 + }, + { + "epoch": 0.03, + "grad_norm": 0.282563380367434, + "learning_rate": 7.804878048780489e-06, + "loss": 1.2453, + "step": 16 + }, + { + "epoch": 0.03, + "grad_norm": 0.22065680088315018, + "learning_rate": 8.292682926829268e-06, + "loss": 1.2491, + "step": 17 + }, + { + "epoch": 0.03, + "grad_norm": 0.22777800877980184, + "learning_rate": 8.78048780487805e-06, + "loss": 1.2655, + "step": 18 + }, + { + "epoch": 0.03, + "grad_norm": 0.22145212540177928, + "learning_rate": 9.268292682926831e-06, + "loss": 1.2413, + "step": 19 + }, + { + "epoch": 0.04, + "grad_norm": 0.22482351883112714, + "learning_rate": 9.756097560975611e-06, + "loss": 1.2653, + "step": 20 + }, + { + "epoch": 0.04, + "grad_norm": 0.20823080508385733, + "learning_rate": 1.024390243902439e-05, + "loss": 1.2374, + "step": 21 + }, + { + "epoch": 0.04, + "grad_norm": 0.26025492562935737, + "learning_rate": 1.0731707317073172e-05, + "loss": 1.2065, + "step": 22 + }, + { + "epoch": 0.04, + "grad_norm": 0.2150252124176173, + "learning_rate": 1.1219512195121953e-05, + "loss": 1.2782, + "step": 23 + }, + { + "epoch": 0.04, + "grad_norm": 0.2505915177425618, + "learning_rate": 1.1707317073170731e-05, + "loss": 1.2742, + "step": 24 + }, + { + "epoch": 0.05, + "grad_norm": 0.20129223044786942, + "learning_rate": 1.2195121951219513e-05, + "loss": 1.3366, + "step": 25 + }, + { + "epoch": 0.05, + "grad_norm": 0.1973508510397107, + "learning_rate": 1.2682926829268294e-05, + "loss": 1.2476, + "step": 26 + }, + { + "epoch": 0.05, + "grad_norm": 0.27103325392437194, + "learning_rate": 1.3170731707317076e-05, + "loss": 1.2325, + "step": 27 + }, + { + "epoch": 0.05, + "grad_norm": 0.17954976411006285, + "learning_rate": 1.3658536585365855e-05, + "loss": 1.2523, + "step": 28 + }, + { + "epoch": 0.05, + "grad_norm": 0.22216997851088888, + "learning_rate": 1.4146341463414635e-05, + "loss": 1.3297, + "step": 29 + }, + { + "epoch": 0.05, + "grad_norm": 0.2071458864548587, + "learning_rate": 1.4634146341463415e-05, + "loss": 1.2127, + "step": 30 + }, + { + "epoch": 0.06, + "grad_norm": 0.18039422081622164, + "learning_rate": 1.5121951219512196e-05, + "loss": 1.2509, + "step": 31 + }, + { + "epoch": 0.06, + "grad_norm": 0.18631254372974412, + "learning_rate": 1.5609756097560978e-05, + "loss": 1.2247, + "step": 32 + }, + { + "epoch": 0.06, + "grad_norm": 0.18843872523649827, + "learning_rate": 1.6097560975609757e-05, + "loss": 1.195, + "step": 33 + }, + { + "epoch": 0.06, + "grad_norm": 0.2163847267778325, + "learning_rate": 1.6585365853658537e-05, + "loss": 1.2179, + "step": 34 + }, + { + "epoch": 0.06, + "grad_norm": 0.19687688475496104, + "learning_rate": 1.7073170731707317e-05, + "loss": 1.2763, + "step": 35 + }, + { + "epoch": 0.07, + "grad_norm": 0.20409643064887947, + "learning_rate": 1.75609756097561e-05, + "loss": 1.253, + "step": 36 + }, + { + "epoch": 0.07, + "grad_norm": 0.1879182661759335, + "learning_rate": 1.804878048780488e-05, + "loss": 1.2586, + "step": 37 + }, + { + "epoch": 0.07, + "grad_norm": 0.19400648948514373, + "learning_rate": 1.8536585365853663e-05, + "loss": 1.2154, + "step": 38 + }, + { + "epoch": 0.07, + "grad_norm": 0.1878879343148452, + "learning_rate": 1.902439024390244e-05, + "loss": 1.2304, + "step": 39 + }, + { + "epoch": 0.07, + "grad_norm": 0.17687475469924052, + "learning_rate": 1.9512195121951222e-05, + "loss": 1.2351, + "step": 40 + }, + { + "epoch": 0.07, + "grad_norm": 0.18223935625384885, + "learning_rate": 2e-05, + "loss": 1.2222, + "step": 41 + }, + { + "epoch": 0.08, + "grad_norm": 0.1943061629408338, + "learning_rate": 2.048780487804878e-05, + "loss": 1.2044, + "step": 42 + }, + { + "epoch": 0.08, + "grad_norm": 0.17027514338700078, + "learning_rate": 2.0975609756097564e-05, + "loss": 1.1548, + "step": 43 + }, + { + "epoch": 0.08, + "grad_norm": 0.18553769630586192, + "learning_rate": 2.1463414634146344e-05, + "loss": 1.2721, + "step": 44 + }, + { + "epoch": 0.08, + "grad_norm": 0.19732826914228765, + "learning_rate": 2.1951219512195124e-05, + "loss": 1.3097, + "step": 45 + }, + { + "epoch": 0.08, + "grad_norm": 0.18714230986631472, + "learning_rate": 2.2439024390243907e-05, + "loss": 1.2662, + "step": 46 + }, + { + "epoch": 0.09, + "grad_norm": 0.19988987568002223, + "learning_rate": 2.2926829268292683e-05, + "loss": 1.2904, + "step": 47 + }, + { + "epoch": 0.09, + "grad_norm": 0.17744650133390918, + "learning_rate": 2.3414634146341463e-05, + "loss": 1.1825, + "step": 48 + }, + { + "epoch": 0.09, + "grad_norm": 0.16576734763834533, + "learning_rate": 2.3902439024390246e-05, + "loss": 1.1858, + "step": 49 + }, + { + "epoch": 0.09, + "grad_norm": 0.179591794065527, + "learning_rate": 2.4390243902439026e-05, + "loss": 1.2711, + "step": 50 + }, + { + "epoch": 0.09, + "grad_norm": 0.17923464471176911, + "learning_rate": 2.4878048780487805e-05, + "loss": 1.2289, + "step": 51 + }, + { + "epoch": 0.1, + "grad_norm": 0.18991742907836837, + "learning_rate": 2.536585365853659e-05, + "loss": 1.3097, + "step": 52 + }, + { + "epoch": 0.1, + "grad_norm": 0.19849796137254636, + "learning_rate": 2.5853658536585368e-05, + "loss": 1.2489, + "step": 53 + }, + { + "epoch": 0.1, + "grad_norm": 0.17452371110976383, + "learning_rate": 2.634146341463415e-05, + "loss": 1.2461, + "step": 54 + }, + { + "epoch": 0.1, + "grad_norm": 0.17671022353085036, + "learning_rate": 2.682926829268293e-05, + "loss": 1.153, + "step": 55 + }, + { + "epoch": 0.1, + "grad_norm": 0.36820559192096686, + "learning_rate": 2.731707317073171e-05, + "loss": 1.2431, + "step": 56 + }, + { + "epoch": 0.1, + "grad_norm": 0.20331468526494198, + "learning_rate": 2.7804878048780487e-05, + "loss": 1.2575, + "step": 57 + }, + { + "epoch": 0.11, + "grad_norm": 0.2402486598118377, + "learning_rate": 2.829268292682927e-05, + "loss": 1.2538, + "step": 58 + }, + { + "epoch": 0.11, + "grad_norm": 0.2549409484173144, + "learning_rate": 2.878048780487805e-05, + "loss": 1.2065, + "step": 59 + }, + { + "epoch": 0.11, + "grad_norm": 0.2053105349872685, + "learning_rate": 2.926829268292683e-05, + "loss": 1.2094, + "step": 60 + }, + { + "epoch": 0.11, + "grad_norm": 0.17971910872957886, + "learning_rate": 2.9756097560975613e-05, + "loss": 1.228, + "step": 61 + }, + { + "epoch": 0.11, + "grad_norm": 0.1885853654992973, + "learning_rate": 3.0243902439024392e-05, + "loss": 1.2286, + "step": 62 + }, + { + "epoch": 0.12, + "grad_norm": 0.1848524571968613, + "learning_rate": 3.073170731707317e-05, + "loss": 1.2718, + "step": 63 + }, + { + "epoch": 0.12, + "grad_norm": 0.18734105883548513, + "learning_rate": 3.1219512195121955e-05, + "loss": 1.2357, + "step": 64 + }, + { + "epoch": 0.12, + "grad_norm": 0.17774668052121825, + "learning_rate": 3.170731707317074e-05, + "loss": 1.1509, + "step": 65 + }, + { + "epoch": 0.12, + "grad_norm": 0.17890968008080646, + "learning_rate": 3.2195121951219514e-05, + "loss": 1.1924, + "step": 66 + }, + { + "epoch": 0.12, + "grad_norm": 0.18249273371332375, + "learning_rate": 3.268292682926829e-05, + "loss": 1.2545, + "step": 67 + }, + { + "epoch": 0.12, + "grad_norm": 0.21064122671902577, + "learning_rate": 3.3170731707317074e-05, + "loss": 1.2832, + "step": 68 + }, + { + "epoch": 0.13, + "grad_norm": 0.1820064171955093, + "learning_rate": 3.365853658536586e-05, + "loss": 1.2071, + "step": 69 + }, + { + "epoch": 0.13, + "grad_norm": 0.16996662800553433, + "learning_rate": 3.414634146341463e-05, + "loss": 1.2073, + "step": 70 + }, + { + "epoch": 0.13, + "grad_norm": 0.1618669302922445, + "learning_rate": 3.4634146341463416e-05, + "loss": 1.1289, + "step": 71 + }, + { + "epoch": 0.13, + "grad_norm": 0.18948744950985544, + "learning_rate": 3.51219512195122e-05, + "loss": 1.2915, + "step": 72 + }, + { + "epoch": 0.13, + "grad_norm": 0.18326143691603383, + "learning_rate": 3.5609756097560976e-05, + "loss": 1.2238, + "step": 73 + }, + { + "epoch": 0.14, + "grad_norm": 0.17410704510700503, + "learning_rate": 3.609756097560976e-05, + "loss": 1.1784, + "step": 74 + }, + { + "epoch": 0.14, + "grad_norm": 0.1983667344995625, + "learning_rate": 3.658536585365854e-05, + "loss": 1.2452, + "step": 75 + }, + { + "epoch": 0.14, + "grad_norm": 0.3416310763369357, + "learning_rate": 3.7073170731707325e-05, + "loss": 1.1972, + "step": 76 + }, + { + "epoch": 0.14, + "grad_norm": 0.2776466983511955, + "learning_rate": 3.75609756097561e-05, + "loss": 1.3121, + "step": 77 + }, + { + "epoch": 0.14, + "grad_norm": 0.20026129636576834, + "learning_rate": 3.804878048780488e-05, + "loss": 1.2436, + "step": 78 + }, + { + "epoch": 0.14, + "grad_norm": 0.21064549243917835, + "learning_rate": 3.853658536585366e-05, + "loss": 1.2064, + "step": 79 + }, + { + "epoch": 0.15, + "grad_norm": 0.22119482175714267, + "learning_rate": 3.9024390243902444e-05, + "loss": 1.2715, + "step": 80 + }, + { + "epoch": 0.15, + "grad_norm": 0.23047133748844142, + "learning_rate": 3.951219512195122e-05, + "loss": 1.2888, + "step": 81 + }, + { + "epoch": 0.15, + "grad_norm": 0.18741863156973176, + "learning_rate": 4e-05, + "loss": 1.248, + "step": 82 + }, + { + "epoch": 0.15, + "grad_norm": 0.1747859810629604, + "learning_rate": 4.0487804878048786e-05, + "loss": 1.1683, + "step": 83 + }, + { + "epoch": 0.15, + "grad_norm": 0.1896944798413341, + "learning_rate": 4.097560975609756e-05, + "loss": 1.2155, + "step": 84 + }, + { + "epoch": 0.16, + "grad_norm": 0.18724128114363303, + "learning_rate": 4.1463414634146346e-05, + "loss": 1.2273, + "step": 85 + }, + { + "epoch": 0.16, + "grad_norm": 0.17368125504855478, + "learning_rate": 4.195121951219513e-05, + "loss": 1.224, + "step": 86 + }, + { + "epoch": 0.16, + "grad_norm": 0.18371141013625703, + "learning_rate": 4.2439024390243905e-05, + "loss": 1.2294, + "step": 87 + }, + { + "epoch": 0.16, + "grad_norm": 0.1791029365673714, + "learning_rate": 4.292682926829269e-05, + "loss": 1.2895, + "step": 88 + }, + { + "epoch": 0.16, + "grad_norm": 0.20259974283859655, + "learning_rate": 4.341463414634147e-05, + "loss": 1.1841, + "step": 89 + }, + { + "epoch": 0.16, + "grad_norm": 0.17457456183272174, + "learning_rate": 4.390243902439025e-05, + "loss": 1.2357, + "step": 90 + }, + { + "epoch": 0.17, + "grad_norm": 0.1815824380789748, + "learning_rate": 4.439024390243903e-05, + "loss": 1.2304, + "step": 91 + }, + { + "epoch": 0.17, + "grad_norm": 0.17566480599583392, + "learning_rate": 4.4878048780487814e-05, + "loss": 1.242, + "step": 92 + }, + { + "epoch": 0.17, + "grad_norm": 0.18422975005984474, + "learning_rate": 4.536585365853658e-05, + "loss": 1.2177, + "step": 93 + }, + { + "epoch": 0.17, + "grad_norm": 0.16796781877940678, + "learning_rate": 4.5853658536585366e-05, + "loss": 1.1482, + "step": 94 + }, + { + "epoch": 0.17, + "grad_norm": 0.18636131653783305, + "learning_rate": 4.634146341463415e-05, + "loss": 1.1758, + "step": 95 + }, + { + "epoch": 0.18, + "grad_norm": 0.1823665700289814, + "learning_rate": 4.6829268292682926e-05, + "loss": 1.289, + "step": 96 + }, + { + "epoch": 0.18, + "grad_norm": 0.1719900691262439, + "learning_rate": 4.731707317073171e-05, + "loss": 1.1626, + "step": 97 + }, + { + "epoch": 0.18, + "grad_norm": 0.17937994168039778, + "learning_rate": 4.780487804878049e-05, + "loss": 1.175, + "step": 98 + }, + { + "epoch": 0.18, + "grad_norm": 0.16631851422106986, + "learning_rate": 4.829268292682927e-05, + "loss": 1.2177, + "step": 99 + }, + { + "epoch": 0.18, + "grad_norm": 0.19143696232800309, + "learning_rate": 4.878048780487805e-05, + "loss": 1.3071, + "step": 100 + }, + { + "epoch": 0.18, + "grad_norm": 0.17859506638780318, + "learning_rate": 4.9268292682926835e-05, + "loss": 1.2351, + "step": 101 + }, + { + "epoch": 0.19, + "grad_norm": 0.18381520321248196, + "learning_rate": 4.975609756097561e-05, + "loss": 1.2342, + "step": 102 + }, + { + "epoch": 0.19, + "grad_norm": 0.17968218683773912, + "learning_rate": 5.0243902439024394e-05, + "loss": 1.2074, + "step": 103 + }, + { + "epoch": 0.19, + "grad_norm": 0.18139489969339018, + "learning_rate": 5.073170731707318e-05, + "loss": 1.1558, + "step": 104 + }, + { + "epoch": 0.19, + "grad_norm": 0.17366624842514394, + "learning_rate": 5.121951219512195e-05, + "loss": 1.1897, + "step": 105 + }, + { + "epoch": 0.19, + "grad_norm": 0.16034845455223745, + "learning_rate": 5.1707317073170736e-05, + "loss": 1.179, + "step": 106 + }, + { + "epoch": 0.2, + "grad_norm": 0.17583069577827776, + "learning_rate": 5.219512195121952e-05, + "loss": 1.1856, + "step": 107 + }, + { + "epoch": 0.2, + "grad_norm": 0.1853758076989552, + "learning_rate": 5.26829268292683e-05, + "loss": 1.2072, + "step": 108 + }, + { + "epoch": 0.2, + "grad_norm": 0.19597443965936462, + "learning_rate": 5.317073170731708e-05, + "loss": 1.2271, + "step": 109 + }, + { + "epoch": 0.2, + "grad_norm": 0.1899206334098331, + "learning_rate": 5.365853658536586e-05, + "loss": 1.1961, + "step": 110 + }, + { + "epoch": 0.2, + "grad_norm": 0.17463763837757018, + "learning_rate": 5.4146341463414645e-05, + "loss": 1.2049, + "step": 111 + }, + { + "epoch": 0.2, + "grad_norm": 0.20431371701229986, + "learning_rate": 5.463414634146342e-05, + "loss": 1.2891, + "step": 112 + }, + { + "epoch": 0.21, + "grad_norm": 0.1814475107638498, + "learning_rate": 5.51219512195122e-05, + "loss": 1.2346, + "step": 113 + }, + { + "epoch": 0.21, + "grad_norm": 0.1883849423207823, + "learning_rate": 5.5609756097560974e-05, + "loss": 1.244, + "step": 114 + }, + { + "epoch": 0.21, + "grad_norm": 0.1857258128640568, + "learning_rate": 5.609756097560976e-05, + "loss": 1.2669, + "step": 115 + }, + { + "epoch": 0.21, + "grad_norm": 0.1740768514118401, + "learning_rate": 5.658536585365854e-05, + "loss": 1.2414, + "step": 116 + }, + { + "epoch": 0.21, + "grad_norm": 0.1919320335584178, + "learning_rate": 5.7073170731707317e-05, + "loss": 1.2886, + "step": 117 + }, + { + "epoch": 0.22, + "grad_norm": 0.18288775167828136, + "learning_rate": 5.75609756097561e-05, + "loss": 1.1875, + "step": 118 + }, + { + "epoch": 0.22, + "grad_norm": 0.18208588867750863, + "learning_rate": 5.804878048780488e-05, + "loss": 1.2388, + "step": 119 + }, + { + "epoch": 0.22, + "grad_norm": 0.1743260015658331, + "learning_rate": 5.853658536585366e-05, + "loss": 1.1762, + "step": 120 + }, + { + "epoch": 0.22, + "grad_norm": 0.17856046291517946, + "learning_rate": 5.902439024390244e-05, + "loss": 1.2888, + "step": 121 + }, + { + "epoch": 0.22, + "grad_norm": 0.17493794870966536, + "learning_rate": 5.9512195121951225e-05, + "loss": 1.2222, + "step": 122 + }, + { + "epoch": 0.22, + "grad_norm": 0.1909202655203384, + "learning_rate": 6.000000000000001e-05, + "loss": 1.2414, + "step": 123 + }, + { + "epoch": 0.23, + "grad_norm": 0.18345819482834988, + "learning_rate": 6.0487804878048785e-05, + "loss": 1.2756, + "step": 124 + }, + { + "epoch": 0.23, + "grad_norm": 0.2057069352956621, + "learning_rate": 6.097560975609757e-05, + "loss": 1.261, + "step": 125 + }, + { + "epoch": 0.23, + "grad_norm": 0.299775882469108, + "learning_rate": 6.146341463414634e-05, + "loss": 1.2566, + "step": 126 + }, + { + "epoch": 0.23, + "grad_norm": 0.1869687633018095, + "learning_rate": 6.195121951219513e-05, + "loss": 1.3039, + "step": 127 + }, + { + "epoch": 0.23, + "grad_norm": 0.17747149926197442, + "learning_rate": 6.243902439024391e-05, + "loss": 1.2524, + "step": 128 + }, + { + "epoch": 0.24, + "grad_norm": 0.17885157788044242, + "learning_rate": 6.29268292682927e-05, + "loss": 1.2455, + "step": 129 + }, + { + "epoch": 0.24, + "grad_norm": 0.17617298187845123, + "learning_rate": 6.341463414634148e-05, + "loss": 1.2009, + "step": 130 + }, + { + "epoch": 0.24, + "grad_norm": 0.20164176323497066, + "learning_rate": 6.390243902439025e-05, + "loss": 1.2634, + "step": 131 + }, + { + "epoch": 0.24, + "grad_norm": 0.20459903417307612, + "learning_rate": 6.439024390243903e-05, + "loss": 1.1963, + "step": 132 + }, + { + "epoch": 0.24, + "grad_norm": 0.1863755486334296, + "learning_rate": 6.487804878048781e-05, + "loss": 1.2387, + "step": 133 + }, + { + "epoch": 0.25, + "grad_norm": 0.19265866140295207, + "learning_rate": 6.536585365853658e-05, + "loss": 1.2688, + "step": 134 + }, + { + "epoch": 0.25, + "grad_norm": 0.1823425868969493, + "learning_rate": 6.585365853658536e-05, + "loss": 1.2041, + "step": 135 + }, + { + "epoch": 0.25, + "grad_norm": 0.2016853266472781, + "learning_rate": 6.634146341463415e-05, + "loss": 1.1223, + "step": 136 + }, + { + "epoch": 0.25, + "grad_norm": 0.17282675192463448, + "learning_rate": 6.682926829268293e-05, + "loss": 1.1879, + "step": 137 + }, + { + "epoch": 0.25, + "grad_norm": 0.17398811693399288, + "learning_rate": 6.731707317073171e-05, + "loss": 1.2682, + "step": 138 + }, + { + "epoch": 0.25, + "grad_norm": 0.18516916965434696, + "learning_rate": 6.78048780487805e-05, + "loss": 1.1666, + "step": 139 + }, + { + "epoch": 0.26, + "grad_norm": 0.1852213129647933, + "learning_rate": 6.829268292682927e-05, + "loss": 1.2501, + "step": 140 + }, + { + "epoch": 0.26, + "grad_norm": 0.17915948766591883, + "learning_rate": 6.878048780487805e-05, + "loss": 1.2264, + "step": 141 + }, + { + "epoch": 0.26, + "grad_norm": 0.21599939417233183, + "learning_rate": 6.926829268292683e-05, + "loss": 1.2376, + "step": 142 + }, + { + "epoch": 0.26, + "grad_norm": 0.17839304459521851, + "learning_rate": 6.975609756097562e-05, + "loss": 1.2353, + "step": 143 + }, + { + "epoch": 0.26, + "grad_norm": 0.20826913231380875, + "learning_rate": 7.02439024390244e-05, + "loss": 1.1901, + "step": 144 + }, + { + "epoch": 0.27, + "grad_norm": 0.20788894913361589, + "learning_rate": 7.073170731707318e-05, + "loss": 1.2577, + "step": 145 + }, + { + "epoch": 0.27, + "grad_norm": 0.18420055842301297, + "learning_rate": 7.121951219512195e-05, + "loss": 1.1393, + "step": 146 + }, + { + "epoch": 0.27, + "grad_norm": 0.19903048468685589, + "learning_rate": 7.170731707317073e-05, + "loss": 1.2321, + "step": 147 + }, + { + "epoch": 0.27, + "grad_norm": 0.19074116314985748, + "learning_rate": 7.219512195121952e-05, + "loss": 1.1912, + "step": 148 + }, + { + "epoch": 0.27, + "grad_norm": 0.2353816469403903, + "learning_rate": 7.26829268292683e-05, + "loss": 1.28, + "step": 149 + }, + { + "epoch": 0.27, + "grad_norm": 0.21634875684769345, + "learning_rate": 7.317073170731708e-05, + "loss": 1.3312, + "step": 150 + }, + { + "epoch": 0.28, + "grad_norm": 0.18290969006743918, + "learning_rate": 7.365853658536587e-05, + "loss": 1.2214, + "step": 151 + }, + { + "epoch": 0.28, + "grad_norm": 0.18484243897545208, + "learning_rate": 7.414634146341465e-05, + "loss": 1.1895, + "step": 152 + }, + { + "epoch": 0.28, + "grad_norm": 0.21882343112978872, + "learning_rate": 7.463414634146342e-05, + "loss": 1.2219, + "step": 153 + }, + { + "epoch": 0.28, + "grad_norm": 0.19868284379241205, + "learning_rate": 7.51219512195122e-05, + "loss": 1.2176, + "step": 154 + }, + { + "epoch": 0.28, + "grad_norm": 0.20912516312950613, + "learning_rate": 7.560975609756097e-05, + "loss": 1.242, + "step": 155 + }, + { + "epoch": 0.29, + "grad_norm": 0.23811880045549916, + "learning_rate": 7.609756097560976e-05, + "loss": 1.2838, + "step": 156 + }, + { + "epoch": 0.29, + "grad_norm": 0.19511077122033713, + "learning_rate": 7.658536585365854e-05, + "loss": 1.1594, + "step": 157 + }, + { + "epoch": 0.29, + "grad_norm": 0.20094129399534238, + "learning_rate": 7.707317073170732e-05, + "loss": 1.2966, + "step": 158 + }, + { + "epoch": 0.29, + "grad_norm": 0.19366245038292418, + "learning_rate": 7.75609756097561e-05, + "loss": 1.2246, + "step": 159 + }, + { + "epoch": 0.29, + "grad_norm": 0.19409570223867306, + "learning_rate": 7.804878048780489e-05, + "loss": 1.2312, + "step": 160 + }, + { + "epoch": 0.29, + "grad_norm": 0.2087258457033805, + "learning_rate": 7.853658536585366e-05, + "loss": 1.2169, + "step": 161 + }, + { + "epoch": 0.3, + "grad_norm": 0.18765223996270428, + "learning_rate": 7.902439024390244e-05, + "loss": 1.2383, + "step": 162 + }, + { + "epoch": 0.3, + "grad_norm": 0.20734180224147242, + "learning_rate": 7.951219512195122e-05, + "loss": 1.2587, + "step": 163 + }, + { + "epoch": 0.3, + "grad_norm": 0.24690929540287834, + "learning_rate": 8e-05, + "loss": 1.1951, + "step": 164 + }, + { + "epoch": 0.3, + "grad_norm": 0.2003538797619543, + "learning_rate": 7.999990914797545e-05, + "loss": 1.1982, + "step": 165 + }, + { + "epoch": 0.3, + "grad_norm": 0.22469075613510484, + "learning_rate": 7.99996365923145e-05, + "loss": 1.2355, + "step": 166 + }, + { + "epoch": 0.31, + "grad_norm": 0.21870100788336058, + "learning_rate": 7.999918233425526e-05, + "loss": 1.1103, + "step": 167 + }, + { + "epoch": 0.31, + "grad_norm": 0.20939989594131886, + "learning_rate": 7.999854637586122e-05, + "loss": 1.1966, + "step": 168 + }, + { + "epoch": 0.31, + "grad_norm": 0.43108211416237796, + "learning_rate": 7.999772872002132e-05, + "loss": 1.2882, + "step": 169 + }, + { + "epoch": 0.31, + "grad_norm": 0.27045413432174487, + "learning_rate": 7.999672937044984e-05, + "loss": 1.2399, + "step": 170 + }, + { + "epoch": 0.31, + "grad_norm": 0.19700483036740515, + "learning_rate": 7.999554833168642e-05, + "loss": 1.202, + "step": 171 + }, + { + "epoch": 0.31, + "grad_norm": 0.3335979493370708, + "learning_rate": 7.999418560909604e-05, + "loss": 1.1995, + "step": 172 + }, + { + "epoch": 0.32, + "grad_norm": 0.3165803974474567, + "learning_rate": 7.999264120886902e-05, + "loss": 1.1569, + "step": 173 + }, + { + "epoch": 0.32, + "grad_norm": 0.1951699080346223, + "learning_rate": 7.999091513802093e-05, + "loss": 1.1778, + "step": 174 + }, + { + "epoch": 0.32, + "grad_norm": 0.2087559121749787, + "learning_rate": 7.998900740439265e-05, + "loss": 1.1736, + "step": 175 + }, + { + "epoch": 0.32, + "grad_norm": 0.20345180977460478, + "learning_rate": 7.998691801665024e-05, + "loss": 1.2281, + "step": 176 + }, + { + "epoch": 0.32, + "grad_norm": 0.24617644827252333, + "learning_rate": 7.998464698428495e-05, + "loss": 1.2072, + "step": 177 + }, + { + "epoch": 0.33, + "grad_norm": 0.2469050959356265, + "learning_rate": 7.998219431761318e-05, + "loss": 1.2242, + "step": 178 + }, + { + "epoch": 0.33, + "grad_norm": 0.19529317748460623, + "learning_rate": 7.997956002777642e-05, + "loss": 1.2567, + "step": 179 + }, + { + "epoch": 0.33, + "grad_norm": 0.19048389491381376, + "learning_rate": 7.99767441267412e-05, + "loss": 1.2982, + "step": 180 + }, + { + "epoch": 0.33, + "grad_norm": 0.2085799116493225, + "learning_rate": 7.997374662729904e-05, + "loss": 1.1254, + "step": 181 + }, + { + "epoch": 0.33, + "grad_norm": 0.20636853256378995, + "learning_rate": 7.997056754306636e-05, + "loss": 1.2435, + "step": 182 + }, + { + "epoch": 0.33, + "grad_norm": 0.20590016382290252, + "learning_rate": 7.99672068884845e-05, + "loss": 1.2658, + "step": 183 + }, + { + "epoch": 0.34, + "grad_norm": 0.1931166169764433, + "learning_rate": 7.996366467881955e-05, + "loss": 1.1637, + "step": 184 + }, + { + "epoch": 0.34, + "grad_norm": 0.18873318157988098, + "learning_rate": 7.995994093016237e-05, + "loss": 1.1335, + "step": 185 + }, + { + "epoch": 0.34, + "grad_norm": 0.19210254625199108, + "learning_rate": 7.995603565942846e-05, + "loss": 1.1928, + "step": 186 + }, + { + "epoch": 0.34, + "grad_norm": 0.2130986479765664, + "learning_rate": 7.995194888435792e-05, + "loss": 1.2158, + "step": 187 + }, + { + "epoch": 0.34, + "grad_norm": 0.22003854501814088, + "learning_rate": 7.994768062351532e-05, + "loss": 1.2288, + "step": 188 + }, + { + "epoch": 0.35, + "grad_norm": 0.20330803191993058, + "learning_rate": 7.994323089628968e-05, + "loss": 1.2426, + "step": 189 + }, + { + "epoch": 0.35, + "grad_norm": 0.20567314642208634, + "learning_rate": 7.993859972289434e-05, + "loss": 1.2649, + "step": 190 + }, + { + "epoch": 0.35, + "grad_norm": 0.21556663727342962, + "learning_rate": 7.993378712436686e-05, + "loss": 1.2545, + "step": 191 + }, + { + "epoch": 0.35, + "grad_norm": 0.20309165469109888, + "learning_rate": 7.992879312256897e-05, + "loss": 1.3338, + "step": 192 + }, + { + "epoch": 0.35, + "grad_norm": 0.19574356669421325, + "learning_rate": 7.992361774018641e-05, + "loss": 1.278, + "step": 193 + }, + { + "epoch": 0.35, + "grad_norm": 0.2763613746722313, + "learning_rate": 7.991826100072891e-05, + "loss": 1.2571, + "step": 194 + }, + { + "epoch": 0.36, + "grad_norm": 0.19346552479915102, + "learning_rate": 7.991272292852996e-05, + "loss": 1.2027, + "step": 195 + }, + { + "epoch": 0.36, + "grad_norm": 0.2281167812123908, + "learning_rate": 7.990700354874683e-05, + "loss": 1.2586, + "step": 196 + }, + { + "epoch": 0.36, + "grad_norm": 0.19699013712137542, + "learning_rate": 7.990110288736042e-05, + "loss": 1.1371, + "step": 197 + }, + { + "epoch": 0.36, + "grad_norm": 0.21768209981475933, + "learning_rate": 7.989502097117503e-05, + "loss": 1.2522, + "step": 198 + }, + { + "epoch": 0.36, + "grad_norm": 0.21335427847754582, + "learning_rate": 7.988875782781838e-05, + "loss": 1.2437, + "step": 199 + }, + { + "epoch": 0.37, + "grad_norm": 0.21856710629066897, + "learning_rate": 7.988231348574147e-05, + "loss": 1.2135, + "step": 200 + }, + { + "epoch": 0.37, + "grad_norm": 0.20482062658774797, + "learning_rate": 7.987568797421836e-05, + "loss": 1.1755, + "step": 201 + }, + { + "epoch": 0.37, + "grad_norm": 0.2017756813960897, + "learning_rate": 7.986888132334608e-05, + "loss": 1.1699, + "step": 202 + }, + { + "epoch": 0.37, + "grad_norm": 0.20496443848153809, + "learning_rate": 7.986189356404458e-05, + "loss": 1.2125, + "step": 203 + }, + { + "epoch": 0.37, + "grad_norm": 0.2134603800558358, + "learning_rate": 7.985472472805643e-05, + "loss": 1.2391, + "step": 204 + }, + { + "epoch": 0.37, + "grad_norm": 0.2364175573420861, + "learning_rate": 7.98473748479468e-05, + "loss": 1.2384, + "step": 205 + }, + { + "epoch": 0.38, + "grad_norm": 0.1872419861598724, + "learning_rate": 7.983984395710326e-05, + "loss": 1.1457, + "step": 206 + }, + { + "epoch": 0.38, + "grad_norm": 0.28222194007095774, + "learning_rate": 7.983213208973566e-05, + "loss": 1.2952, + "step": 207 + }, + { + "epoch": 0.38, + "grad_norm": 0.1916094851162064, + "learning_rate": 7.982423928087593e-05, + "loss": 1.1763, + "step": 208 + }, + { + "epoch": 0.38, + "grad_norm": 0.18446245256166657, + "learning_rate": 7.981616556637795e-05, + "loss": 1.1863, + "step": 209 + }, + { + "epoch": 0.38, + "grad_norm": 0.195191961022491, + "learning_rate": 7.980791098291737e-05, + "loss": 1.2036, + "step": 210 + }, + { + "epoch": 0.39, + "grad_norm": 0.2652439657825496, + "learning_rate": 7.979947556799151e-05, + "loss": 1.2834, + "step": 211 + }, + { + "epoch": 0.39, + "grad_norm": 0.24308438957843412, + "learning_rate": 7.979085935991906e-05, + "loss": 1.234, + "step": 212 + }, + { + "epoch": 0.39, + "grad_norm": 0.21294701043622016, + "learning_rate": 7.978206239784004e-05, + "loss": 1.3006, + "step": 213 + }, + { + "epoch": 0.39, + "grad_norm": 0.25809277041859524, + "learning_rate": 7.977308472171553e-05, + "loss": 1.2272, + "step": 214 + }, + { + "epoch": 0.39, + "grad_norm": 0.193463860107294, + "learning_rate": 7.976392637232754e-05, + "loss": 1.2295, + "step": 215 + }, + { + "epoch": 0.4, + "grad_norm": 0.2150023760609626, + "learning_rate": 7.975458739127877e-05, + "loss": 1.2135, + "step": 216 + }, + { + "epoch": 0.4, + "grad_norm": 0.22590495955605894, + "learning_rate": 7.974506782099253e-05, + "loss": 1.2532, + "step": 217 + }, + { + "epoch": 0.4, + "grad_norm": 0.21023744668403702, + "learning_rate": 7.973536770471242e-05, + "loss": 1.2472, + "step": 218 + }, + { + "epoch": 0.4, + "grad_norm": 0.2345749799511543, + "learning_rate": 7.972548708650218e-05, + "loss": 1.1791, + "step": 219 + }, + { + "epoch": 0.4, + "grad_norm": 0.2158876734005217, + "learning_rate": 7.971542601124553e-05, + "loss": 1.2483, + "step": 220 + }, + { + "epoch": 0.4, + "grad_norm": 0.29455339949432446, + "learning_rate": 7.970518452464593e-05, + "loss": 1.2894, + "step": 221 + }, + { + "epoch": 0.41, + "grad_norm": 0.23983708730626851, + "learning_rate": 7.969476267322636e-05, + "loss": 1.271, + "step": 222 + }, + { + "epoch": 0.41, + "grad_norm": 0.1922400905426158, + "learning_rate": 7.968416050432912e-05, + "loss": 1.2139, + "step": 223 + }, + { + "epoch": 0.41, + "grad_norm": 0.2238136844422931, + "learning_rate": 7.967337806611568e-05, + "loss": 1.2655, + "step": 224 + }, + { + "epoch": 0.41, + "grad_norm": 0.21230292828267672, + "learning_rate": 7.966241540756631e-05, + "loss": 1.2406, + "step": 225 + }, + { + "epoch": 0.41, + "grad_norm": 0.26656119419070456, + "learning_rate": 7.965127257848004e-05, + "loss": 1.2595, + "step": 226 + }, + { + "epoch": 0.42, + "grad_norm": 0.22381385502992684, + "learning_rate": 7.963994962947426e-05, + "loss": 1.1737, + "step": 227 + }, + { + "epoch": 0.42, + "grad_norm": 0.20056702203994298, + "learning_rate": 7.962844661198462e-05, + "loss": 1.1969, + "step": 228 + }, + { + "epoch": 0.42, + "grad_norm": 0.20148701321526885, + "learning_rate": 7.961676357826478e-05, + "loss": 1.2151, + "step": 229 + }, + { + "epoch": 0.42, + "grad_norm": 0.20034834807028637, + "learning_rate": 7.960490058138604e-05, + "loss": 1.1455, + "step": 230 + }, + { + "epoch": 0.42, + "grad_norm": 0.21050838521846033, + "learning_rate": 7.959285767523732e-05, + "loss": 1.2223, + "step": 231 + }, + { + "epoch": 0.42, + "grad_norm": 0.20904772138969777, + "learning_rate": 7.95806349145247e-05, + "loss": 1.2534, + "step": 232 + }, + { + "epoch": 0.43, + "grad_norm": 0.20307877304792957, + "learning_rate": 7.956823235477134e-05, + "loss": 1.1352, + "step": 233 + }, + { + "epoch": 0.43, + "grad_norm": 0.20501105270897094, + "learning_rate": 7.95556500523171e-05, + "loss": 1.2031, + "step": 234 + }, + { + "epoch": 0.43, + "grad_norm": 0.19800586972038586, + "learning_rate": 7.954288806431838e-05, + "loss": 1.2567, + "step": 235 + }, + { + "epoch": 0.43, + "grad_norm": 0.2175102450594135, + "learning_rate": 7.952994644874777e-05, + "loss": 1.2538, + "step": 236 + }, + { + "epoch": 0.43, + "grad_norm": 0.22698189300067595, + "learning_rate": 7.951682526439391e-05, + "loss": 1.3088, + "step": 237 + }, + { + "epoch": 0.44, + "grad_norm": 0.19208392014975315, + "learning_rate": 7.950352457086109e-05, + "loss": 1.2336, + "step": 238 + }, + { + "epoch": 0.44, + "grad_norm": 0.27004086334319655, + "learning_rate": 7.949004442856905e-05, + "loss": 1.2012, + "step": 239 + }, + { + "epoch": 0.44, + "grad_norm": 0.23420974954538043, + "learning_rate": 7.947638489875272e-05, + "loss": 1.2244, + "step": 240 + }, + { + "epoch": 0.44, + "grad_norm": 0.20514399124802024, + "learning_rate": 7.946254604346186e-05, + "loss": 1.2548, + "step": 241 + }, + { + "epoch": 0.44, + "grad_norm": 0.19334973602372896, + "learning_rate": 7.944852792556092e-05, + "loss": 1.2104, + "step": 242 + }, + { + "epoch": 0.44, + "grad_norm": 0.1992640714537956, + "learning_rate": 7.943433060872858e-05, + "loss": 1.2628, + "step": 243 + }, + { + "epoch": 0.45, + "grad_norm": 0.203284617090413, + "learning_rate": 7.941995415745761e-05, + "loss": 1.2002, + "step": 244 + }, + { + "epoch": 0.45, + "grad_norm": 0.22795306969682058, + "learning_rate": 7.94053986370545e-05, + "loss": 1.2215, + "step": 245 + }, + { + "epoch": 0.45, + "grad_norm": 0.20789041346838505, + "learning_rate": 7.939066411363915e-05, + "loss": 1.0998, + "step": 246 + }, + { + "epoch": 0.45, + "grad_norm": 0.22354868884742066, + "learning_rate": 7.937575065414464e-05, + "loss": 1.2564, + "step": 247 + }, + { + "epoch": 0.45, + "grad_norm": 0.21176392726647736, + "learning_rate": 7.936065832631687e-05, + "loss": 1.2816, + "step": 248 + }, + { + "epoch": 0.46, + "grad_norm": 0.19967179557235587, + "learning_rate": 7.934538719871427e-05, + "loss": 1.1961, + "step": 249 + }, + { + "epoch": 0.46, + "grad_norm": 0.210819577350627, + "learning_rate": 7.932993734070747e-05, + "loss": 1.2167, + "step": 250 + }, + { + "epoch": 0.46, + "grad_norm": 0.21537794551756187, + "learning_rate": 7.931430882247903e-05, + "loss": 1.2341, + "step": 251 + }, + { + "epoch": 0.46, + "grad_norm": 0.22850872387256574, + "learning_rate": 7.929850171502304e-05, + "loss": 1.1686, + "step": 252 + }, + { + "epoch": 0.46, + "grad_norm": 0.22380366415076383, + "learning_rate": 7.928251609014493e-05, + "loss": 1.1462, + "step": 253 + }, + { + "epoch": 0.46, + "grad_norm": 0.22426923149036065, + "learning_rate": 7.926635202046102e-05, + "loss": 1.1792, + "step": 254 + }, + { + "epoch": 0.47, + "grad_norm": 0.42082703321103965, + "learning_rate": 7.925000957939822e-05, + "loss": 1.2718, + "step": 255 + }, + { + "epoch": 0.47, + "grad_norm": 0.2235432774854074, + "learning_rate": 7.92334888411937e-05, + "loss": 1.2598, + "step": 256 + }, + { + "epoch": 0.47, + "grad_norm": 0.281644028934108, + "learning_rate": 7.92167898808946e-05, + "loss": 1.2205, + "step": 257 + }, + { + "epoch": 0.47, + "grad_norm": 0.2037705143888748, + "learning_rate": 7.919991277435763e-05, + "loss": 1.1737, + "step": 258 + }, + { + "epoch": 0.47, + "grad_norm": 0.20917419230028977, + "learning_rate": 7.918285759824879e-05, + "loss": 1.2035, + "step": 259 + }, + { + "epoch": 0.48, + "grad_norm": 0.20510847570635518, + "learning_rate": 7.916562443004292e-05, + "loss": 1.2135, + "step": 260 + }, + { + "epoch": 0.48, + "grad_norm": 0.25172483071092466, + "learning_rate": 7.914821334802342e-05, + "loss": 1.2218, + "step": 261 + }, + { + "epoch": 0.48, + "grad_norm": 0.21102706700634313, + "learning_rate": 7.91306244312819e-05, + "loss": 1.1738, + "step": 262 + }, + { + "epoch": 0.48, + "grad_norm": 0.22626060872645815, + "learning_rate": 7.911285775971781e-05, + "loss": 1.238, + "step": 263 + }, + { + "epoch": 0.48, + "grad_norm": 0.22448567539778486, + "learning_rate": 7.909491341403805e-05, + "loss": 1.2404, + "step": 264 + }, + { + "epoch": 0.48, + "grad_norm": 0.2019099786139193, + "learning_rate": 7.907679147575661e-05, + "loss": 1.213, + "step": 265 + }, + { + "epoch": 0.49, + "grad_norm": 0.24307234839096267, + "learning_rate": 7.905849202719422e-05, + "loss": 1.2322, + "step": 266 + }, + { + "epoch": 0.49, + "grad_norm": 0.19801890521743487, + "learning_rate": 7.904001515147802e-05, + "loss": 1.2448, + "step": 267 + }, + { + "epoch": 0.49, + "grad_norm": 0.2102742273575385, + "learning_rate": 7.902136093254106e-05, + "loss": 1.1657, + "step": 268 + }, + { + "epoch": 0.49, + "grad_norm": 0.2173464476815016, + "learning_rate": 7.900252945512201e-05, + "loss": 1.2549, + "step": 269 + }, + { + "epoch": 0.49, + "grad_norm": 0.20957275458699595, + "learning_rate": 7.898352080476479e-05, + "loss": 1.2536, + "step": 270 + }, + { + "epoch": 0.5, + "grad_norm": 0.20691966388952363, + "learning_rate": 7.896433506781811e-05, + "loss": 1.2661, + "step": 271 + }, + { + "epoch": 0.5, + "grad_norm": 0.2276662275112648, + "learning_rate": 7.894497233143509e-05, + "loss": 1.2409, + "step": 272 + }, + { + "epoch": 0.5, + "grad_norm": 0.23854109569301263, + "learning_rate": 7.892543268357297e-05, + "loss": 1.2681, + "step": 273 + }, + { + "epoch": 0.5, + "grad_norm": 0.2233864156677627, + "learning_rate": 7.890571621299252e-05, + "loss": 1.1687, + "step": 274 + }, + { + "epoch": 0.5, + "grad_norm": 0.20114129147925475, + "learning_rate": 7.888582300925787e-05, + "loss": 1.2184, + "step": 275 + }, + { + "epoch": 0.5, + "grad_norm": 0.2154654670569462, + "learning_rate": 7.886575316273586e-05, + "loss": 1.1982, + "step": 276 + }, + { + "epoch": 0.51, + "grad_norm": 0.2292982209343639, + "learning_rate": 7.884550676459583e-05, + "loss": 1.2129, + "step": 277 + }, + { + "epoch": 0.51, + "grad_norm": 0.21302713135229548, + "learning_rate": 7.882508390680908e-05, + "loss": 1.1605, + "step": 278 + }, + { + "epoch": 0.51, + "grad_norm": 0.2123661020671048, + "learning_rate": 7.88044846821485e-05, + "loss": 1.2308, + "step": 279 + }, + { + "epoch": 0.51, + "grad_norm": 0.2080577410800404, + "learning_rate": 7.878370918418818e-05, + "loss": 1.2195, + "step": 280 + }, + { + "epoch": 0.51, + "grad_norm": 0.19663901881127385, + "learning_rate": 7.876275750730289e-05, + "loss": 1.1591, + "step": 281 + }, + { + "epoch": 0.52, + "grad_norm": 0.20534502031312163, + "learning_rate": 7.874162974666776e-05, + "loss": 1.2664, + "step": 282 + }, + { + "epoch": 0.52, + "grad_norm": 0.23240445399513837, + "learning_rate": 7.872032599825779e-05, + "loss": 1.2151, + "step": 283 + }, + { + "epoch": 0.52, + "grad_norm": 0.2672527316717507, + "learning_rate": 7.86988463588474e-05, + "loss": 1.2406, + "step": 284 + }, + { + "epoch": 0.52, + "grad_norm": 0.19893903058743695, + "learning_rate": 7.867719092601003e-05, + "loss": 1.1291, + "step": 285 + }, + { + "epoch": 0.52, + "grad_norm": 0.33275268109930917, + "learning_rate": 7.865535979811768e-05, + "loss": 1.1406, + "step": 286 + }, + { + "epoch": 0.52, + "grad_norm": 0.2373619455690358, + "learning_rate": 7.863335307434045e-05, + "loss": 1.2799, + "step": 287 + }, + { + "epoch": 0.53, + "grad_norm": 0.263235735390858, + "learning_rate": 7.861117085464612e-05, + "loss": 1.2415, + "step": 288 + }, + { + "epoch": 0.53, + "grad_norm": 0.25884281780784324, + "learning_rate": 7.858881323979965e-05, + "loss": 1.3919, + "step": 289 + }, + { + "epoch": 0.53, + "grad_norm": 0.25426288332255736, + "learning_rate": 7.85662803313628e-05, + "loss": 1.174, + "step": 290 + }, + { + "epoch": 0.53, + "grad_norm": 0.26655405527881243, + "learning_rate": 7.854357223169356e-05, + "loss": 1.2806, + "step": 291 + }, + { + "epoch": 0.53, + "grad_norm": 0.20909844432349833, + "learning_rate": 7.852068904394579e-05, + "loss": 1.2627, + "step": 292 + }, + { + "epoch": 0.54, + "grad_norm": 0.21307115068935759, + "learning_rate": 7.849763087206866e-05, + "loss": 1.1879, + "step": 293 + }, + { + "epoch": 0.54, + "grad_norm": 0.25009949471398946, + "learning_rate": 7.847439782080628e-05, + "loss": 1.2881, + "step": 294 + }, + { + "epoch": 0.54, + "grad_norm": 0.20960783418679174, + "learning_rate": 7.845098999569712e-05, + "loss": 1.2723, + "step": 295 + }, + { + "epoch": 0.54, + "grad_norm": 0.24968832437925104, + "learning_rate": 7.842740750307362e-05, + "loss": 1.2029, + "step": 296 + }, + { + "epoch": 0.54, + "grad_norm": 0.22981196585125677, + "learning_rate": 7.84036504500616e-05, + "loss": 1.1695, + "step": 297 + }, + { + "epoch": 0.55, + "grad_norm": 0.2320606844751365, + "learning_rate": 7.837971894457991e-05, + "loss": 1.2317, + "step": 298 + }, + { + "epoch": 0.55, + "grad_norm": 0.23051459673906124, + "learning_rate": 7.835561309533981e-05, + "loss": 1.2046, + "step": 299 + }, + { + "epoch": 0.55, + "grad_norm": 0.2510027231060586, + "learning_rate": 7.833133301184457e-05, + "loss": 1.199, + "step": 300 + }, + { + "epoch": 0.55, + "grad_norm": 0.23601180466018787, + "learning_rate": 7.830687880438895e-05, + "loss": 1.1755, + "step": 301 + }, + { + "epoch": 0.55, + "grad_norm": 0.24740820934385369, + "learning_rate": 7.828225058405864e-05, + "loss": 1.2054, + "step": 302 + }, + { + "epoch": 0.55, + "grad_norm": 0.23065372979111173, + "learning_rate": 7.825744846272984e-05, + "loss": 1.2066, + "step": 303 + }, + { + "epoch": 0.56, + "grad_norm": 0.22385077334838213, + "learning_rate": 7.823247255306866e-05, + "loss": 1.2147, + "step": 304 + }, + { + "epoch": 0.56, + "grad_norm": 0.42981213948386104, + "learning_rate": 7.820732296853074e-05, + "loss": 1.2314, + "step": 305 + }, + { + "epoch": 0.56, + "grad_norm": 0.21122844902751076, + "learning_rate": 7.818199982336058e-05, + "loss": 1.1462, + "step": 306 + }, + { + "epoch": 0.56, + "grad_norm": 0.23374869692118933, + "learning_rate": 7.815650323259117e-05, + "loss": 1.2051, + "step": 307 + }, + { + "epoch": 0.56, + "grad_norm": 0.21662363795962128, + "learning_rate": 7.813083331204332e-05, + "loss": 1.1575, + "step": 308 + }, + { + "epoch": 0.57, + "grad_norm": 0.2088315773384112, + "learning_rate": 7.810499017832526e-05, + "loss": 1.1316, + "step": 309 + }, + { + "epoch": 0.57, + "grad_norm": 0.2095238410730976, + "learning_rate": 7.807897394883203e-05, + "loss": 1.2087, + "step": 310 + }, + { + "epoch": 0.57, + "grad_norm": 0.22672932127256515, + "learning_rate": 7.805278474174499e-05, + "loss": 1.2512, + "step": 311 + }, + { + "epoch": 0.57, + "grad_norm": 0.21873052340922736, + "learning_rate": 7.802642267603126e-05, + "loss": 1.1909, + "step": 312 + }, + { + "epoch": 0.57, + "grad_norm": 0.219814521916342, + "learning_rate": 7.79998878714432e-05, + "loss": 1.1669, + "step": 313 + }, + { + "epoch": 0.57, + "grad_norm": 0.3049426027257317, + "learning_rate": 7.797318044851786e-05, + "loss": 1.1797, + "step": 314 + }, + { + "epoch": 0.58, + "grad_norm": 0.22309435690065985, + "learning_rate": 7.794630052857638e-05, + "loss": 1.1417, + "step": 315 + }, + { + "epoch": 0.58, + "grad_norm": 0.3891885169154885, + "learning_rate": 7.791924823372354e-05, + "loss": 1.2369, + "step": 316 + }, + { + "epoch": 0.58, + "grad_norm": 0.24780269452456372, + "learning_rate": 7.789202368684711e-05, + "loss": 1.2521, + "step": 317 + }, + { + "epoch": 0.58, + "grad_norm": 0.21660460720269362, + "learning_rate": 7.786462701161738e-05, + "loss": 1.2151, + "step": 318 + }, + { + "epoch": 0.58, + "grad_norm": 0.23635409466561857, + "learning_rate": 7.783705833248649e-05, + "loss": 1.2363, + "step": 319 + }, + { + "epoch": 0.59, + "grad_norm": 0.2616135839903218, + "learning_rate": 7.780931777468797e-05, + "loss": 1.2428, + "step": 320 + }, + { + "epoch": 0.59, + "grad_norm": 0.21461059159245083, + "learning_rate": 7.77814054642361e-05, + "loss": 1.1434, + "step": 321 + }, + { + "epoch": 0.59, + "grad_norm": 0.25348824286656163, + "learning_rate": 7.775332152792539e-05, + "loss": 1.2368, + "step": 322 + }, + { + "epoch": 0.59, + "grad_norm": 0.22275034726331247, + "learning_rate": 7.772506609332995e-05, + "loss": 1.1827, + "step": 323 + }, + { + "epoch": 0.59, + "grad_norm": 0.25030821228147526, + "learning_rate": 7.769663928880298e-05, + "loss": 1.2428, + "step": 324 + }, + { + "epoch": 0.59, + "grad_norm": 0.22251804398745534, + "learning_rate": 7.766804124347608e-05, + "loss": 1.1889, + "step": 325 + }, + { + "epoch": 0.6, + "grad_norm": 0.23381455520411995, + "learning_rate": 7.763927208725879e-05, + "loss": 1.2115, + "step": 326 + }, + { + "epoch": 0.6, + "grad_norm": 0.27341902651946226, + "learning_rate": 7.761033195083791e-05, + "loss": 1.2535, + "step": 327 + }, + { + "epoch": 0.6, + "grad_norm": 0.24862471659814522, + "learning_rate": 7.758122096567694e-05, + "loss": 1.2128, + "step": 328 + }, + { + "epoch": 0.6, + "grad_norm": 0.2251357082045494, + "learning_rate": 7.755193926401547e-05, + "loss": 1.2334, + "step": 329 + }, + { + "epoch": 0.6, + "grad_norm": 0.3173274941622932, + "learning_rate": 7.752248697886857e-05, + "loss": 1.226, + "step": 330 + }, + { + "epoch": 0.61, + "grad_norm": 0.23056440717672175, + "learning_rate": 7.74928642440263e-05, + "loss": 1.2339, + "step": 331 + }, + { + "epoch": 0.61, + "grad_norm": 0.2801507500859342, + "learning_rate": 7.746307119405286e-05, + "loss": 1.287, + "step": 332 + }, + { + "epoch": 0.61, + "grad_norm": 0.2267818430426272, + "learning_rate": 7.743310796428622e-05, + "loss": 1.1916, + "step": 333 + }, + { + "epoch": 0.61, + "grad_norm": 0.2777329160365585, + "learning_rate": 7.74029746908374e-05, + "loss": 1.252, + "step": 334 + }, + { + "epoch": 0.61, + "grad_norm": 0.25289169762353, + "learning_rate": 7.737267151058983e-05, + "loss": 1.2153, + "step": 335 + }, + { + "epoch": 0.61, + "grad_norm": 0.2424670686901653, + "learning_rate": 7.734219856119875e-05, + "loss": 1.2227, + "step": 336 + }, + { + "epoch": 0.62, + "grad_norm": 0.22747092217441645, + "learning_rate": 7.731155598109067e-05, + "loss": 1.19, + "step": 337 + }, + { + "epoch": 0.62, + "grad_norm": 0.2307810940100189, + "learning_rate": 7.728074390946257e-05, + "loss": 1.1818, + "step": 338 + }, + { + "epoch": 0.62, + "grad_norm": 0.2583402574655623, + "learning_rate": 7.724976248628142e-05, + "loss": 1.1608, + "step": 339 + }, + { + "epoch": 0.62, + "grad_norm": 0.22140209760890694, + "learning_rate": 7.721861185228347e-05, + "loss": 1.1245, + "step": 340 + }, + { + "epoch": 0.62, + "grad_norm": 0.25859310758244686, + "learning_rate": 7.718729214897362e-05, + "loss": 1.2247, + "step": 341 + }, + { + "epoch": 0.63, + "grad_norm": 0.26371179531372124, + "learning_rate": 7.715580351862482e-05, + "loss": 1.2128, + "step": 342 + }, + { + "epoch": 0.63, + "grad_norm": 0.26575541302851047, + "learning_rate": 7.712414610427733e-05, + "loss": 1.2443, + "step": 343 + }, + { + "epoch": 0.63, + "grad_norm": 0.269978305197599, + "learning_rate": 7.709232004973816e-05, + "loss": 1.2231, + "step": 344 + }, + { + "epoch": 0.63, + "grad_norm": 0.26583998705977047, + "learning_rate": 7.70603254995804e-05, + "loss": 1.2476, + "step": 345 + }, + { + "epoch": 0.63, + "grad_norm": 0.24256062164066097, + "learning_rate": 7.702816259914253e-05, + "loss": 1.2901, + "step": 346 + }, + { + "epoch": 0.63, + "grad_norm": 0.3463123472658915, + "learning_rate": 7.699583149452779e-05, + "loss": 1.3277, + "step": 347 + }, + { + "epoch": 0.64, + "grad_norm": 0.2269096590531878, + "learning_rate": 7.696333233260345e-05, + "loss": 1.2047, + "step": 348 + }, + { + "epoch": 0.64, + "grad_norm": 0.25136883001050025, + "learning_rate": 7.693066526100031e-05, + "loss": 1.1619, + "step": 349 + }, + { + "epoch": 0.64, + "grad_norm": 0.2565112571116145, + "learning_rate": 7.68978304281118e-05, + "loss": 1.2389, + "step": 350 + }, + { + "epoch": 0.64, + "grad_norm": 0.22175779550828703, + "learning_rate": 7.686482798309349e-05, + "loss": 1.2238, + "step": 351 + }, + { + "epoch": 0.64, + "grad_norm": 0.22588304332216555, + "learning_rate": 7.683165807586234e-05, + "loss": 1.174, + "step": 352 + }, + { + "epoch": 0.65, + "grad_norm": 0.24889474296529737, + "learning_rate": 7.6798320857096e-05, + "loss": 1.2366, + "step": 353 + }, + { + "epoch": 0.65, + "grad_norm": 0.27339703806525034, + "learning_rate": 7.676481647823214e-05, + "loss": 1.2356, + "step": 354 + }, + { + "epoch": 0.65, + "grad_norm": 0.23424666722888365, + "learning_rate": 7.673114509146782e-05, + "loss": 1.2089, + "step": 355 + }, + { + "epoch": 0.65, + "grad_norm": 0.27978285392461766, + "learning_rate": 7.66973068497587e-05, + "loss": 1.2609, + "step": 356 + }, + { + "epoch": 0.65, + "grad_norm": 0.2509423350138824, + "learning_rate": 7.666330190681844e-05, + "loss": 1.1777, + "step": 357 + }, + { + "epoch": 0.65, + "grad_norm": 0.23007730927468031, + "learning_rate": 7.662913041711793e-05, + "loss": 1.154, + "step": 358 + }, + { + "epoch": 0.66, + "grad_norm": 0.2438648674953112, + "learning_rate": 7.659479253588462e-05, + "loss": 1.2257, + "step": 359 + }, + { + "epoch": 0.66, + "grad_norm": 0.28816093242092233, + "learning_rate": 7.65602884191018e-05, + "loss": 1.2558, + "step": 360 + }, + { + "epoch": 0.66, + "grad_norm": 0.24972815300596035, + "learning_rate": 7.652561822350793e-05, + "loss": 1.2837, + "step": 361 + }, + { + "epoch": 0.66, + "grad_norm": 0.2543189139697063, + "learning_rate": 7.649078210659587e-05, + "loss": 1.2193, + "step": 362 + }, + { + "epoch": 0.66, + "grad_norm": 0.2237937956718952, + "learning_rate": 7.645578022661224e-05, + "loss": 1.2237, + "step": 363 + }, + { + "epoch": 0.67, + "grad_norm": 0.29742029408787396, + "learning_rate": 7.642061274255657e-05, + "loss": 1.2116, + "step": 364 + }, + { + "epoch": 0.67, + "grad_norm": 0.2462883147335493, + "learning_rate": 7.638527981418075e-05, + "loss": 1.1827, + "step": 365 + }, + { + "epoch": 0.67, + "grad_norm": 0.2647802498907096, + "learning_rate": 7.634978160198817e-05, + "loss": 1.2739, + "step": 366 + }, + { + "epoch": 0.67, + "grad_norm": 0.22360398779217264, + "learning_rate": 7.631411826723306e-05, + "loss": 1.2185, + "step": 367 + }, + { + "epoch": 0.67, + "grad_norm": 0.2635048004593543, + "learning_rate": 7.627828997191973e-05, + "loss": 1.2317, + "step": 368 + }, + { + "epoch": 0.67, + "grad_norm": 0.2764803449917684, + "learning_rate": 7.624229687880184e-05, + "loss": 1.1923, + "step": 369 + }, + { + "epoch": 0.68, + "grad_norm": 0.25724943233414527, + "learning_rate": 7.620613915138166e-05, + "loss": 1.2218, + "step": 370 + }, + { + "epoch": 0.68, + "grad_norm": 0.2858318045794755, + "learning_rate": 7.61698169539093e-05, + "loss": 1.1496, + "step": 371 + }, + { + "epoch": 0.68, + "grad_norm": 0.23547216647460364, + "learning_rate": 7.613333045138206e-05, + "loss": 1.1905, + "step": 372 + }, + { + "epoch": 0.68, + "grad_norm": 0.22984814903684375, + "learning_rate": 7.609667980954355e-05, + "loss": 1.2009, + "step": 373 + }, + { + "epoch": 0.68, + "grad_norm": 0.2551903754079084, + "learning_rate": 7.605986519488301e-05, + "loss": 1.2042, + "step": 374 + }, + { + "epoch": 0.69, + "grad_norm": 0.2508257410125616, + "learning_rate": 7.602288677463457e-05, + "loss": 1.2468, + "step": 375 + }, + { + "epoch": 0.69, + "grad_norm": 0.25324577774935964, + "learning_rate": 7.598574471677644e-05, + "loss": 1.2603, + "step": 376 + }, + { + "epoch": 0.69, + "grad_norm": 0.35888776531769967, + "learning_rate": 7.59484391900302e-05, + "loss": 1.1929, + "step": 377 + }, + { + "epoch": 0.69, + "grad_norm": 0.22048517191014724, + "learning_rate": 7.591097036385994e-05, + "loss": 1.1783, + "step": 378 + }, + { + "epoch": 0.69, + "grad_norm": 0.2781160412746083, + "learning_rate": 7.587333840847162e-05, + "loss": 1.3397, + "step": 379 + }, + { + "epoch": 0.7, + "grad_norm": 0.24033046830332258, + "learning_rate": 7.583554349481222e-05, + "loss": 1.2436, + "step": 380 + }, + { + "epoch": 0.7, + "grad_norm": 0.26413762380260003, + "learning_rate": 7.579758579456893e-05, + "loss": 1.1917, + "step": 381 + }, + { + "epoch": 0.7, + "grad_norm": 0.2390937887338632, + "learning_rate": 7.575946548016847e-05, + "loss": 1.2186, + "step": 382 + }, + { + "epoch": 0.7, + "grad_norm": 0.25131263043429275, + "learning_rate": 7.572118272477622e-05, + "loss": 1.2538, + "step": 383 + }, + { + "epoch": 0.7, + "grad_norm": 0.223974104870702, + "learning_rate": 7.568273770229546e-05, + "loss": 1.2165, + "step": 384 + }, + { + "epoch": 0.7, + "grad_norm": 0.25840356830252875, + "learning_rate": 7.564413058736663e-05, + "loss": 1.1848, + "step": 385 + }, + { + "epoch": 0.71, + "grad_norm": 0.2723156683076603, + "learning_rate": 7.560536155536641e-05, + "loss": 1.1982, + "step": 386 + }, + { + "epoch": 0.71, + "grad_norm": 0.265687427976889, + "learning_rate": 7.556643078240708e-05, + "loss": 1.231, + "step": 387 + }, + { + "epoch": 0.71, + "grad_norm": 0.25152762080976077, + "learning_rate": 7.552733844533562e-05, + "loss": 1.1974, + "step": 388 + }, + { + "epoch": 0.71, + "grad_norm": 0.2366049485053541, + "learning_rate": 7.548808472173292e-05, + "loss": 1.3119, + "step": 389 + }, + { + "epoch": 0.71, + "grad_norm": 0.22092196577077122, + "learning_rate": 7.5448669789913e-05, + "loss": 1.195, + "step": 390 + }, + { + "epoch": 0.72, + "grad_norm": 0.22667521540462374, + "learning_rate": 7.540909382892217e-05, + "loss": 1.1431, + "step": 391 + }, + { + "epoch": 0.72, + "grad_norm": 0.25432207282646513, + "learning_rate": 7.536935701853823e-05, + "loss": 1.2173, + "step": 392 + }, + { + "epoch": 0.72, + "grad_norm": 0.29950506457923864, + "learning_rate": 7.53294595392697e-05, + "loss": 1.1962, + "step": 393 + }, + { + "epoch": 0.72, + "grad_norm": 0.24735689607229913, + "learning_rate": 7.528940157235487e-05, + "loss": 1.2053, + "step": 394 + }, + { + "epoch": 0.72, + "grad_norm": 0.24394198607459663, + "learning_rate": 7.524918329976114e-05, + "loss": 1.1979, + "step": 395 + }, + { + "epoch": 0.72, + "grad_norm": 0.2630369372689188, + "learning_rate": 7.520880490418409e-05, + "loss": 1.2111, + "step": 396 + }, + { + "epoch": 0.73, + "grad_norm": 0.26275028416291457, + "learning_rate": 7.516826656904664e-05, + "loss": 1.2133, + "step": 397 + }, + { + "epoch": 0.73, + "grad_norm": 0.23938074620956928, + "learning_rate": 7.512756847849831e-05, + "loss": 1.1355, + "step": 398 + }, + { + "epoch": 0.73, + "grad_norm": 0.3724960610098138, + "learning_rate": 7.508671081741428e-05, + "loss": 1.2572, + "step": 399 + }, + { + "epoch": 0.73, + "grad_norm": 0.24161685847894723, + "learning_rate": 7.504569377139462e-05, + "loss": 1.1706, + "step": 400 + }, + { + "epoch": 0.73, + "grad_norm": 0.26121591322670523, + "learning_rate": 7.50045175267634e-05, + "loss": 1.2135, + "step": 401 + }, + { + "epoch": 0.74, + "grad_norm": 0.2465579498164775, + "learning_rate": 7.496318227056788e-05, + "loss": 1.1641, + "step": 402 + }, + { + "epoch": 0.74, + "grad_norm": 0.2556288696122787, + "learning_rate": 7.492168819057767e-05, + "loss": 1.2939, + "step": 403 + }, + { + "epoch": 0.74, + "grad_norm": 0.261481216336303, + "learning_rate": 7.488003547528382e-05, + "loss": 1.2026, + "step": 404 + }, + { + "epoch": 0.74, + "grad_norm": 0.2389415135676362, + "learning_rate": 7.483822431389799e-05, + "loss": 1.2131, + "step": 405 + }, + { + "epoch": 0.74, + "grad_norm": 0.2559201956627192, + "learning_rate": 7.479625489635162e-05, + "loss": 1.1246, + "step": 406 + }, + { + "epoch": 0.74, + "grad_norm": 0.27127932491822604, + "learning_rate": 7.475412741329504e-05, + "loss": 1.2429, + "step": 407 + }, + { + "epoch": 0.75, + "grad_norm": 0.27006004008695594, + "learning_rate": 7.47118420560966e-05, + "loss": 1.2388, + "step": 408 + }, + { + "epoch": 0.75, + "grad_norm": 0.23716823297200537, + "learning_rate": 7.466939901684182e-05, + "loss": 1.1264, + "step": 409 + }, + { + "epoch": 0.75, + "grad_norm": 0.2885373898669248, + "learning_rate": 7.462679848833252e-05, + "loss": 1.2786, + "step": 410 + }, + { + "epoch": 0.75, + "grad_norm": 0.49215227598639927, + "learning_rate": 7.458404066408588e-05, + "loss": 1.2386, + "step": 411 + }, + { + "epoch": 0.75, + "grad_norm": 0.24235735604947403, + "learning_rate": 7.454112573833368e-05, + "loss": 1.1423, + "step": 412 + }, + { + "epoch": 0.76, + "grad_norm": 0.2584614748054343, + "learning_rate": 7.449805390602127e-05, + "loss": 1.2669, + "step": 413 + }, + { + "epoch": 0.76, + "grad_norm": 0.23806123085998873, + "learning_rate": 7.445482536280684e-05, + "loss": 1.1763, + "step": 414 + }, + { + "epoch": 0.76, + "grad_norm": 0.24459517607786851, + "learning_rate": 7.441144030506043e-05, + "loss": 1.198, + "step": 415 + }, + { + "epoch": 0.76, + "grad_norm": 0.25801616402700395, + "learning_rate": 7.436789892986304e-05, + "loss": 1.2136, + "step": 416 + }, + { + "epoch": 0.76, + "grad_norm": 0.2814819942392514, + "learning_rate": 7.432420143500578e-05, + "loss": 1.2398, + "step": 417 + }, + { + "epoch": 0.76, + "grad_norm": 0.22134709322606153, + "learning_rate": 7.428034801898893e-05, + "loss": 1.1592, + "step": 418 + }, + { + "epoch": 0.77, + "grad_norm": 0.2899677536995633, + "learning_rate": 7.42363388810211e-05, + "loss": 1.2296, + "step": 419 + }, + { + "epoch": 0.77, + "grad_norm": 0.24005943230262294, + "learning_rate": 7.419217422101822e-05, + "loss": 1.2223, + "step": 420 + }, + { + "epoch": 0.77, + "grad_norm": 0.26417562369496167, + "learning_rate": 7.414785423960275e-05, + "loss": 1.2261, + "step": 421 + }, + { + "epoch": 0.77, + "grad_norm": 0.2580815883535521, + "learning_rate": 7.410337913810271e-05, + "loss": 1.2021, + "step": 422 + }, + { + "epoch": 0.77, + "grad_norm": 0.25242217589496435, + "learning_rate": 7.405874911855071e-05, + "loss": 1.239, + "step": 423 + }, + { + "epoch": 0.78, + "grad_norm": 0.21991733999839932, + "learning_rate": 7.401396438368315e-05, + "loss": 1.1716, + "step": 424 + }, + { + "epoch": 0.78, + "grad_norm": 0.40116538322720213, + "learning_rate": 7.396902513693924e-05, + "loss": 1.2773, + "step": 425 + }, + { + "epoch": 0.78, + "grad_norm": 0.277333939455099, + "learning_rate": 7.392393158246002e-05, + "loss": 1.2574, + "step": 426 + }, + { + "epoch": 0.78, + "grad_norm": 0.27146087746385755, + "learning_rate": 7.387868392508756e-05, + "loss": 1.2243, + "step": 427 + }, + { + "epoch": 0.78, + "grad_norm": 0.255881055620786, + "learning_rate": 7.38332823703639e-05, + "loss": 1.223, + "step": 428 + }, + { + "epoch": 0.78, + "grad_norm": 0.24807364856677255, + "learning_rate": 7.378772712453021e-05, + "loss": 1.1985, + "step": 429 + }, + { + "epoch": 0.79, + "grad_norm": 0.25746257617764423, + "learning_rate": 7.37420183945258e-05, + "loss": 1.2502, + "step": 430 + }, + { + "epoch": 0.79, + "grad_norm": 0.28851991049982234, + "learning_rate": 7.369615638798722e-05, + "loss": 1.2535, + "step": 431 + }, + { + "epoch": 0.79, + "grad_norm": 0.24113389811604363, + "learning_rate": 7.365014131324725e-05, + "loss": 1.2227, + "step": 432 + }, + { + "epoch": 0.79, + "grad_norm": 0.2414465151257969, + "learning_rate": 7.360397337933405e-05, + "loss": 1.1884, + "step": 433 + }, + { + "epoch": 0.79, + "grad_norm": 0.2735463134699831, + "learning_rate": 7.355765279597011e-05, + "loss": 1.2756, + "step": 434 + }, + { + "epoch": 0.8, + "grad_norm": 0.2588437452987293, + "learning_rate": 7.351117977357139e-05, + "loss": 1.2108, + "step": 435 + }, + { + "epoch": 0.8, + "grad_norm": 0.26573294117796553, + "learning_rate": 7.346455452324629e-05, + "loss": 1.1821, + "step": 436 + }, + { + "epoch": 0.8, + "grad_norm": 0.2555476577827304, + "learning_rate": 7.341777725679473e-05, + "loss": 1.1937, + "step": 437 + }, + { + "epoch": 0.8, + "grad_norm": 0.2867704132108098, + "learning_rate": 7.337084818670716e-05, + "loss": 1.2272, + "step": 438 + }, + { + "epoch": 0.8, + "grad_norm": 0.27726678115981157, + "learning_rate": 7.332376752616367e-05, + "loss": 1.2331, + "step": 439 + }, + { + "epoch": 0.8, + "grad_norm": 0.26955338021079955, + "learning_rate": 7.32765354890329e-05, + "loss": 1.1731, + "step": 440 + }, + { + "epoch": 0.81, + "grad_norm": 0.25250321202536524, + "learning_rate": 7.322915228987116e-05, + "loss": 1.2653, + "step": 441 + }, + { + "epoch": 0.81, + "grad_norm": 0.24748844179765395, + "learning_rate": 7.318161814392143e-05, + "loss": 1.24, + "step": 442 + }, + { + "epoch": 0.81, + "grad_norm": 0.28177805247356325, + "learning_rate": 7.313393326711239e-05, + "loss": 1.185, + "step": 443 + }, + { + "epoch": 0.81, + "grad_norm": 0.24093242000396312, + "learning_rate": 7.30860978760574e-05, + "loss": 1.1994, + "step": 444 + }, + { + "epoch": 0.81, + "grad_norm": 0.26277803901457075, + "learning_rate": 7.30381121880536e-05, + "loss": 1.212, + "step": 445 + }, + { + "epoch": 0.82, + "grad_norm": 0.2506524258682433, + "learning_rate": 7.298997642108079e-05, + "loss": 1.2421, + "step": 446 + }, + { + "epoch": 0.82, + "grad_norm": 0.2840599700015824, + "learning_rate": 7.294169079380061e-05, + "loss": 1.1818, + "step": 447 + }, + { + "epoch": 0.82, + "grad_norm": 0.24892184038117549, + "learning_rate": 7.289325552555538e-05, + "loss": 1.1916, + "step": 448 + }, + { + "epoch": 0.82, + "grad_norm": 0.2700898428541357, + "learning_rate": 7.284467083636722e-05, + "loss": 1.2517, + "step": 449 + }, + { + "epoch": 0.82, + "grad_norm": 0.2617848546539419, + "learning_rate": 7.279593694693698e-05, + "loss": 1.2063, + "step": 450 + }, + { + "epoch": 0.82, + "grad_norm": 0.2698278585334131, + "learning_rate": 7.274705407864332e-05, + "loss": 1.194, + "step": 451 + }, + { + "epoch": 0.83, + "grad_norm": 0.23678313024953834, + "learning_rate": 7.26980224535416e-05, + "loss": 1.2349, + "step": 452 + }, + { + "epoch": 0.83, + "grad_norm": 0.24851875792002978, + "learning_rate": 7.264884229436293e-05, + "loss": 1.1758, + "step": 453 + }, + { + "epoch": 0.83, + "grad_norm": 0.24122080121681125, + "learning_rate": 7.259951382451318e-05, + "loss": 1.1962, + "step": 454 + }, + { + "epoch": 0.83, + "grad_norm": 0.22741322959884405, + "learning_rate": 7.25500372680719e-05, + "loss": 1.1702, + "step": 455 + }, + { + "epoch": 0.83, + "grad_norm": 0.2297475610861458, + "learning_rate": 7.250041284979137e-05, + "loss": 1.1466, + "step": 456 + }, + { + "epoch": 0.84, + "grad_norm": 0.3057605989721467, + "learning_rate": 7.245064079509553e-05, + "loss": 1.246, + "step": 457 + }, + { + "epoch": 0.84, + "grad_norm": 0.2719638501597136, + "learning_rate": 7.240072133007899e-05, + "loss": 1.2184, + "step": 458 + }, + { + "epoch": 0.84, + "grad_norm": 0.2436807816414479, + "learning_rate": 7.235065468150593e-05, + "loss": 1.2324, + "step": 459 + }, + { + "epoch": 0.84, + "grad_norm": 0.23436349430255515, + "learning_rate": 7.23004410768092e-05, + "loss": 1.1813, + "step": 460 + }, + { + "epoch": 0.84, + "grad_norm": 0.2398940990211377, + "learning_rate": 7.22500807440892e-05, + "loss": 1.1924, + "step": 461 + }, + { + "epoch": 0.84, + "grad_norm": 0.2605716625062531, + "learning_rate": 7.219957391211281e-05, + "loss": 1.182, + "step": 462 + }, + { + "epoch": 0.85, + "grad_norm": 0.260462524570941, + "learning_rate": 7.214892081031244e-05, + "loss": 1.2136, + "step": 463 + }, + { + "epoch": 0.85, + "grad_norm": 0.21979766512306334, + "learning_rate": 7.209812166878491e-05, + "loss": 1.2066, + "step": 464 + }, + { + "epoch": 0.85, + "grad_norm": 0.23324453647530663, + "learning_rate": 7.204717671829051e-05, + "loss": 1.1657, + "step": 465 + }, + { + "epoch": 0.85, + "grad_norm": 0.2529434935507481, + "learning_rate": 7.199608619025177e-05, + "loss": 1.2093, + "step": 466 + }, + { + "epoch": 0.85, + "grad_norm": 0.25371701891720116, + "learning_rate": 7.194485031675265e-05, + "loss": 1.2225, + "step": 467 + }, + { + "epoch": 0.86, + "grad_norm": 0.23272423066292103, + "learning_rate": 7.189346933053725e-05, + "loss": 1.1721, + "step": 468 + }, + { + "epoch": 0.86, + "grad_norm": 0.25122928735587546, + "learning_rate": 7.184194346500892e-05, + "loss": 1.2537, + "step": 469 + }, + { + "epoch": 0.86, + "grad_norm": 0.2159270875490409, + "learning_rate": 7.179027295422913e-05, + "loss": 1.197, + "step": 470 + }, + { + "epoch": 0.86, + "grad_norm": 0.2633111059076544, + "learning_rate": 7.173845803291636e-05, + "loss": 1.1721, + "step": 471 + }, + { + "epoch": 0.86, + "grad_norm": 0.30555936322098703, + "learning_rate": 7.168649893644517e-05, + "loss": 1.3011, + "step": 472 + }, + { + "epoch": 0.87, + "grad_norm": 0.23492670111453726, + "learning_rate": 7.163439590084502e-05, + "loss": 1.1601, + "step": 473 + }, + { + "epoch": 0.87, + "grad_norm": 0.26602734263721806, + "learning_rate": 7.158214916279923e-05, + "loss": 1.2808, + "step": 474 + }, + { + "epoch": 0.87, + "grad_norm": 0.3182695007856262, + "learning_rate": 7.152975895964386e-05, + "loss": 1.2967, + "step": 475 + }, + { + "epoch": 0.87, + "grad_norm": 0.2785021674736721, + "learning_rate": 7.147722552936673e-05, + "loss": 1.1789, + "step": 476 + }, + { + "epoch": 0.87, + "grad_norm": 0.279474303138652, + "learning_rate": 7.142454911060627e-05, + "loss": 1.2596, + "step": 477 + }, + { + "epoch": 0.87, + "grad_norm": 0.2556980144910755, + "learning_rate": 7.137172994265044e-05, + "loss": 1.2426, + "step": 478 + }, + { + "epoch": 0.88, + "grad_norm": 0.3311256331993533, + "learning_rate": 7.131876826543565e-05, + "loss": 1.2059, + "step": 479 + }, + { + "epoch": 0.88, + "grad_norm": 0.26467296197775253, + "learning_rate": 7.12656643195457e-05, + "loss": 1.2482, + "step": 480 + }, + { + "epoch": 0.88, + "grad_norm": 0.27444885274652553, + "learning_rate": 7.121241834621064e-05, + "loss": 1.2528, + "step": 481 + }, + { + "epoch": 0.88, + "grad_norm": 0.2572283861115396, + "learning_rate": 7.115903058730567e-05, + "loss": 1.1849, + "step": 482 + }, + { + "epoch": 0.88, + "grad_norm": 0.2677065778235683, + "learning_rate": 7.11055012853501e-05, + "loss": 1.2011, + "step": 483 + }, + { + "epoch": 0.89, + "grad_norm": 0.29470622036742816, + "learning_rate": 7.105183068350619e-05, + "loss": 1.2398, + "step": 484 + }, + { + "epoch": 0.89, + "grad_norm": 0.27609230248969197, + "learning_rate": 7.099801902557811e-05, + "loss": 1.2259, + "step": 485 + }, + { + "epoch": 0.89, + "grad_norm": 0.24248634168099284, + "learning_rate": 7.094406655601073e-05, + "loss": 1.2282, + "step": 486 + }, + { + "epoch": 0.89, + "grad_norm": 0.2765941767688746, + "learning_rate": 7.088997351988865e-05, + "loss": 1.2319, + "step": 487 + }, + { + "epoch": 0.89, + "grad_norm": 0.29347776909858947, + "learning_rate": 7.083574016293493e-05, + "loss": 1.1765, + "step": 488 + }, + { + "epoch": 0.89, + "grad_norm": 0.285370295424537, + "learning_rate": 7.078136673151008e-05, + "loss": 1.26, + "step": 489 + }, + { + "epoch": 0.9, + "grad_norm": 0.29408734903836536, + "learning_rate": 7.072685347261093e-05, + "loss": 1.226, + "step": 490 + }, + { + "epoch": 0.9, + "grad_norm": 0.27437470239205813, + "learning_rate": 7.067220063386947e-05, + "loss": 1.1976, + "step": 491 + }, + { + "epoch": 0.9, + "grad_norm": 0.2680770258777871, + "learning_rate": 7.061740846355176e-05, + "loss": 1.1915, + "step": 492 + }, + { + "epoch": 0.9, + "grad_norm": 0.27200362879502954, + "learning_rate": 7.056247721055678e-05, + "loss": 1.2002, + "step": 493 + }, + { + "epoch": 0.9, + "grad_norm": 0.2637811092577037, + "learning_rate": 7.050740712441528e-05, + "loss": 1.287, + "step": 494 + }, + { + "epoch": 0.91, + "grad_norm": 0.24657959209271266, + "learning_rate": 7.045219845528875e-05, + "loss": 1.2284, + "step": 495 + }, + { + "epoch": 0.91, + "grad_norm": 0.25311992110358666, + "learning_rate": 7.039685145396812e-05, + "loss": 1.1616, + "step": 496 + }, + { + "epoch": 0.91, + "grad_norm": 0.2564633694193358, + "learning_rate": 7.034136637187275e-05, + "loss": 1.2067, + "step": 497 + }, + { + "epoch": 0.91, + "grad_norm": 0.2446797651174144, + "learning_rate": 7.028574346104926e-05, + "loss": 1.2284, + "step": 498 + }, + { + "epoch": 0.91, + "grad_norm": 0.2592751463399255, + "learning_rate": 7.022998297417034e-05, + "loss": 1.2371, + "step": 499 + }, + { + "epoch": 0.91, + "grad_norm": 0.2500713943206808, + "learning_rate": 7.017408516453365e-05, + "loss": 1.1061, + "step": 500 + }, + { + "epoch": 0.92, + "grad_norm": 0.2812266276040743, + "learning_rate": 7.011805028606064e-05, + "loss": 1.1949, + "step": 501 + }, + { + "epoch": 0.92, + "grad_norm": 0.298829667668083, + "learning_rate": 7.006187859329544e-05, + "loss": 1.2313, + "step": 502 + }, + { + "epoch": 0.92, + "grad_norm": 0.26518768159745104, + "learning_rate": 7.000557034140361e-05, + "loss": 1.2246, + "step": 503 + }, + { + "epoch": 0.92, + "grad_norm": 0.3037280360760458, + "learning_rate": 6.994912578617113e-05, + "loss": 1.1617, + "step": 504 + }, + { + "epoch": 0.92, + "grad_norm": 0.2726903109255714, + "learning_rate": 6.989254518400309e-05, + "loss": 1.2415, + "step": 505 + }, + { + "epoch": 0.93, + "grad_norm": 0.25568082003046966, + "learning_rate": 6.98358287919226e-05, + "loss": 1.1817, + "step": 506 + }, + { + "epoch": 0.93, + "grad_norm": 0.25633294893705044, + "learning_rate": 6.97789768675696e-05, + "loss": 1.2149, + "step": 507 + }, + { + "epoch": 0.93, + "grad_norm": 0.28291439435087123, + "learning_rate": 6.972198966919972e-05, + "loss": 1.1578, + "step": 508 + }, + { + "epoch": 0.93, + "grad_norm": 0.27195184756655516, + "learning_rate": 6.966486745568308e-05, + "loss": 1.2355, + "step": 509 + }, + { + "epoch": 0.93, + "grad_norm": 0.239159568376005, + "learning_rate": 6.960761048650312e-05, + "loss": 1.1688, + "step": 510 + }, + { + "epoch": 0.93, + "grad_norm": 0.22961475425949177, + "learning_rate": 6.955021902175543e-05, + "loss": 1.2094, + "step": 511 + }, + { + "epoch": 0.94, + "grad_norm": 0.27443773600741117, + "learning_rate": 6.949269332214651e-05, + "loss": 1.2559, + "step": 512 + }, + { + "epoch": 0.94, + "grad_norm": 0.26230551832002097, + "learning_rate": 6.94350336489927e-05, + "loss": 1.2121, + "step": 513 + }, + { + "epoch": 0.94, + "grad_norm": 0.2716742985303849, + "learning_rate": 6.937724026421892e-05, + "loss": 1.2444, + "step": 514 + }, + { + "epoch": 0.94, + "grad_norm": 0.2537850139439542, + "learning_rate": 6.931931343035742e-05, + "loss": 1.1327, + "step": 515 + }, + { + "epoch": 0.94, + "grad_norm": 0.28599587967496826, + "learning_rate": 6.926125341054676e-05, + "loss": 1.2236, + "step": 516 + }, + { + "epoch": 0.95, + "grad_norm": 0.26780654378470103, + "learning_rate": 6.920306046853043e-05, + "loss": 1.2295, + "step": 517 + }, + { + "epoch": 0.95, + "grad_norm": 0.23606296888412015, + "learning_rate": 6.914473486865577e-05, + "loss": 1.1543, + "step": 518 + }, + { + "epoch": 0.95, + "grad_norm": 0.34976881174240837, + "learning_rate": 6.90862768758727e-05, + "loss": 1.2067, + "step": 519 + }, + { + "epoch": 0.95, + "grad_norm": 0.2481257873494882, + "learning_rate": 6.902768675573258e-05, + "loss": 1.2188, + "step": 520 + }, + { + "epoch": 0.95, + "grad_norm": 0.2996395778117021, + "learning_rate": 6.896896477438699e-05, + "loss": 1.2326, + "step": 521 + }, + { + "epoch": 0.95, + "grad_norm": 0.8839768816333193, + "learning_rate": 6.891011119858643e-05, + "loss": 1.2435, + "step": 522 + }, + { + "epoch": 0.96, + "grad_norm": 0.2851882482058998, + "learning_rate": 6.885112629567927e-05, + "loss": 1.2644, + "step": 523 + }, + { + "epoch": 0.96, + "grad_norm": 0.2813663482913699, + "learning_rate": 6.879201033361035e-05, + "loss": 1.2309, + "step": 524 + }, + { + "epoch": 0.96, + "grad_norm": 0.3257551560135454, + "learning_rate": 6.873276358091996e-05, + "loss": 1.2755, + "step": 525 + }, + { + "epoch": 0.96, + "grad_norm": 0.28930479952494365, + "learning_rate": 6.867338630674247e-05, + "loss": 1.1962, + "step": 526 + }, + { + "epoch": 0.96, + "grad_norm": 0.3077462996938649, + "learning_rate": 6.861387878080511e-05, + "loss": 1.2402, + "step": 527 + }, + { + "epoch": 0.97, + "grad_norm": 0.2848900193452761, + "learning_rate": 6.855424127342688e-05, + "loss": 1.2748, + "step": 528 + }, + { + "epoch": 0.97, + "grad_norm": 0.4765938812802202, + "learning_rate": 6.849447405551718e-05, + "loss": 1.2226, + "step": 529 + }, + { + "epoch": 0.97, + "grad_norm": 0.53184473292579, + "learning_rate": 6.843457739857467e-05, + "loss": 1.2347, + "step": 530 + }, + { + "epoch": 0.97, + "grad_norm": 0.6416239346492343, + "learning_rate": 6.837455157468596e-05, + "loss": 1.2429, + "step": 531 + }, + { + "epoch": 0.97, + "grad_norm": 0.3188092712502773, + "learning_rate": 6.831439685652442e-05, + "loss": 1.216, + "step": 532 + }, + { + "epoch": 0.97, + "grad_norm": 0.3527495731006385, + "learning_rate": 6.825411351734895e-05, + "loss": 1.1682, + "step": 533 + }, + { + "epoch": 0.98, + "grad_norm": 0.29603753744741856, + "learning_rate": 6.819370183100274e-05, + "loss": 1.1434, + "step": 534 + }, + { + "epoch": 0.98, + "grad_norm": 0.5252450389976622, + "learning_rate": 6.813316207191198e-05, + "loss": 1.1943, + "step": 535 + }, + { + "epoch": 0.98, + "grad_norm": 0.32999419558659937, + "learning_rate": 6.807249451508466e-05, + "loss": 1.192, + "step": 536 + }, + { + "epoch": 0.98, + "grad_norm": 0.3650175469778724, + "learning_rate": 6.801169943610929e-05, + "loss": 1.2141, + "step": 537 + }, + { + "epoch": 0.98, + "grad_norm": 1.0643532150783557, + "learning_rate": 6.795077711115368e-05, + "loss": 1.2253, + "step": 538 + }, + { + "epoch": 0.99, + "grad_norm": 0.5041310609130145, + "learning_rate": 6.788972781696363e-05, + "loss": 1.278, + "step": 539 + }, + { + "epoch": 0.99, + "grad_norm": 0.5123058164360991, + "learning_rate": 6.782855183086177e-05, + "loss": 1.2231, + "step": 540 + }, + { + "epoch": 0.99, + "grad_norm": 0.3533015702394419, + "learning_rate": 6.776724943074619e-05, + "loss": 1.2072, + "step": 541 + }, + { + "epoch": 0.99, + "grad_norm": 0.30253964625417207, + "learning_rate": 6.770582089508927e-05, + "loss": 1.1382, + "step": 542 + }, + { + "epoch": 0.99, + "grad_norm": 0.348991618828202, + "learning_rate": 6.764426650293633e-05, + "loss": 1.2079, + "step": 543 + }, + { + "epoch": 0.99, + "grad_norm": 0.46017440578788743, + "learning_rate": 6.758258653390444e-05, + "loss": 1.1813, + "step": 544 + }, + { + "epoch": 1.0, + "grad_norm": 0.31962101755594885, + "learning_rate": 6.75207812681811e-05, + "loss": 1.1339, + "step": 545 + }, + { + "epoch": 1.0, + "grad_norm": 0.37092024548285923, + "learning_rate": 6.745885098652298e-05, + "loss": 1.2591, + "step": 546 + }, + { + "epoch": 1.0, + "grad_norm": 0.32347106450715835, + "learning_rate": 6.739679597025466e-05, + "loss": 1.2017, + "step": 547 + }, + { + "epoch": 1.0, + "grad_norm": 0.39250187112342494, + "learning_rate": 6.733461650126733e-05, + "loss": 1.0933, + "step": 548 + }, + { + "epoch": 1.0, + "grad_norm": 0.473522452217324, + "learning_rate": 6.727231286201752e-05, + "loss": 1.1124, + "step": 549 + }, + { + "epoch": 1.01, + "grad_norm": 0.4809062179622052, + "learning_rate": 6.720988533552582e-05, + "loss": 1.1585, + "step": 550 + }, + { + "epoch": 1.01, + "grad_norm": 0.3529662801059162, + "learning_rate": 6.714733420537559e-05, + "loss": 1.0501, + "step": 551 + }, + { + "epoch": 1.01, + "grad_norm": 0.5958247214391118, + "learning_rate": 6.708465975571168e-05, + "loss": 1.1086, + "step": 552 + }, + { + "epoch": 1.01, + "grad_norm": 0.5341364205022454, + "learning_rate": 6.70218622712391e-05, + "loss": 1.0518, + "step": 553 + }, + { + "epoch": 1.01, + "grad_norm": 0.3601805724462006, + "learning_rate": 6.695894203722181e-05, + "loss": 1.1779, + "step": 554 + }, + { + "epoch": 1.02, + "grad_norm": 0.43410190338280613, + "learning_rate": 6.68958993394813e-05, + "loss": 1.093, + "step": 555 + }, + { + "epoch": 1.02, + "grad_norm": 0.46217742572873594, + "learning_rate": 6.683273446439546e-05, + "loss": 1.0117, + "step": 556 + }, + { + "epoch": 1.02, + "grad_norm": 0.8591682373623357, + "learning_rate": 6.676944769889708e-05, + "loss": 1.1002, + "step": 557 + }, + { + "epoch": 1.02, + "grad_norm": 0.7383229487622726, + "learning_rate": 6.670603933047272e-05, + "loss": 1.0779, + "step": 558 + }, + { + "epoch": 1.02, + "grad_norm": 0.5965305891207813, + "learning_rate": 6.664250964716131e-05, + "loss": 1.0889, + "step": 559 + }, + { + "epoch": 1.02, + "grad_norm": 0.6030858606684543, + "learning_rate": 6.657885893755288e-05, + "loss": 1.0982, + "step": 560 + }, + { + "epoch": 1.03, + "grad_norm": 0.4644510682398409, + "learning_rate": 6.65150874907872e-05, + "loss": 1.1004, + "step": 561 + }, + { + "epoch": 1.03, + "grad_norm": 0.43943285132452564, + "learning_rate": 6.645119559655254e-05, + "loss": 1.0536, + "step": 562 + }, + { + "epoch": 1.03, + "grad_norm": 0.4456395978600012, + "learning_rate": 6.638718354508427e-05, + "loss": 1.0733, + "step": 563 + }, + { + "epoch": 1.03, + "grad_norm": 0.3303824433217466, + "learning_rate": 6.632305162716365e-05, + "loss": 1.0552, + "step": 564 + }, + { + "epoch": 1.03, + "grad_norm": 0.3617704823170143, + "learning_rate": 6.62588001341164e-05, + "loss": 1.1092, + "step": 565 + }, + { + "epoch": 1.04, + "grad_norm": 0.4465013349903427, + "learning_rate": 6.619442935781141e-05, + "loss": 1.0781, + "step": 566 + }, + { + "epoch": 1.04, + "grad_norm": 0.48516780613791277, + "learning_rate": 6.612993959065947e-05, + "loss": 1.0686, + "step": 567 + }, + { + "epoch": 1.04, + "grad_norm": 0.38867820318536633, + "learning_rate": 6.606533112561186e-05, + "loss": 1.1215, + "step": 568 + }, + { + "epoch": 1.04, + "grad_norm": 0.38566119820378336, + "learning_rate": 6.600060425615907e-05, + "loss": 1.1213, + "step": 569 + }, + { + "epoch": 1.04, + "grad_norm": 0.35534855445058544, + "learning_rate": 6.593575927632947e-05, + "loss": 1.0955, + "step": 570 + }, + { + "epoch": 1.04, + "grad_norm": 0.38124406233349717, + "learning_rate": 6.587079648068795e-05, + "loss": 1.0659, + "step": 571 + }, + { + "epoch": 1.05, + "grad_norm": 0.454750160923548, + "learning_rate": 6.580571616433457e-05, + "loss": 1.1149, + "step": 572 + }, + { + "epoch": 1.05, + "grad_norm": 0.35353190088025255, + "learning_rate": 6.574051862290325e-05, + "loss": 1.0388, + "step": 573 + }, + { + "epoch": 1.05, + "grad_norm": 0.3249395594793626, + "learning_rate": 6.567520415256045e-05, + "loss": 1.0784, + "step": 574 + }, + { + "epoch": 1.05, + "grad_norm": 0.40078898818247227, + "learning_rate": 6.560977305000375e-05, + "loss": 1.0859, + "step": 575 + }, + { + "epoch": 1.05, + "grad_norm": 0.4115264795060035, + "learning_rate": 6.554422561246054e-05, + "loss": 1.1828, + "step": 576 + }, + { + "epoch": 1.06, + "grad_norm": 0.30090229228069215, + "learning_rate": 6.54785621376867e-05, + "loss": 1.0901, + "step": 577 + }, + { + "epoch": 1.06, + "grad_norm": 0.28827860350299206, + "learning_rate": 6.541278292396523e-05, + "loss": 1.0277, + "step": 578 + }, + { + "epoch": 1.06, + "grad_norm": 0.34690404488996757, + "learning_rate": 6.534688827010484e-05, + "loss": 1.048, + "step": 579 + }, + { + "epoch": 1.06, + "grad_norm": 0.29943113556644785, + "learning_rate": 6.528087847543867e-05, + "loss": 1.0646, + "step": 580 + }, + { + "epoch": 1.06, + "grad_norm": 0.37318202575874415, + "learning_rate": 6.521475383982291e-05, + "loss": 1.1091, + "step": 581 + }, + { + "epoch": 1.06, + "grad_norm": 0.3049663659203959, + "learning_rate": 6.51485146636354e-05, + "loss": 1.0552, + "step": 582 + }, + { + "epoch": 1.07, + "grad_norm": 0.3342407867509692, + "learning_rate": 6.508216124777431e-05, + "loss": 1.2227, + "step": 583 + }, + { + "epoch": 1.07, + "grad_norm": 0.3348396047855952, + "learning_rate": 6.501569389365674e-05, + "loss": 1.0861, + "step": 584 + }, + { + "epoch": 1.07, + "grad_norm": 0.30951429367513383, + "learning_rate": 6.494911290321737e-05, + "loss": 1.0461, + "step": 585 + }, + { + "epoch": 1.07, + "grad_norm": 0.33898401361064606, + "learning_rate": 6.488241857890711e-05, + "loss": 1.0854, + "step": 586 + }, + { + "epoch": 1.07, + "grad_norm": 0.4901462068263497, + "learning_rate": 6.481561122369164e-05, + "loss": 1.1012, + "step": 587 + }, + { + "epoch": 1.08, + "grad_norm": 0.3179574879809652, + "learning_rate": 6.474869114105018e-05, + "loss": 1.0451, + "step": 588 + }, + { + "epoch": 1.08, + "grad_norm": 0.32159328915060714, + "learning_rate": 6.468165863497395e-05, + "loss": 1.0458, + "step": 589 + }, + { + "epoch": 1.08, + "grad_norm": 0.36462235008537297, + "learning_rate": 6.461451400996491e-05, + "loss": 1.1247, + "step": 590 + }, + { + "epoch": 1.08, + "grad_norm": 0.5373862753611778, + "learning_rate": 6.454725757103432e-05, + "loss": 1.0542, + "step": 591 + }, + { + "epoch": 1.08, + "grad_norm": 0.3160409270291303, + "learning_rate": 6.447988962370133e-05, + "loss": 1.0829, + "step": 592 + }, + { + "epoch": 1.08, + "grad_norm": 0.390452102978435, + "learning_rate": 6.441241047399169e-05, + "loss": 1.192, + "step": 593 + }, + { + "epoch": 1.09, + "grad_norm": 0.3802122712014928, + "learning_rate": 6.434482042843627e-05, + "loss": 1.1153, + "step": 594 + }, + { + "epoch": 1.09, + "grad_norm": 0.4081584328242501, + "learning_rate": 6.427711979406966e-05, + "loss": 1.1635, + "step": 595 + }, + { + "epoch": 1.09, + "grad_norm": 0.3791962989638633, + "learning_rate": 6.420930887842889e-05, + "loss": 1.1581, + "step": 596 + }, + { + "epoch": 1.09, + "grad_norm": 0.33239440056484193, + "learning_rate": 6.414138798955189e-05, + "loss": 1.0926, + "step": 597 + }, + { + "epoch": 1.09, + "grad_norm": 0.3279881540815014, + "learning_rate": 6.407335743597616e-05, + "loss": 1.1386, + "step": 598 + }, + { + "epoch": 1.1, + "grad_norm": 0.30309644763750837, + "learning_rate": 6.40052175267374e-05, + "loss": 1.0523, + "step": 599 + }, + { + "epoch": 1.1, + "grad_norm": 0.3349097308403333, + "learning_rate": 6.393696857136801e-05, + "loss": 1.0815, + "step": 600 + }, + { + "epoch": 1.1, + "grad_norm": 0.3288227593556618, + "learning_rate": 6.386861087989581e-05, + "loss": 1.015, + "step": 601 + }, + { + "epoch": 1.1, + "grad_norm": 0.36685586740843157, + "learning_rate": 6.380014476284255e-05, + "loss": 1.1232, + "step": 602 + }, + { + "epoch": 1.1, + "grad_norm": 0.3620977714204643, + "learning_rate": 6.373157053122243e-05, + "loss": 1.1138, + "step": 603 + }, + { + "epoch": 1.1, + "grad_norm": 0.3130587018197183, + "learning_rate": 6.366288849654091e-05, + "loss": 1.1255, + "step": 604 + }, + { + "epoch": 1.11, + "grad_norm": 0.3602737087072766, + "learning_rate": 6.359409897079303e-05, + "loss": 1.0282, + "step": 605 + }, + { + "epoch": 1.11, + "grad_norm": 0.31168852571991945, + "learning_rate": 6.352520226646222e-05, + "loss": 1.0779, + "step": 606 + }, + { + "epoch": 1.11, + "grad_norm": 0.3516045580189353, + "learning_rate": 6.345619869651871e-05, + "loss": 1.1028, + "step": 607 + }, + { + "epoch": 1.11, + "grad_norm": 0.3231857927563657, + "learning_rate": 6.33870885744182e-05, + "loss": 1.1202, + "step": 608 + }, + { + "epoch": 1.11, + "grad_norm": 0.30205205129701157, + "learning_rate": 6.331787221410041e-05, + "loss": 1.1369, + "step": 609 + }, + { + "epoch": 1.12, + "grad_norm": 0.3198359813888166, + "learning_rate": 6.32485499299877e-05, + "loss": 1.1763, + "step": 610 + }, + { + "epoch": 1.12, + "grad_norm": 0.3128641370321787, + "learning_rate": 6.31791220369835e-05, + "loss": 1.0223, + "step": 611 + }, + { + "epoch": 1.12, + "grad_norm": 0.2989105616213649, + "learning_rate": 6.31095888504711e-05, + "loss": 1.0358, + "step": 612 + }, + { + "epoch": 1.12, + "grad_norm": 0.3103537906853337, + "learning_rate": 6.303995068631203e-05, + "loss": 1.1261, + "step": 613 + }, + { + "epoch": 1.12, + "grad_norm": 0.28598715532508207, + "learning_rate": 6.297020786084467e-05, + "loss": 1.0629, + "step": 614 + }, + { + "epoch": 1.12, + "grad_norm": 0.29809789918093255, + "learning_rate": 6.290036069088288e-05, + "loss": 1.035, + "step": 615 + }, + { + "epoch": 1.13, + "grad_norm": 0.33765270252261453, + "learning_rate": 6.283040949371451e-05, + "loss": 1.1221, + "step": 616 + }, + { + "epoch": 1.13, + "grad_norm": 0.3424617501293415, + "learning_rate": 6.276035458709993e-05, + "loss": 1.155, + "step": 617 + }, + { + "epoch": 1.13, + "grad_norm": 0.3799189737987811, + "learning_rate": 6.269019628927067e-05, + "loss": 1.0701, + "step": 618 + }, + { + "epoch": 1.13, + "grad_norm": 0.3358898935253196, + "learning_rate": 6.261993491892791e-05, + "loss": 1.1649, + "step": 619 + }, + { + "epoch": 1.13, + "grad_norm": 0.31569979424117356, + "learning_rate": 6.254957079524099e-05, + "loss": 1.0633, + "step": 620 + }, + { + "epoch": 1.14, + "grad_norm": 0.3002168156888237, + "learning_rate": 6.247910423784609e-05, + "loss": 1.0846, + "step": 621 + }, + { + "epoch": 1.14, + "grad_norm": 0.3097238823450595, + "learning_rate": 6.24085355668447e-05, + "loss": 1.0808, + "step": 622 + }, + { + "epoch": 1.14, + "grad_norm": 0.3120312761417578, + "learning_rate": 6.233786510280212e-05, + "loss": 1.0142, + "step": 623 + }, + { + "epoch": 1.14, + "grad_norm": 0.3335343015064923, + "learning_rate": 6.22670931667461e-05, + "loss": 1.0674, + "step": 624 + }, + { + "epoch": 1.14, + "grad_norm": 0.3234062304634526, + "learning_rate": 6.219622008016533e-05, + "loss": 1.0981, + "step": 625 + }, + { + "epoch": 1.14, + "grad_norm": 0.32152678786547273, + "learning_rate": 6.212524616500798e-05, + "loss": 1.0244, + "step": 626 + }, + { + "epoch": 1.15, + "grad_norm": 0.39031977608147594, + "learning_rate": 6.205417174368023e-05, + "loss": 1.1205, + "step": 627 + }, + { + "epoch": 1.15, + "grad_norm": 0.3806189090017157, + "learning_rate": 6.198299713904485e-05, + "loss": 1.1134, + "step": 628 + }, + { + "epoch": 1.15, + "grad_norm": 0.2978349276971668, + "learning_rate": 6.191172267441967e-05, + "loss": 1.0088, + "step": 629 + }, + { + "epoch": 1.15, + "grad_norm": 0.3190354077382501, + "learning_rate": 6.184034867357617e-05, + "loss": 1.108, + "step": 630 + }, + { + "epoch": 1.15, + "grad_norm": 0.32633048665038994, + "learning_rate": 6.176887546073797e-05, + "loss": 1.0825, + "step": 631 + }, + { + "epoch": 1.16, + "grad_norm": 0.3428026413020903, + "learning_rate": 6.169730336057939e-05, + "loss": 1.0765, + "step": 632 + }, + { + "epoch": 1.16, + "grad_norm": 0.3475737151929015, + "learning_rate": 6.162563269822391e-05, + "loss": 1.0693, + "step": 633 + }, + { + "epoch": 1.16, + "grad_norm": 0.3870252154591392, + "learning_rate": 6.15538637992428e-05, + "loss": 1.1081, + "step": 634 + }, + { + "epoch": 1.16, + "grad_norm": 0.33597355193652834, + "learning_rate": 6.148199698965352e-05, + "loss": 1.0893, + "step": 635 + }, + { + "epoch": 1.16, + "grad_norm": 0.30805894179787247, + "learning_rate": 6.141003259591834e-05, + "loss": 1.0995, + "step": 636 + }, + { + "epoch": 1.17, + "grad_norm": 0.3025073882734066, + "learning_rate": 6.133797094494281e-05, + "loss": 1.0388, + "step": 637 + }, + { + "epoch": 1.17, + "grad_norm": 0.3524395196391662, + "learning_rate": 6.126581236407429e-05, + "loss": 1.1196, + "step": 638 + }, + { + "epoch": 1.17, + "grad_norm": 0.3377646188130345, + "learning_rate": 6.119355718110039e-05, + "loss": 1.0382, + "step": 639 + }, + { + "epoch": 1.17, + "grad_norm": 0.35508400659785483, + "learning_rate": 6.112120572424763e-05, + "loss": 1.1402, + "step": 640 + }, + { + "epoch": 1.17, + "grad_norm": 0.3454418793700457, + "learning_rate": 6.104875832217982e-05, + "loss": 1.1032, + "step": 641 + }, + { + "epoch": 1.17, + "grad_norm": 0.32629806837059866, + "learning_rate": 6.097621530399661e-05, + "loss": 1.0959, + "step": 642 + }, + { + "epoch": 1.18, + "grad_norm": 0.3329536837751315, + "learning_rate": 6.090357699923202e-05, + "loss": 1.0467, + "step": 643 + }, + { + "epoch": 1.18, + "grad_norm": 0.32302233828349475, + "learning_rate": 6.083084373785287e-05, + "loss": 1.0858, + "step": 644 + }, + { + "epoch": 1.18, + "grad_norm": 0.3310358826507611, + "learning_rate": 6.075801585025739e-05, + "loss": 1.0715, + "step": 645 + }, + { + "epoch": 1.18, + "grad_norm": 0.319322035854079, + "learning_rate": 6.068509366727362e-05, + "loss": 1.177, + "step": 646 + }, + { + "epoch": 1.18, + "grad_norm": 0.3065230667302707, + "learning_rate": 6.061207752015797e-05, + "loss": 1.0649, + "step": 647 + }, + { + "epoch": 1.19, + "grad_norm": 0.29926795565748227, + "learning_rate": 6.053896774059368e-05, + "loss": 1.1325, + "step": 648 + }, + { + "epoch": 1.19, + "grad_norm": 0.3556069634279046, + "learning_rate": 6.046576466068931e-05, + "loss": 1.1366, + "step": 649 + }, + { + "epoch": 1.19, + "grad_norm": 0.3189191131461966, + "learning_rate": 6.039246861297727e-05, + "loss": 1.0693, + "step": 650 + }, + { + "epoch": 1.19, + "grad_norm": 0.3347197156648834, + "learning_rate": 6.031907993041227e-05, + "loss": 1.1009, + "step": 651 + }, + { + "epoch": 1.19, + "grad_norm": 0.32274156348185445, + "learning_rate": 6.0245598946369826e-05, + "loss": 1.1675, + "step": 652 + }, + { + "epoch": 1.19, + "grad_norm": 0.35534089035455224, + "learning_rate": 6.017202599464476e-05, + "loss": 1.1723, + "step": 653 + }, + { + "epoch": 1.2, + "grad_norm": 0.3106026578570133, + "learning_rate": 6.009836140944965e-05, + "loss": 1.0954, + "step": 654 + }, + { + "epoch": 1.2, + "grad_norm": 0.3309144454564729, + "learning_rate": 6.002460552541331e-05, + "loss": 1.0209, + "step": 655 + }, + { + "epoch": 1.2, + "grad_norm": 0.3023619281400003, + "learning_rate": 5.9950758677579345e-05, + "loss": 1.0363, + "step": 656 + }, + { + "epoch": 1.2, + "grad_norm": 0.3311182880219704, + "learning_rate": 5.987682120140451e-05, + "loss": 1.0515, + "step": 657 + }, + { + "epoch": 1.2, + "grad_norm": 0.33396486010030413, + "learning_rate": 5.980279343275729e-05, + "loss": 1.1251, + "step": 658 + }, + { + "epoch": 1.21, + "grad_norm": 0.3465764556678002, + "learning_rate": 5.97286757079163e-05, + "loss": 1.165, + "step": 659 + }, + { + "epoch": 1.21, + "grad_norm": 0.304193441363374, + "learning_rate": 5.965446836356882e-05, + "loss": 1.0228, + "step": 660 + }, + { + "epoch": 1.21, + "grad_norm": 0.3415149030413082, + "learning_rate": 5.9580171736809224e-05, + "loss": 1.0742, + "step": 661 + }, + { + "epoch": 1.21, + "grad_norm": 0.33138658321132064, + "learning_rate": 5.950578616513746e-05, + "loss": 1.0843, + "step": 662 + }, + { + "epoch": 1.21, + "grad_norm": 0.30774403421162994, + "learning_rate": 5.943131198645752e-05, + "loss": 1.065, + "step": 663 + }, + { + "epoch": 1.21, + "grad_norm": 0.3428877492183819, + "learning_rate": 5.9356749539075885e-05, + "loss": 1.1101, + "step": 664 + }, + { + "epoch": 1.22, + "grad_norm": 0.3621290546130101, + "learning_rate": 5.928209916170003e-05, + "loss": 1.1372, + "step": 665 + }, + { + "epoch": 1.22, + "grad_norm": 0.3482375945469884, + "learning_rate": 5.9207361193436865e-05, + "loss": 1.132, + "step": 666 + }, + { + "epoch": 1.22, + "grad_norm": 0.31754384974068384, + "learning_rate": 5.9132535973791156e-05, + "loss": 1.148, + "step": 667 + }, + { + "epoch": 1.22, + "grad_norm": 0.36003834782050365, + "learning_rate": 5.9057623842664044e-05, + "loss": 1.1099, + "step": 668 + }, + { + "epoch": 1.22, + "grad_norm": 0.2963701622969662, + "learning_rate": 5.8982625140351464e-05, + "loss": 1.0755, + "step": 669 + }, + { + "epoch": 1.23, + "grad_norm": 0.32579569606066516, + "learning_rate": 5.8907540207542616e-05, + "loss": 1.0809, + "step": 670 + }, + { + "epoch": 1.23, + "grad_norm": 0.4247563451753457, + "learning_rate": 5.8832369385318416e-05, + "loss": 1.097, + "step": 671 + }, + { + "epoch": 1.23, + "grad_norm": 0.33076932102169776, + "learning_rate": 5.875711301514992e-05, + "loss": 1.1078, + "step": 672 + }, + { + "epoch": 1.23, + "grad_norm": 0.3609238032332309, + "learning_rate": 5.8681771438896815e-05, + "loss": 1.1031, + "step": 673 + }, + { + "epoch": 1.23, + "grad_norm": 0.325159585649425, + "learning_rate": 5.860634499880583e-05, + "loss": 1.0707, + "step": 674 + }, + { + "epoch": 1.23, + "grad_norm": 0.4620687271068983, + "learning_rate": 5.853083403750922e-05, + "loss": 1.1017, + "step": 675 + }, + { + "epoch": 1.24, + "grad_norm": 0.33485279064365936, + "learning_rate": 5.845523889802316e-05, + "loss": 1.0989, + "step": 676 + }, + { + "epoch": 1.24, + "grad_norm": 0.30952573170841513, + "learning_rate": 5.8379559923746214e-05, + "loss": 1.0393, + "step": 677 + }, + { + "epoch": 1.24, + "grad_norm": 0.33498605810588283, + "learning_rate": 5.830379745845781e-05, + "loss": 1.1259, + "step": 678 + }, + { + "epoch": 1.24, + "grad_norm": 0.35771921163037307, + "learning_rate": 5.822795184631659e-05, + "loss": 1.0815, + "step": 679 + }, + { + "epoch": 1.24, + "grad_norm": 0.3329650192347647, + "learning_rate": 5.815202343185894e-05, + "loss": 1.1344, + "step": 680 + }, + { + "epoch": 1.25, + "grad_norm": 0.3356634465845771, + "learning_rate": 5.807601255999736e-05, + "loss": 1.1297, + "step": 681 + }, + { + "epoch": 1.25, + "grad_norm": 0.3289442034151235, + "learning_rate": 5.7999919576018934e-05, + "loss": 1.022, + "step": 682 + }, + { + "epoch": 1.25, + "grad_norm": 0.3207007334784113, + "learning_rate": 5.7923744825583745e-05, + "loss": 1.0571, + "step": 683 + }, + { + "epoch": 1.25, + "grad_norm": 0.3582460325329284, + "learning_rate": 5.7847488654723304e-05, + "loss": 1.0778, + "step": 684 + }, + { + "epoch": 1.25, + "grad_norm": 0.3563317666176927, + "learning_rate": 5.777115140983899e-05, + "loss": 1.1003, + "step": 685 + }, + { + "epoch": 1.25, + "grad_norm": 3.4694912945702105, + "learning_rate": 5.769473343770047e-05, + "loss": 1.121, + "step": 686 + }, + { + "epoch": 1.26, + "grad_norm": 0.43002349520483113, + "learning_rate": 5.761823508544411e-05, + "loss": 1.0765, + "step": 687 + }, + { + "epoch": 1.26, + "grad_norm": 0.39467783104839754, + "learning_rate": 5.754165670057142e-05, + "loss": 1.0788, + "step": 688 + }, + { + "epoch": 1.26, + "grad_norm": 0.39629029674867916, + "learning_rate": 5.7464998630947464e-05, + "loss": 1.0812, + "step": 689 + }, + { + "epoch": 1.26, + "grad_norm": 0.3880152093965208, + "learning_rate": 5.738826122479929e-05, + "loss": 1.1228, + "step": 690 + }, + { + "epoch": 1.26, + "grad_norm": 0.3777874121959188, + "learning_rate": 5.7311444830714324e-05, + "loss": 1.0907, + "step": 691 + }, + { + "epoch": 1.27, + "grad_norm": 0.38004041653523696, + "learning_rate": 5.723454979763882e-05, + "loss": 1.1263, + "step": 692 + }, + { + "epoch": 1.27, + "grad_norm": 0.37049672627797636, + "learning_rate": 5.7157576474876246e-05, + "loss": 1.1438, + "step": 693 + }, + { + "epoch": 1.27, + "grad_norm": 0.32973606103437614, + "learning_rate": 5.7080525212085725e-05, + "loss": 1.0553, + "step": 694 + }, + { + "epoch": 1.27, + "grad_norm": 0.31674639252070325, + "learning_rate": 5.700339635928038e-05, + "loss": 1.06, + "step": 695 + }, + { + "epoch": 1.27, + "grad_norm": 0.32282199426553837, + "learning_rate": 5.692619026682588e-05, + "loss": 1.0841, + "step": 696 + }, + { + "epoch": 1.27, + "grad_norm": 0.4810882958061859, + "learning_rate": 5.684890728543869e-05, + "loss": 1.0803, + "step": 697 + }, + { + "epoch": 1.28, + "grad_norm": 0.3995638550178378, + "learning_rate": 5.6771547766184566e-05, + "loss": 1.1187, + "step": 698 + }, + { + "epoch": 1.28, + "grad_norm": 0.35264932960583484, + "learning_rate": 5.669411206047699e-05, + "loss": 1.0641, + "step": 699 + }, + { + "epoch": 1.28, + "grad_norm": 0.35240640524733, + "learning_rate": 5.661660052007547e-05, + "loss": 1.076, + "step": 700 + }, + { + "epoch": 1.28, + "grad_norm": 0.3540694609860389, + "learning_rate": 5.653901349708401e-05, + "loss": 1.1369, + "step": 701 + }, + { + "epoch": 1.28, + "grad_norm": 0.3196055112925304, + "learning_rate": 5.646135134394955e-05, + "loss": 1.0677, + "step": 702 + }, + { + "epoch": 1.29, + "grad_norm": 0.4214141007955914, + "learning_rate": 5.6383614413460266e-05, + "loss": 1.1139, + "step": 703 + }, + { + "epoch": 1.29, + "grad_norm": 0.3625611311798579, + "learning_rate": 5.630580305874402e-05, + "loss": 1.1845, + "step": 704 + }, + { + "epoch": 1.29, + "grad_norm": 0.3425208672181188, + "learning_rate": 5.62279176332668e-05, + "loss": 1.174, + "step": 705 + }, + { + "epoch": 1.29, + "grad_norm": 0.3108419862818321, + "learning_rate": 5.6149958490830996e-05, + "loss": 1.0331, + "step": 706 + }, + { + "epoch": 1.29, + "grad_norm": 0.3274644181571904, + "learning_rate": 5.607192598557394e-05, + "loss": 1.0664, + "step": 707 + }, + { + "epoch": 1.29, + "grad_norm": 0.346218197215145, + "learning_rate": 5.599382047196617e-05, + "loss": 1.2088, + "step": 708 + }, + { + "epoch": 1.3, + "grad_norm": 0.328497632267458, + "learning_rate": 5.591564230480989e-05, + "loss": 1.0287, + "step": 709 + }, + { + "epoch": 1.3, + "grad_norm": 0.3708173720611468, + "learning_rate": 5.583739183923732e-05, + "loss": 1.0883, + "step": 710 + }, + { + "epoch": 1.3, + "grad_norm": 0.3631427403535479, + "learning_rate": 5.575906943070915e-05, + "loss": 1.1155, + "step": 711 + }, + { + "epoch": 1.3, + "grad_norm": 0.3305201458598695, + "learning_rate": 5.5680675435012834e-05, + "loss": 1.0958, + "step": 712 + }, + { + "epoch": 1.3, + "grad_norm": 0.34978833532083714, + "learning_rate": 5.5602210208261036e-05, + "loss": 1.1437, + "step": 713 + }, + { + "epoch": 1.31, + "grad_norm": 0.3510553882510229, + "learning_rate": 5.552367410688999e-05, + "loss": 1.0941, + "step": 714 + }, + { + "epoch": 1.31, + "grad_norm": 0.3523747462465078, + "learning_rate": 5.544506748765789e-05, + "loss": 1.1289, + "step": 715 + }, + { + "epoch": 1.31, + "grad_norm": 0.38262637783927445, + "learning_rate": 5.5366390707643266e-05, + "loss": 1.099, + "step": 716 + }, + { + "epoch": 1.31, + "grad_norm": 0.38620065989073454, + "learning_rate": 5.528764412424334e-05, + "loss": 1.083, + "step": 717 + }, + { + "epoch": 1.31, + "grad_norm": 0.3401355276121096, + "learning_rate": 5.520882809517245e-05, + "loss": 1.028, + "step": 718 + }, + { + "epoch": 1.32, + "grad_norm": 0.3392061008943934, + "learning_rate": 5.512994297846039e-05, + "loss": 1.1083, + "step": 719 + }, + { + "epoch": 1.32, + "grad_norm": 0.34219480421015414, + "learning_rate": 5.505098913245077e-05, + "loss": 1.1108, + "step": 720 + }, + { + "epoch": 1.32, + "grad_norm": 0.3275058061553761, + "learning_rate": 5.497196691579945e-05, + "loss": 1.111, + "step": 721 + }, + { + "epoch": 1.32, + "grad_norm": 0.36800249746509384, + "learning_rate": 5.489287668747283e-05, + "loss": 1.1221, + "step": 722 + }, + { + "epoch": 1.32, + "grad_norm": 0.4129005533101575, + "learning_rate": 5.481371880674628e-05, + "loss": 1.0966, + "step": 723 + }, + { + "epoch": 1.32, + "grad_norm": 0.36563906596251655, + "learning_rate": 5.4734493633202505e-05, + "loss": 1.0927, + "step": 724 + }, + { + "epoch": 1.33, + "grad_norm": 0.3614650536839971, + "learning_rate": 5.465520152672986e-05, + "loss": 1.13, + "step": 725 + }, + { + "epoch": 1.33, + "grad_norm": 0.36419665098633497, + "learning_rate": 5.4575842847520765e-05, + "loss": 1.1183, + "step": 726 + }, + { + "epoch": 1.33, + "grad_norm": 0.34490689807258995, + "learning_rate": 5.449641795607005e-05, + "loss": 1.0919, + "step": 727 + }, + { + "epoch": 1.33, + "grad_norm": 0.3627643746876298, + "learning_rate": 5.441692721317334e-05, + "loss": 1.0411, + "step": 728 + }, + { + "epoch": 1.33, + "grad_norm": 0.323620411949565, + "learning_rate": 5.433737097992537e-05, + "loss": 1.0725, + "step": 729 + }, + { + "epoch": 1.34, + "grad_norm": 0.3521599501824965, + "learning_rate": 5.425774961771838e-05, + "loss": 1.0926, + "step": 730 + }, + { + "epoch": 1.34, + "grad_norm": 0.3302390546764222, + "learning_rate": 5.417806348824047e-05, + "loss": 1.0468, + "step": 731 + }, + { + "epoch": 1.34, + "grad_norm": 0.3833325802616019, + "learning_rate": 5.4098312953473956e-05, + "loss": 1.1291, + "step": 732 + }, + { + "epoch": 1.34, + "grad_norm": 0.3708621126835512, + "learning_rate": 5.401849837569372e-05, + "loss": 1.0887, + "step": 733 + }, + { + "epoch": 1.34, + "grad_norm": 0.3625834373416278, + "learning_rate": 5.393862011746555e-05, + "loss": 1.0981, + "step": 734 + }, + { + "epoch": 1.34, + "grad_norm": 0.3583343965080617, + "learning_rate": 5.385867854164451e-05, + "loss": 1.1021, + "step": 735 + }, + { + "epoch": 1.35, + "grad_norm": 0.34598320594096066, + "learning_rate": 5.377867401137332e-05, + "loss": 1.1376, + "step": 736 + }, + { + "epoch": 1.35, + "grad_norm": 0.3046382791315433, + "learning_rate": 5.369860689008066e-05, + "loss": 1.0206, + "step": 737 + }, + { + "epoch": 1.35, + "grad_norm": 0.34464948380043725, + "learning_rate": 5.3618477541479505e-05, + "loss": 1.1084, + "step": 738 + }, + { + "epoch": 1.35, + "grad_norm": 0.3203242519627101, + "learning_rate": 5.353828632956557e-05, + "loss": 1.0731, + "step": 739 + }, + { + "epoch": 1.35, + "grad_norm": 0.3431169960355163, + "learning_rate": 5.3458033618615516e-05, + "loss": 1.091, + "step": 740 + }, + { + "epoch": 1.36, + "grad_norm": 0.33492074521678705, + "learning_rate": 5.337771977318543e-05, + "loss": 1.1112, + "step": 741 + }, + { + "epoch": 1.36, + "grad_norm": 0.32576546585541344, + "learning_rate": 5.3297345158109086e-05, + "loss": 1.0993, + "step": 742 + }, + { + "epoch": 1.36, + "grad_norm": 0.3410007245037574, + "learning_rate": 5.3216910138496286e-05, + "loss": 1.094, + "step": 743 + }, + { + "epoch": 1.36, + "grad_norm": 0.34891180680896833, + "learning_rate": 5.313641507973128e-05, + "loss": 1.1331, + "step": 744 + }, + { + "epoch": 1.36, + "grad_norm": 0.37135766946717214, + "learning_rate": 5.3055860347471006e-05, + "loss": 1.1, + "step": 745 + }, + { + "epoch": 1.36, + "grad_norm": 0.3465019415478411, + "learning_rate": 5.297524630764349e-05, + "loss": 1.1256, + "step": 746 + }, + { + "epoch": 1.37, + "grad_norm": 0.37035388481626563, + "learning_rate": 5.289457332644615e-05, + "loss": 1.0366, + "step": 747 + }, + { + "epoch": 1.37, + "grad_norm": 0.33853883270759155, + "learning_rate": 5.281384177034421e-05, + "loss": 1.0547, + "step": 748 + }, + { + "epoch": 1.37, + "grad_norm": 0.364306618627317, + "learning_rate": 5.2733052006068897e-05, + "loss": 1.0768, + "step": 749 + }, + { + "epoch": 1.37, + "grad_norm": 0.4021754315731627, + "learning_rate": 5.2652204400615916e-05, + "loss": 1.1382, + "step": 750 + }, + { + "epoch": 1.37, + "grad_norm": 0.3332185389039008, + "learning_rate": 5.257129932124368e-05, + "loss": 1.0815, + "step": 751 + }, + { + "epoch": 1.38, + "grad_norm": 0.3453105709879854, + "learning_rate": 5.249033713547173e-05, + "loss": 1.1109, + "step": 752 + }, + { + "epoch": 1.38, + "grad_norm": 0.3385397539717797, + "learning_rate": 5.2409318211078966e-05, + "loss": 1.0529, + "step": 753 + }, + { + "epoch": 1.38, + "grad_norm": 0.33197994450130447, + "learning_rate": 5.232824291610206e-05, + "loss": 1.0721, + "step": 754 + }, + { + "epoch": 1.38, + "grad_norm": 0.32836289576124167, + "learning_rate": 5.224711161883375e-05, + "loss": 1.0459, + "step": 755 + }, + { + "epoch": 1.38, + "grad_norm": 0.32491620058831744, + "learning_rate": 5.216592468782117e-05, + "loss": 1.0897, + "step": 756 + }, + { + "epoch": 1.38, + "grad_norm": 0.3137879047811153, + "learning_rate": 5.2084682491864155e-05, + "loss": 1.096, + "step": 757 + }, + { + "epoch": 1.39, + "grad_norm": 0.3356938043023012, + "learning_rate": 5.200338540001364e-05, + "loss": 1.0827, + "step": 758 + }, + { + "epoch": 1.39, + "grad_norm": 0.36044340490819055, + "learning_rate": 5.192203378156984e-05, + "loss": 1.0617, + "step": 759 + }, + { + "epoch": 1.39, + "grad_norm": 0.34674262047888293, + "learning_rate": 5.184062800608077e-05, + "loss": 1.1267, + "step": 760 + }, + { + "epoch": 1.39, + "grad_norm": 0.32469442322149333, + "learning_rate": 5.1759168443340375e-05, + "loss": 1.1483, + "step": 761 + }, + { + "epoch": 1.39, + "grad_norm": 0.3290384307774216, + "learning_rate": 5.167765546338698e-05, + "loss": 1.047, + "step": 762 + }, + { + "epoch": 1.4, + "grad_norm": 0.31637612188770403, + "learning_rate": 5.1596089436501525e-05, + "loss": 1.0311, + "step": 763 + }, + { + "epoch": 1.4, + "grad_norm": 0.3168693829641207, + "learning_rate": 5.151447073320597e-05, + "loss": 1.1405, + "step": 764 + }, + { + "epoch": 1.4, + "grad_norm": 0.34322421571238926, + "learning_rate": 5.143279972426153e-05, + "loss": 1.1428, + "step": 765 + }, + { + "epoch": 1.4, + "grad_norm": 0.3291030435830325, + "learning_rate": 5.1351076780667026e-05, + "loss": 1.0473, + "step": 766 + }, + { + "epoch": 1.4, + "grad_norm": 0.33772039158758044, + "learning_rate": 5.1269302273657195e-05, + "loss": 1.0909, + "step": 767 + }, + { + "epoch": 1.4, + "grad_norm": 0.3802031736890876, + "learning_rate": 5.118747657470102e-05, + "loss": 1.1482, + "step": 768 + }, + { + "epoch": 1.41, + "grad_norm": 0.3296067628997962, + "learning_rate": 5.1105600055500025e-05, + "loss": 1.0085, + "step": 769 + }, + { + "epoch": 1.41, + "grad_norm": 0.3707139982828035, + "learning_rate": 5.102367308798658e-05, + "loss": 1.0746, + "step": 770 + }, + { + "epoch": 1.41, + "grad_norm": 0.3378537316757011, + "learning_rate": 5.094169604432225e-05, + "loss": 1.0482, + "step": 771 + }, + { + "epoch": 1.41, + "grad_norm": 0.4008417246255145, + "learning_rate": 5.085966929689601e-05, + "loss": 1.1065, + "step": 772 + }, + { + "epoch": 1.41, + "grad_norm": 0.3244385106988064, + "learning_rate": 5.077759321832271e-05, + "loss": 1.0827, + "step": 773 + }, + { + "epoch": 1.42, + "grad_norm": 0.37228575732812336, + "learning_rate": 5.0695468181441215e-05, + "loss": 1.1146, + "step": 774 + }, + { + "epoch": 1.42, + "grad_norm": 0.33761714797540276, + "learning_rate": 5.061329455931283e-05, + "loss": 1.092, + "step": 775 + }, + { + "epoch": 1.42, + "grad_norm": 0.3158158390913494, + "learning_rate": 5.053107272521955e-05, + "loss": 1.1058, + "step": 776 + }, + { + "epoch": 1.42, + "grad_norm": 0.3691501929738938, + "learning_rate": 5.044880305266239e-05, + "loss": 1.1599, + "step": 777 + }, + { + "epoch": 1.42, + "grad_norm": 0.33730914019805525, + "learning_rate": 5.0366485915359645e-05, + "loss": 1.0615, + "step": 778 + }, + { + "epoch": 1.42, + "grad_norm": 0.34970059240017, + "learning_rate": 5.0284121687245257e-05, + "loss": 1.1475, + "step": 779 + }, + { + "epoch": 1.43, + "grad_norm": 0.3374028029407197, + "learning_rate": 5.020171074246707e-05, + "loss": 1.0926, + "step": 780 + }, + { + "epoch": 1.43, + "grad_norm": 0.3350020681123992, + "learning_rate": 5.011925345538514e-05, + "loss": 1.1276, + "step": 781 + }, + { + "epoch": 1.43, + "grad_norm": 0.3224228965786606, + "learning_rate": 5.003675020057003e-05, + "loss": 1.0183, + "step": 782 + }, + { + "epoch": 1.43, + "grad_norm": 0.3357310714740298, + "learning_rate": 4.995420135280114e-05, + "loss": 1.1114, + "step": 783 + }, + { + "epoch": 1.43, + "grad_norm": 0.3590203255363759, + "learning_rate": 4.9871607287064966e-05, + "loss": 1.1504, + "step": 784 + }, + { + "epoch": 1.44, + "grad_norm": 0.33011195419611655, + "learning_rate": 4.9788968378553396e-05, + "loss": 1.0826, + "step": 785 + }, + { + "epoch": 1.44, + "grad_norm": 0.31088868195439445, + "learning_rate": 4.970628500266207e-05, + "loss": 1.0704, + "step": 786 + }, + { + "epoch": 1.44, + "grad_norm": 0.3144996103179409, + "learning_rate": 4.962355753498858e-05, + "loss": 1.1403, + "step": 787 + }, + { + "epoch": 1.44, + "grad_norm": 0.3147269555419068, + "learning_rate": 4.954078635133081e-05, + "loss": 1.0898, + "step": 788 + }, + { + "epoch": 1.44, + "grad_norm": 0.3280151747783868, + "learning_rate": 4.945797182768524e-05, + "loss": 1.1115, + "step": 789 + }, + { + "epoch": 1.44, + "grad_norm": 0.3551996569232493, + "learning_rate": 4.937511434024524e-05, + "loss": 1.1731, + "step": 790 + }, + { + "epoch": 1.45, + "grad_norm": 0.343863208057807, + "learning_rate": 4.9292214265399336e-05, + "loss": 1.0866, + "step": 791 + }, + { + "epoch": 1.45, + "grad_norm": 0.37316699385322466, + "learning_rate": 4.920927197972949e-05, + "loss": 1.1083, + "step": 792 + }, + { + "epoch": 1.45, + "grad_norm": 0.3635739774067832, + "learning_rate": 4.9126287860009453e-05, + "loss": 1.1393, + "step": 793 + }, + { + "epoch": 1.45, + "grad_norm": 0.3755910554972886, + "learning_rate": 4.9043262283202974e-05, + "loss": 1.1624, + "step": 794 + }, + { + "epoch": 1.45, + "grad_norm": 0.3635899120146823, + "learning_rate": 4.8960195626462145e-05, + "loss": 1.2095, + "step": 795 + }, + { + "epoch": 1.46, + "grad_norm": 0.3642202684342816, + "learning_rate": 4.8877088267125664e-05, + "loss": 1.1099, + "step": 796 + }, + { + "epoch": 1.46, + "grad_norm": 0.3339946548799316, + "learning_rate": 4.879394058271712e-05, + "loss": 1.1157, + "step": 797 + }, + { + "epoch": 1.46, + "grad_norm": 0.3457189703100475, + "learning_rate": 4.871075295094329e-05, + "loss": 1.129, + "step": 798 + }, + { + "epoch": 1.46, + "grad_norm": 0.3550931839691424, + "learning_rate": 4.862752574969241e-05, + "loss": 1.076, + "step": 799 + }, + { + "epoch": 1.46, + "grad_norm": 0.36139108917966734, + "learning_rate": 4.8544259357032475e-05, + "loss": 1.1577, + "step": 800 + }, + { + "epoch": 1.0, + "grad_norm": 0.39569703665247874, + "learning_rate": 4.8460954151209486e-05, + "loss": 1.0543, + "step": 801 + }, + { + "epoch": 1.0, + "grad_norm": 0.3879033670170866, + "learning_rate": 4.837761051064579e-05, + "loss": 1.0688, + "step": 802 + }, + { + "epoch": 1.01, + "grad_norm": 0.3796846713967255, + "learning_rate": 4.8294228813938285e-05, + "loss": 0.9911, + "step": 803 + }, + { + "epoch": 1.01, + "grad_norm": 0.4007831430409375, + "learning_rate": 4.8210809439856804e-05, + "loss": 1.0126, + "step": 804 + }, + { + "epoch": 1.01, + "grad_norm": 0.37588078665500885, + "learning_rate": 4.8127352767342276e-05, + "loss": 0.9302, + "step": 805 + }, + { + "epoch": 1.01, + "grad_norm": 0.4078509175013281, + "learning_rate": 4.8043859175505095e-05, + "loss": 0.9982, + "step": 806 + }, + { + "epoch": 1.01, + "grad_norm": 0.379096046185539, + "learning_rate": 4.7960329043623344e-05, + "loss": 1.0035, + "step": 807 + }, + { + "epoch": 1.01, + "grad_norm": 0.3813938568133554, + "learning_rate": 4.787676275114111e-05, + "loss": 0.9579, + "step": 808 + }, + { + "epoch": 1.02, + "grad_norm": 0.3686863564511168, + "learning_rate": 4.779316067766673e-05, + "loss": 1.0105, + "step": 809 + }, + { + "epoch": 1.02, + "grad_norm": 0.4263940878847523, + "learning_rate": 4.770952320297109e-05, + "loss": 1.0677, + "step": 810 + }, + { + "epoch": 1.02, + "grad_norm": 0.37178778374665006, + "learning_rate": 4.7625850706985886e-05, + "loss": 1.0019, + "step": 811 + }, + { + "epoch": 1.02, + "grad_norm": 0.36803355429187945, + "learning_rate": 4.7542143569801894e-05, + "loss": 0.9937, + "step": 812 + }, + { + "epoch": 1.02, + "grad_norm": 0.3897072472941179, + "learning_rate": 4.745840217166725e-05, + "loss": 1.0877, + "step": 813 + }, + { + "epoch": 1.03, + "grad_norm": 0.35571833841716255, + "learning_rate": 4.737462689298577e-05, + "loss": 1.0015, + "step": 814 + }, + { + "epoch": 1.03, + "grad_norm": 0.38930229991094323, + "learning_rate": 4.7290818114315086e-05, + "loss": 1.028, + "step": 815 + }, + { + "epoch": 1.03, + "grad_norm": 0.411005007105147, + "learning_rate": 4.72069762163651e-05, + "loss": 1.0068, + "step": 816 + }, + { + "epoch": 1.03, + "grad_norm": 0.3980240190337736, + "learning_rate": 4.7123101579996106e-05, + "loss": 0.9919, + "step": 817 + }, + { + "epoch": 1.03, + "grad_norm": 0.36369517703115467, + "learning_rate": 4.7039194586217136e-05, + "loss": 0.967, + "step": 818 + }, + { + "epoch": 1.03, + "grad_norm": 0.38591148840458894, + "learning_rate": 4.695525561618418e-05, + "loss": 0.9743, + "step": 819 + }, + { + "epoch": 1.04, + "grad_norm": 0.45873135108949337, + "learning_rate": 4.687128505119853e-05, + "loss": 1.0516, + "step": 820 + }, + { + "epoch": 1.04, + "grad_norm": 0.3866330351411308, + "learning_rate": 4.6787283272704966e-05, + "loss": 0.9939, + "step": 821 + }, + { + "epoch": 1.04, + "grad_norm": 0.4620340173291326, + "learning_rate": 4.670325066229009e-05, + "loss": 1.0526, + "step": 822 + }, + { + "epoch": 1.04, + "grad_norm": 0.38877454299870284, + "learning_rate": 4.661918760168052e-05, + "loss": 0.9904, + "step": 823 + }, + { + "epoch": 1.04, + "grad_norm": 0.3880489386116793, + "learning_rate": 4.653509447274121e-05, + "loss": 0.9623, + "step": 824 + }, + { + "epoch": 1.05, + "grad_norm": 0.3827392356186151, + "learning_rate": 4.6450971657473743e-05, + "loss": 1.0772, + "step": 825 + }, + { + "epoch": 1.05, + "grad_norm": 0.4132814641854327, + "learning_rate": 4.63668195380145e-05, + "loss": 1.0533, + "step": 826 + }, + { + "epoch": 1.05, + "grad_norm": 0.3703610182402835, + "learning_rate": 4.628263849663301e-05, + "loss": 0.9336, + "step": 827 + }, + { + "epoch": 1.05, + "grad_norm": 0.4152053683299823, + "learning_rate": 4.619842891573016e-05, + "loss": 0.9801, + "step": 828 + }, + { + "epoch": 1.05, + "grad_norm": 0.41791059043554274, + "learning_rate": 4.6114191177836514e-05, + "loss": 1.0617, + "step": 829 + }, + { + "epoch": 1.05, + "grad_norm": 0.46363896517299136, + "learning_rate": 4.6029925665610524e-05, + "loss": 0.9687, + "step": 830 + }, + { + "epoch": 1.06, + "grad_norm": 0.41141959057512445, + "learning_rate": 4.59456327618368e-05, + "loss": 1.0965, + "step": 831 + }, + { + "epoch": 1.06, + "grad_norm": 0.3789192764519836, + "learning_rate": 4.5861312849424386e-05, + "loss": 0.9793, + "step": 832 + }, + { + "epoch": 1.06, + "grad_norm": 0.4047291581107866, + "learning_rate": 4.5776966311405035e-05, + "loss": 1.0342, + "step": 833 + }, + { + "epoch": 1.06, + "grad_norm": 0.4425157400959256, + "learning_rate": 4.5692593530931416e-05, + "loss": 1.0892, + "step": 834 + }, + { + "epoch": 1.06, + "grad_norm": 0.3707332144806616, + "learning_rate": 4.560819489127545e-05, + "loss": 0.9815, + "step": 835 + }, + { + "epoch": 1.07, + "grad_norm": 0.3897444102572823, + "learning_rate": 4.552377077582646e-05, + "loss": 0.9884, + "step": 836 + }, + { + "epoch": 1.07, + "grad_norm": 0.42725787957019346, + "learning_rate": 4.543932156808959e-05, + "loss": 0.9972, + "step": 837 + }, + { + "epoch": 1.07, + "grad_norm": 0.40615269781820007, + "learning_rate": 4.535484765168386e-05, + "loss": 0.9529, + "step": 838 + }, + { + "epoch": 1.07, + "grad_norm": 0.3505829736050887, + "learning_rate": 4.527034941034063e-05, + "loss": 0.9492, + "step": 839 + }, + { + "epoch": 1.07, + "grad_norm": 0.36688064686440497, + "learning_rate": 4.51858272279017e-05, + "loss": 0.9592, + "step": 840 + }, + { + "epoch": 1.07, + "grad_norm": 0.4043468777955929, + "learning_rate": 4.5101281488317634e-05, + "loss": 1.048, + "step": 841 + }, + { + "epoch": 1.08, + "grad_norm": 0.3811489793242706, + "learning_rate": 4.501671257564602e-05, + "loss": 1.0138, + "step": 842 + }, + { + "epoch": 1.08, + "grad_norm": 0.39813004142325986, + "learning_rate": 4.49321208740497e-05, + "loss": 1.071, + "step": 843 + }, + { + "epoch": 1.08, + "grad_norm": 0.3809751022095503, + "learning_rate": 4.484750676779504e-05, + "loss": 1.0351, + "step": 844 + }, + { + "epoch": 1.08, + "grad_norm": 0.384312178013823, + "learning_rate": 4.4762870641250185e-05, + "loss": 0.9737, + "step": 845 + }, + { + "epoch": 1.08, + "grad_norm": 0.40769404907923557, + "learning_rate": 4.467821287888331e-05, + "loss": 0.9659, + "step": 846 + }, + { + "epoch": 1.09, + "grad_norm": 0.39594136851937817, + "learning_rate": 4.459353386526086e-05, + "loss": 0.9405, + "step": 847 + }, + { + "epoch": 1.09, + "grad_norm": 0.37180161011562185, + "learning_rate": 4.450883398504584e-05, + "loss": 1.0732, + "step": 848 + }, + { + "epoch": 1.09, + "grad_norm": 0.3772603623154663, + "learning_rate": 4.442411362299602e-05, + "loss": 0.9646, + "step": 849 + }, + { + "epoch": 1.09, + "grad_norm": 0.4346142368506476, + "learning_rate": 4.433937316396224e-05, + "loss": 0.9572, + "step": 850 + }, + { + "epoch": 1.09, + "grad_norm": 0.3997258084612474, + "learning_rate": 4.425461299288659e-05, + "loss": 0.9492, + "step": 851 + }, + { + "epoch": 1.1, + "grad_norm": 0.41245476865247155, + "learning_rate": 4.416983349480073e-05, + "loss": 0.8732, + "step": 852 + }, + { + "epoch": 1.1, + "grad_norm": 0.6761499297939195, + "learning_rate": 4.408503505482412e-05, + "loss": 1.0425, + "step": 853 + }, + { + "epoch": 1.1, + "grad_norm": 0.40340979486858985, + "learning_rate": 4.400021805816225e-05, + "loss": 0.9596, + "step": 854 + }, + { + "epoch": 1.1, + "grad_norm": 0.43290732392699666, + "learning_rate": 4.391538289010493e-05, + "loss": 1.0123, + "step": 855 + }, + { + "epoch": 1.1, + "grad_norm": 0.36878054442190156, + "learning_rate": 4.383052993602448e-05, + "loss": 0.9448, + "step": 856 + }, + { + "epoch": 1.1, + "grad_norm": 0.7146145128961262, + "learning_rate": 4.374565958137404e-05, + "loss": 1.0342, + "step": 857 + }, + { + "epoch": 1.11, + "grad_norm": 0.44429357586145607, + "learning_rate": 4.3660772211685775e-05, + "loss": 1.0436, + "step": 858 + }, + { + "epoch": 1.11, + "grad_norm": 0.4565751973640598, + "learning_rate": 4.357586821256918e-05, + "loss": 1.0311, + "step": 859 + }, + { + "epoch": 1.11, + "grad_norm": 0.3919991236654277, + "learning_rate": 4.349094796970925e-05, + "loss": 1.1401, + "step": 860 + }, + { + "epoch": 1.11, + "grad_norm": 0.4347441949284011, + "learning_rate": 4.3406011868864795e-05, + "loss": 1.0252, + "step": 861 + }, + { + "epoch": 1.11, + "grad_norm": 0.38339976027415407, + "learning_rate": 4.3321060295866635e-05, + "loss": 1.0536, + "step": 862 + }, + { + "epoch": 1.12, + "grad_norm": 0.37688790408195166, + "learning_rate": 4.32360936366159e-05, + "loss": 1.012, + "step": 863 + }, + { + "epoch": 1.12, + "grad_norm": 0.4317538207582504, + "learning_rate": 4.315111227708224e-05, + "loss": 1.0505, + "step": 864 + }, + { + "epoch": 1.12, + "grad_norm": 0.4145324872228796, + "learning_rate": 4.306611660330208e-05, + "loss": 1.0496, + "step": 865 + }, + { + "epoch": 1.12, + "grad_norm": 0.416535227064448, + "learning_rate": 4.298110700137687e-05, + "loss": 0.9628, + "step": 866 + }, + { + "epoch": 1.12, + "grad_norm": 0.46564356187492717, + "learning_rate": 4.2896083857471345e-05, + "loss": 1.0016, + "step": 867 + }, + { + "epoch": 1.12, + "grad_norm": 0.4228980941889828, + "learning_rate": 4.281104755781172e-05, + "loss": 1.0904, + "step": 868 + }, + { + "epoch": 1.13, + "grad_norm": 0.4267821214430208, + "learning_rate": 4.272599848868402e-05, + "loss": 1.0544, + "step": 869 + }, + { + "epoch": 1.13, + "grad_norm": 0.45763332095792075, + "learning_rate": 4.264093703643223e-05, + "loss": 1.0686, + "step": 870 + }, + { + "epoch": 1.13, + "grad_norm": 0.4347555516548761, + "learning_rate": 4.255586358745662e-05, + "loss": 1.0264, + "step": 871 + }, + { + "epoch": 1.13, + "grad_norm": 0.3817726381103066, + "learning_rate": 4.247077852821194e-05, + "loss": 1.0045, + "step": 872 + }, + { + "epoch": 1.13, + "grad_norm": 0.3882808845457995, + "learning_rate": 4.2385682245205685e-05, + "loss": 1.0193, + "step": 873 + }, + { + "epoch": 1.14, + "grad_norm": 0.39410930252966775, + "learning_rate": 4.230057512499634e-05, + "loss": 0.9832, + "step": 874 + }, + { + "epoch": 1.14, + "grad_norm": 0.4373094593907156, + "learning_rate": 4.221545755419159e-05, + "loss": 1.0343, + "step": 875 + }, + { + "epoch": 1.14, + "grad_norm": 0.4462843721698891, + "learning_rate": 4.2130329919446646e-05, + "loss": 1.0324, + "step": 876 + }, + { + "epoch": 1.14, + "grad_norm": 0.4747274247448112, + "learning_rate": 4.20451926074624e-05, + "loss": 0.9903, + "step": 877 + }, + { + "epoch": 1.14, + "grad_norm": 0.4157472897596409, + "learning_rate": 4.196004600498369e-05, + "loss": 0.9266, + "step": 878 + }, + { + "epoch": 1.14, + "grad_norm": 0.41625958088960685, + "learning_rate": 4.1874890498797605e-05, + "loss": 0.9658, + "step": 879 + }, + { + "epoch": 1.15, + "grad_norm": 0.44784944130574333, + "learning_rate": 4.178972647573163e-05, + "loss": 0.9671, + "step": 880 + }, + { + "epoch": 1.15, + "grad_norm": 0.4116839177956385, + "learning_rate": 4.1704554322651975e-05, + "loss": 0.9591, + "step": 881 + }, + { + "epoch": 1.15, + "grad_norm": 0.4025569857639452, + "learning_rate": 4.161937442646176e-05, + "loss": 1.0072, + "step": 882 + }, + { + "epoch": 1.15, + "grad_norm": 0.41518478124763597, + "learning_rate": 4.1534187174099285e-05, + "loss": 1.0275, + "step": 883 + }, + { + "epoch": 1.15, + "grad_norm": 0.3987815564664466, + "learning_rate": 4.1448992952536275e-05, + "loss": 1.0039, + "step": 884 + }, + { + "epoch": 1.16, + "grad_norm": 0.4270378155679982, + "learning_rate": 4.136379214877609e-05, + "loss": 1.0369, + "step": 885 + }, + { + "epoch": 1.16, + "grad_norm": 0.42144733922972777, + "learning_rate": 4.127858514985203e-05, + "loss": 1.0269, + "step": 886 + }, + { + "epoch": 1.16, + "grad_norm": 0.4198664438272548, + "learning_rate": 4.1193372342825494e-05, + "loss": 1.0427, + "step": 887 + }, + { + "epoch": 1.16, + "grad_norm": 0.3985048256281719, + "learning_rate": 4.1108154114784275e-05, + "loss": 1.0702, + "step": 888 + }, + { + "epoch": 1.16, + "grad_norm": 0.605520808292362, + "learning_rate": 4.102293085284083e-05, + "loss": 0.9749, + "step": 889 + }, + { + "epoch": 1.16, + "grad_norm": 0.4150515863924052, + "learning_rate": 4.0937702944130426e-05, + "loss": 1.0231, + "step": 890 + }, + { + "epoch": 1.17, + "grad_norm": 0.3935997576565283, + "learning_rate": 4.085247077580948e-05, + "loss": 1.0014, + "step": 891 + }, + { + "epoch": 1.17, + "grad_norm": 0.399446131403209, + "learning_rate": 4.076723473505374e-05, + "loss": 0.9602, + "step": 892 + }, + { + "epoch": 1.17, + "grad_norm": 0.4406024397129952, + "learning_rate": 4.068199520905655e-05, + "loss": 1.0425, + "step": 893 + }, + { + "epoch": 1.17, + "grad_norm": 0.4036917571496492, + "learning_rate": 4.059675258502709e-05, + "loss": 0.973, + "step": 894 + }, + { + "epoch": 1.17, + "grad_norm": 0.4057196459433299, + "learning_rate": 4.05115072501886e-05, + "loss": 0.9997, + "step": 895 + }, + { + "epoch": 1.18, + "grad_norm": 0.4374124954708759, + "learning_rate": 4.0426259591776645e-05, + "loss": 0.9826, + "step": 896 + }, + { + "epoch": 1.18, + "grad_norm": 0.4545699371285546, + "learning_rate": 4.0341009997037356e-05, + "loss": 1.0554, + "step": 897 + }, + { + "epoch": 1.18, + "grad_norm": 0.4251917031237376, + "learning_rate": 4.025575885322563e-05, + "loss": 1.0217, + "step": 898 + }, + { + "epoch": 1.18, + "grad_norm": 0.3857651901893941, + "learning_rate": 4.0170506547603427e-05, + "loss": 1.0317, + "step": 899 + }, + { + "epoch": 1.18, + "grad_norm": 0.46323573798490897, + "learning_rate": 4.008525346743797e-05, + "loss": 1.0398, + "step": 900 + }, + { + "epoch": 1.18, + "grad_norm": 0.4011541121460918, + "learning_rate": 4e-05, + "loss": 1.0706, + "step": 901 + }, + { + "epoch": 1.19, + "grad_norm": 0.46493281221028004, + "learning_rate": 3.991474653256204e-05, + "loss": 1.0525, + "step": 902 + }, + { + "epoch": 1.19, + "grad_norm": 0.41683080924539023, + "learning_rate": 3.982949345239658e-05, + "loss": 1.0905, + "step": 903 + }, + { + "epoch": 1.19, + "grad_norm": 0.4750350025014512, + "learning_rate": 3.974424114677437e-05, + "loss": 1.049, + "step": 904 + }, + { + "epoch": 1.19, + "grad_norm": 0.3867445073614702, + "learning_rate": 3.965899000296266e-05, + "loss": 0.9624, + "step": 905 + }, + { + "epoch": 1.19, + "grad_norm": 0.378387661131469, + "learning_rate": 3.957374040822335e-05, + "loss": 1.0223, + "step": 906 + }, + { + "epoch": 1.2, + "grad_norm": 0.3905996390559077, + "learning_rate": 3.948849274981141e-05, + "loss": 1.0315, + "step": 907 + }, + { + "epoch": 1.2, + "grad_norm": 0.4139717689498189, + "learning_rate": 3.940324741497291e-05, + "loss": 0.9297, + "step": 908 + }, + { + "epoch": 1.2, + "grad_norm": 0.39086355684921514, + "learning_rate": 3.9318004790943465e-05, + "loss": 0.9684, + "step": 909 + }, + { + "epoch": 1.2, + "grad_norm": 0.4334915643736419, + "learning_rate": 3.923276526494627e-05, + "loss": 0.996, + "step": 910 + }, + { + "epoch": 1.2, + "grad_norm": 0.40782018986229496, + "learning_rate": 3.9147529224190536e-05, + "loss": 1.0875, + "step": 911 + }, + { + "epoch": 1.2, + "grad_norm": 0.43578702386625723, + "learning_rate": 3.906229705586959e-05, + "loss": 1.1214, + "step": 912 + }, + { + "epoch": 1.21, + "grad_norm": 0.414945683409524, + "learning_rate": 3.89770691471592e-05, + "loss": 1.1037, + "step": 913 + }, + { + "epoch": 1.21, + "grad_norm": 0.40665801579679106, + "learning_rate": 3.889184588521573e-05, + "loss": 0.9743, + "step": 914 + }, + { + "epoch": 1.21, + "grad_norm": 0.4064250611574517, + "learning_rate": 3.880662765717453e-05, + "loss": 0.8814, + "step": 915 + }, + { + "epoch": 1.21, + "grad_norm": 0.48023046298843347, + "learning_rate": 3.8721414850147985e-05, + "loss": 0.9663, + "step": 916 + }, + { + "epoch": 1.21, + "grad_norm": 0.42358024833566227, + "learning_rate": 3.8636207851223924e-05, + "loss": 1.0491, + "step": 917 + }, + { + "epoch": 1.22, + "grad_norm": 0.41522494786195835, + "learning_rate": 3.855100704746374e-05, + "loss": 1.033, + "step": 918 + }, + { + "epoch": 1.22, + "grad_norm": 0.40890517696706496, + "learning_rate": 3.8465812825900715e-05, + "loss": 1.0369, + "step": 919 + }, + { + "epoch": 1.22, + "grad_norm": 0.4325851866408538, + "learning_rate": 3.838062557353825e-05, + "loss": 0.9362, + "step": 920 + }, + { + "epoch": 1.22, + "grad_norm": 0.4185860919050069, + "learning_rate": 3.8295445677348025e-05, + "loss": 1.026, + "step": 921 + }, + { + "epoch": 1.22, + "grad_norm": 0.3975762375934804, + "learning_rate": 3.8210273524268375e-05, + "loss": 1.0412, + "step": 922 + }, + { + "epoch": 1.22, + "grad_norm": 0.41725298241987474, + "learning_rate": 3.8125109501202395e-05, + "loss": 1.0004, + "step": 923 + }, + { + "epoch": 1.23, + "grad_norm": 0.455183913149126, + "learning_rate": 3.803995399501632e-05, + "loss": 1.0594, + "step": 924 + }, + { + "epoch": 1.23, + "grad_norm": 0.3993993856483797, + "learning_rate": 3.795480739253761e-05, + "loss": 0.9761, + "step": 925 + }, + { + "epoch": 1.23, + "grad_norm": 0.41638796815161494, + "learning_rate": 3.786967008055337e-05, + "loss": 1.0369, + "step": 926 + }, + { + "epoch": 1.23, + "grad_norm": 0.40015112695810534, + "learning_rate": 3.7784542445808414e-05, + "loss": 1.0271, + "step": 927 + }, + { + "epoch": 1.23, + "grad_norm": 0.3995749494729548, + "learning_rate": 3.769942487500368e-05, + "loss": 1.0613, + "step": 928 + }, + { + "epoch": 1.24, + "grad_norm": 0.4073556267037492, + "learning_rate": 3.761431775479432e-05, + "loss": 1.0528, + "step": 929 + }, + { + "epoch": 1.24, + "grad_norm": 0.44218148822636044, + "learning_rate": 3.752922147178807e-05, + "loss": 1.0742, + "step": 930 + }, + { + "epoch": 1.24, + "grad_norm": 0.4435063485893757, + "learning_rate": 3.744413641254339e-05, + "loss": 1.0825, + "step": 931 + }, + { + "epoch": 1.24, + "grad_norm": 0.46841574994107515, + "learning_rate": 3.735906296356778e-05, + "loss": 1.0471, + "step": 932 + }, + { + "epoch": 1.24, + "grad_norm": 0.40093716627657294, + "learning_rate": 3.727400151131599e-05, + "loss": 1.0474, + "step": 933 + }, + { + "epoch": 1.25, + "grad_norm": 0.3866415067997244, + "learning_rate": 3.71889524421883e-05, + "loss": 1.0209, + "step": 934 + }, + { + "epoch": 1.25, + "grad_norm": 0.4881546110706673, + "learning_rate": 3.710391614252867e-05, + "loss": 1.0768, + "step": 935 + }, + { + "epoch": 1.25, + "grad_norm": 0.4133084639324523, + "learning_rate": 3.701889299862314e-05, + "loss": 1.0423, + "step": 936 + }, + { + "epoch": 1.25, + "grad_norm": 0.40523563084001196, + "learning_rate": 3.6933883396697936e-05, + "loss": 1.005, + "step": 937 + }, + { + "epoch": 1.25, + "grad_norm": 0.38757352418642405, + "learning_rate": 3.684888772291777e-05, + "loss": 0.9659, + "step": 938 + }, + { + "epoch": 1.25, + "grad_norm": 0.421394551890689, + "learning_rate": 3.676390636338411e-05, + "loss": 1.0454, + "step": 939 + }, + { + "epoch": 1.26, + "grad_norm": 0.45693070958342186, + "learning_rate": 3.667893970413337e-05, + "loss": 1.1459, + "step": 940 + }, + { + "epoch": 1.26, + "grad_norm": 0.4172025376377795, + "learning_rate": 3.659398813113522e-05, + "loss": 0.9954, + "step": 941 + }, + { + "epoch": 1.26, + "grad_norm": 0.3871624019510191, + "learning_rate": 3.650905203029075e-05, + "loss": 1.0441, + "step": 942 + }, + { + "epoch": 1.26, + "grad_norm": 0.38541342610032325, + "learning_rate": 3.642413178743083e-05, + "loss": 0.9465, + "step": 943 + }, + { + "epoch": 1.26, + "grad_norm": 0.4208031670525743, + "learning_rate": 3.633922778831423e-05, + "loss": 1.0367, + "step": 944 + }, + { + "epoch": 1.27, + "grad_norm": 0.41867209013040035, + "learning_rate": 3.6254340418625975e-05, + "loss": 1.0868, + "step": 945 + }, + { + "epoch": 1.27, + "grad_norm": 0.431758149074127, + "learning_rate": 3.6169470063975536e-05, + "loss": 1.0689, + "step": 946 + }, + { + "epoch": 1.27, + "grad_norm": 0.4988803338819952, + "learning_rate": 3.608461710989509e-05, + "loss": 1.0879, + "step": 947 + }, + { + "epoch": 1.27, + "grad_norm": 0.4094858411191625, + "learning_rate": 3.5999781941837755e-05, + "loss": 1.0332, + "step": 948 + }, + { + "epoch": 1.27, + "grad_norm": 0.3831847195845155, + "learning_rate": 3.591496494517589e-05, + "loss": 0.9751, + "step": 949 + }, + { + "epoch": 1.27, + "grad_norm": 0.40535692821947267, + "learning_rate": 3.5830166505199284e-05, + "loss": 1.0594, + "step": 950 + }, + { + "epoch": 1.28, + "grad_norm": 0.4875663789389966, + "learning_rate": 3.574538700711343e-05, + "loss": 0.9749, + "step": 951 + }, + { + "epoch": 1.28, + "grad_norm": 0.5155923998285772, + "learning_rate": 3.566062683603778e-05, + "loss": 0.9999, + "step": 952 + }, + { + "epoch": 1.28, + "grad_norm": 0.5280285947816189, + "learning_rate": 3.557588637700399e-05, + "loss": 1.1061, + "step": 953 + }, + { + "epoch": 1.28, + "grad_norm": 0.46573407357796753, + "learning_rate": 3.5491166014954174e-05, + "loss": 1.102, + "step": 954 + }, + { + "epoch": 1.28, + "grad_norm": 0.4122542582865379, + "learning_rate": 3.540646613473915e-05, + "loss": 1.0469, + "step": 955 + }, + { + "epoch": 1.29, + "grad_norm": 0.41414476980823367, + "learning_rate": 3.53217871211167e-05, + "loss": 0.9973, + "step": 956 + }, + { + "epoch": 1.29, + "grad_norm": 0.4030707611608045, + "learning_rate": 3.523712935874983e-05, + "loss": 0.9796, + "step": 957 + }, + { + "epoch": 1.29, + "grad_norm": 0.4235313349747291, + "learning_rate": 3.5152493232204975e-05, + "loss": 1.0601, + "step": 958 + }, + { + "epoch": 1.29, + "grad_norm": 0.4165235178302652, + "learning_rate": 3.5067879125950316e-05, + "loss": 1.0358, + "step": 959 + }, + { + "epoch": 1.29, + "grad_norm": 0.44083984701952955, + "learning_rate": 3.4983287424354e-05, + "loss": 1.0957, + "step": 960 + }, + { + "epoch": 1.29, + "grad_norm": 0.3781161039063518, + "learning_rate": 3.489871851168238e-05, + "loss": 0.9838, + "step": 961 + }, + { + "epoch": 1.3, + "grad_norm": 0.4095747724038915, + "learning_rate": 3.4814172772098314e-05, + "loss": 1.014, + "step": 962 + }, + { + "epoch": 1.3, + "grad_norm": 0.42197119558898466, + "learning_rate": 3.472965058965938e-05, + "loss": 1.0096, + "step": 963 + }, + { + "epoch": 1.3, + "grad_norm": 0.4339963388152155, + "learning_rate": 3.464515234831615e-05, + "loss": 1.0158, + "step": 964 + }, + { + "epoch": 1.3, + "grad_norm": 0.4284638765548976, + "learning_rate": 3.4560678431910424e-05, + "loss": 1.1047, + "step": 965 + }, + { + "epoch": 1.3, + "grad_norm": 0.3935144535755794, + "learning_rate": 3.447622922417355e-05, + "loss": 0.9925, + "step": 966 + }, + { + "epoch": 1.31, + "grad_norm": 0.45884343961025, + "learning_rate": 3.439180510872457e-05, + "loss": 1.0583, + "step": 967 + }, + { + "epoch": 1.31, + "grad_norm": 0.42439320759788374, + "learning_rate": 3.4307406469068604e-05, + "loss": 0.9305, + "step": 968 + }, + { + "epoch": 1.31, + "grad_norm": 0.45770082390324845, + "learning_rate": 3.4223033688594985e-05, + "loss": 1.054, + "step": 969 + }, + { + "epoch": 1.31, + "grad_norm": 0.4284786643981094, + "learning_rate": 3.4138687150575634e-05, + "loss": 0.9409, + "step": 970 + }, + { + "epoch": 1.31, + "grad_norm": 0.41356124058383237, + "learning_rate": 3.4054367238163215e-05, + "loss": 1.0739, + "step": 971 + }, + { + "epoch": 1.31, + "grad_norm": 0.4255832249412624, + "learning_rate": 3.3970074334389496e-05, + "loss": 1.0764, + "step": 972 + }, + { + "epoch": 1.32, + "grad_norm": 0.4337695536142702, + "learning_rate": 3.388580882216349e-05, + "loss": 1.0195, + "step": 973 + }, + { + "epoch": 1.32, + "grad_norm": 0.41363495650922455, + "learning_rate": 3.380157108426985e-05, + "loss": 1.0615, + "step": 974 + }, + { + "epoch": 1.32, + "grad_norm": 0.3950691247686479, + "learning_rate": 3.371736150336701e-05, + "loss": 1.0283, + "step": 975 + }, + { + "epoch": 1.32, + "grad_norm": 0.4042823691555822, + "learning_rate": 3.3633180461985505e-05, + "loss": 1.0309, + "step": 976 + }, + { + "epoch": 1.32, + "grad_norm": 0.3921158850479399, + "learning_rate": 3.354902834252627e-05, + "loss": 1.068, + "step": 977 + }, + { + "epoch": 1.33, + "grad_norm": 0.38349545732725654, + "learning_rate": 3.346490552725879e-05, + "loss": 1.0886, + "step": 978 + }, + { + "epoch": 1.33, + "grad_norm": 0.38689221457248724, + "learning_rate": 3.33808123983195e-05, + "loss": 0.987, + "step": 979 + }, + { + "epoch": 1.33, + "grad_norm": 0.38660550867425647, + "learning_rate": 3.329674933770992e-05, + "loss": 1.069, + "step": 980 + }, + { + "epoch": 1.33, + "grad_norm": 0.3917593746353493, + "learning_rate": 3.321271672729504e-05, + "loss": 0.9858, + "step": 981 + }, + { + "epoch": 1.33, + "grad_norm": 0.4292314072827653, + "learning_rate": 3.3128714948801474e-05, + "loss": 1.0477, + "step": 982 + }, + { + "epoch": 1.33, + "grad_norm": 0.479414638418211, + "learning_rate": 3.3044744383815835e-05, + "loss": 1.0763, + "step": 983 + }, + { + "epoch": 1.34, + "grad_norm": 0.380831894995463, + "learning_rate": 3.2960805413782884e-05, + "loss": 1.0393, + "step": 984 + }, + { + "epoch": 1.34, + "grad_norm": 0.42402274703362114, + "learning_rate": 3.2876898420003914e-05, + "loss": 1.0837, + "step": 985 + }, + { + "epoch": 1.34, + "grad_norm": 0.4571447203722258, + "learning_rate": 3.279302378363491e-05, + "loss": 1.0594, + "step": 986 + }, + { + "epoch": 1.34, + "grad_norm": 0.3776673281658531, + "learning_rate": 3.270918188568493e-05, + "loss": 1.0121, + "step": 987 + }, + { + "epoch": 1.34, + "grad_norm": 0.4367173448132159, + "learning_rate": 3.262537310701425e-05, + "loss": 0.9612, + "step": 988 + }, + { + "epoch": 1.35, + "grad_norm": 0.43679765208840926, + "learning_rate": 3.254159782833276e-05, + "loss": 1.0565, + "step": 989 + }, + { + "epoch": 1.35, + "grad_norm": 0.4018151260013493, + "learning_rate": 3.2457856430198126e-05, + "loss": 0.9975, + "step": 990 + }, + { + "epoch": 1.35, + "grad_norm": 0.40461959940721076, + "learning_rate": 3.237414929301412e-05, + "loss": 1.0255, + "step": 991 + }, + { + "epoch": 1.35, + "grad_norm": 0.41342378541540653, + "learning_rate": 3.2290476797028926e-05, + "loss": 1.024, + "step": 992 + }, + { + "epoch": 1.35, + "grad_norm": 0.3926173909201105, + "learning_rate": 3.220683932233328e-05, + "loss": 1.0877, + "step": 993 + }, + { + "epoch": 1.35, + "grad_norm": 0.3835623199834992, + "learning_rate": 3.21232372488589e-05, + "loss": 1.0992, + "step": 994 + }, + { + "epoch": 1.36, + "grad_norm": 0.39901809497083496, + "learning_rate": 3.2039670956376656e-05, + "loss": 1.0723, + "step": 995 + }, + { + "epoch": 1.36, + "grad_norm": 0.3979604537466272, + "learning_rate": 3.195614082449492e-05, + "loss": 1.0201, + "step": 996 + }, + { + "epoch": 1.36, + "grad_norm": 0.4057122427176845, + "learning_rate": 3.1872647232657723e-05, + "loss": 1.0885, + "step": 997 + }, + { + "epoch": 1.36, + "grad_norm": 0.39747060350754754, + "learning_rate": 3.17891905601432e-05, + "loss": 1.0544, + "step": 998 + }, + { + "epoch": 1.36, + "grad_norm": 0.4397658078291558, + "learning_rate": 3.1705771186061715e-05, + "loss": 1.0998, + "step": 999 + }, + { + "epoch": 1.37, + "grad_norm": 0.37373547663810053, + "learning_rate": 3.162238948935423e-05, + "loss": 1.0465, + "step": 1000 + }, + { + "epoch": 1.0, + "grad_norm": 0.4042576001255747, + "learning_rate": 3.153904584879052e-05, + "loss": 0.9206, + "step": 1001 + }, + { + "epoch": 1.0, + "grad_norm": 0.4042994886900337, + "learning_rate": 3.1455740642967545e-05, + "loss": 0.975, + "step": 1002 + }, + { + "epoch": 1.01, + "grad_norm": 0.4359721725421234, + "learning_rate": 3.1372474250307594e-05, + "loss": 0.9163, + "step": 1003 + }, + { + "epoch": 1.01, + "grad_norm": 0.4886423524029179, + "learning_rate": 3.128924704905673e-05, + "loss": 0.9956, + "step": 1004 + }, + { + "epoch": 1.01, + "grad_norm": 0.48669990170138744, + "learning_rate": 3.1206059417282894e-05, + "loss": 0.9874, + "step": 1005 + }, + { + "epoch": 1.01, + "grad_norm": 0.41954255928633066, + "learning_rate": 3.1122911732874356e-05, + "loss": 0.8986, + "step": 1006 + }, + { + "epoch": 1.01, + "grad_norm": 0.43363878644039366, + "learning_rate": 3.103980437353787e-05, + "loss": 0.9268, + "step": 1007 + }, + { + "epoch": 1.01, + "grad_norm": 0.5199775120765874, + "learning_rate": 3.0956737716797047e-05, + "loss": 0.9341, + "step": 1008 + }, + { + "epoch": 1.02, + "grad_norm": 0.40735757951139595, + "learning_rate": 3.087371213999056e-05, + "loss": 0.9142, + "step": 1009 + }, + { + "epoch": 1.02, + "grad_norm": 0.44449027493884186, + "learning_rate": 3.079072802027051e-05, + "loss": 0.966, + "step": 1010 + }, + { + "epoch": 1.02, + "grad_norm": 0.46590494286419365, + "learning_rate": 3.070778573460068e-05, + "loss": 0.8768, + "step": 1011 + }, + { + "epoch": 1.02, + "grad_norm": 0.45161453051587425, + "learning_rate": 3.062488565975476e-05, + "loss": 0.9299, + "step": 1012 + }, + { + "epoch": 1.02, + "grad_norm": 0.5022364894382346, + "learning_rate": 3.054202817231477e-05, + "loss": 0.9352, + "step": 1013 + }, + { + "epoch": 1.03, + "grad_norm": 0.46443439138730447, + "learning_rate": 3.0459213648669195e-05, + "loss": 0.8913, + "step": 1014 + }, + { + "epoch": 1.03, + "grad_norm": 0.41932307219261455, + "learning_rate": 3.0376442465011436e-05, + "loss": 0.8968, + "step": 1015 + }, + { + "epoch": 1.03, + "grad_norm": 0.42445864358441704, + "learning_rate": 3.0293714997337927e-05, + "loss": 0.8449, + "step": 1016 + }, + { + "epoch": 1.03, + "grad_norm": 0.4489777773688699, + "learning_rate": 3.0211031621446607e-05, + "loss": 0.927, + "step": 1017 + }, + { + "epoch": 1.03, + "grad_norm": 0.45180577504235525, + "learning_rate": 3.0128392712935044e-05, + "loss": 0.8834, + "step": 1018 + }, + { + "epoch": 1.03, + "grad_norm": 0.44680469596106914, + "learning_rate": 3.0045798647198882e-05, + "loss": 1.0176, + "step": 1019 + }, + { + "epoch": 1.04, + "grad_norm": 0.45747851649657734, + "learning_rate": 2.9963249799429986e-05, + "loss": 0.9036, + "step": 1020 + }, + { + "epoch": 1.04, + "grad_norm": 0.5045904501932169, + "learning_rate": 2.988074654461489e-05, + "loss": 1.0475, + "step": 1021 + }, + { + "epoch": 1.04, + "grad_norm": 0.47086144833942983, + "learning_rate": 2.9798289257532946e-05, + "loss": 0.9596, + "step": 1022 + }, + { + "epoch": 1.04, + "grad_norm": 0.4406706196288816, + "learning_rate": 2.9715878312754767e-05, + "loss": 1.0054, + "step": 1023 + }, + { + "epoch": 1.04, + "grad_norm": 0.44584179061175105, + "learning_rate": 2.9633514084640365e-05, + "loss": 0.8981, + "step": 1024 + }, + { + "epoch": 1.05, + "grad_norm": 0.462343843042957, + "learning_rate": 2.955119694733763e-05, + "loss": 0.974, + "step": 1025 + }, + { + "epoch": 1.05, + "grad_norm": 0.46767265335377156, + "learning_rate": 2.946892727478045e-05, + "loss": 1.0063, + "step": 1026 + }, + { + "epoch": 1.05, + "grad_norm": 0.43250194002958803, + "learning_rate": 2.9386705440687168e-05, + "loss": 0.9332, + "step": 1027 + }, + { + "epoch": 1.05, + "grad_norm": 0.44391321845917453, + "learning_rate": 2.9304531818558795e-05, + "loss": 0.8937, + "step": 1028 + }, + { + "epoch": 1.05, + "grad_norm": 0.45616826414927975, + "learning_rate": 2.9222406781677294e-05, + "loss": 0.869, + "step": 1029 + }, + { + "epoch": 1.05, + "grad_norm": 0.5670635396983207, + "learning_rate": 2.9140330703103992e-05, + "loss": 0.9697, + "step": 1030 + }, + { + "epoch": 1.06, + "grad_norm": 0.4860829361401993, + "learning_rate": 2.905830395567776e-05, + "loss": 0.9677, + "step": 1031 + }, + { + "epoch": 1.06, + "grad_norm": 0.4484206829172443, + "learning_rate": 2.8976326912013422e-05, + "loss": 0.9582, + "step": 1032 + }, + { + "epoch": 1.06, + "grad_norm": 0.46728002332884067, + "learning_rate": 2.8894399944499974e-05, + "loss": 0.9023, + "step": 1033 + }, + { + "epoch": 1.06, + "grad_norm": 0.48539702863685763, + "learning_rate": 2.8812523425299e-05, + "loss": 0.9725, + "step": 1034 + }, + { + "epoch": 1.06, + "grad_norm": 0.42521485032555006, + "learning_rate": 2.873069772634281e-05, + "loss": 0.9525, + "step": 1035 + }, + { + "epoch": 1.07, + "grad_norm": 0.4068824768950637, + "learning_rate": 2.8648923219332997e-05, + "loss": 0.8318, + "step": 1036 + }, + { + "epoch": 1.07, + "grad_norm": 0.45227216852040214, + "learning_rate": 2.856720027573848e-05, + "loss": 1.0211, + "step": 1037 + }, + { + "epoch": 1.07, + "grad_norm": 0.42310911927974604, + "learning_rate": 2.8485529266794043e-05, + "loss": 0.9422, + "step": 1038 + }, + { + "epoch": 1.07, + "grad_norm": 0.4494478185683011, + "learning_rate": 2.8403910563498482e-05, + "loss": 0.9577, + "step": 1039 + }, + { + "epoch": 1.07, + "grad_norm": 0.517885146963669, + "learning_rate": 2.832234453661304e-05, + "loss": 0.9551, + "step": 1040 + }, + { + "epoch": 1.07, + "grad_norm": 0.46117112897797924, + "learning_rate": 2.8240831556659635e-05, + "loss": 0.9336, + "step": 1041 + }, + { + "epoch": 1.08, + "grad_norm": 0.4610208217170147, + "learning_rate": 2.815937199391924e-05, + "loss": 0.926, + "step": 1042 + }, + { + "epoch": 1.08, + "grad_norm": 0.445775414660019, + "learning_rate": 2.807796621843016e-05, + "loss": 0.9737, + "step": 1043 + }, + { + "epoch": 1.08, + "grad_norm": 0.46809555676786746, + "learning_rate": 2.799661459998638e-05, + "loss": 0.9916, + "step": 1044 + }, + { + "epoch": 1.08, + "grad_norm": 0.4366867439876077, + "learning_rate": 2.7915317508135848e-05, + "loss": 0.9549, + "step": 1045 + }, + { + "epoch": 1.08, + "grad_norm": 0.388979809570948, + "learning_rate": 2.7834075312178838e-05, + "loss": 0.8967, + "step": 1046 + }, + { + "epoch": 1.09, + "grad_norm": 0.45918748975994583, + "learning_rate": 2.775288838116626e-05, + "loss": 1.032, + "step": 1047 + }, + { + "epoch": 1.09, + "grad_norm": 0.4607131980517622, + "learning_rate": 2.767175708389794e-05, + "loss": 0.9638, + "step": 1048 + }, + { + "epoch": 1.09, + "grad_norm": 0.4583573438714022, + "learning_rate": 2.759068178892105e-05, + "loss": 0.8574, + "step": 1049 + }, + { + "epoch": 1.09, + "grad_norm": 0.4506028295056863, + "learning_rate": 2.750966286452828e-05, + "loss": 0.904, + "step": 1050 + } + ], + "logging_steps": 1.0, + "max_steps": 1638, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 50, + "total_flos": 1088318406131712.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-1050/training_args.bin b/checkpoint-1050/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..2a09fdd266ae9810360e648350362ad7d2be5739 --- /dev/null +++ b/checkpoint-1050/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a909eb0ea0915df22d4edfd0cec39e5fb548124f1fa8377ddc8d0ea5dab0d0cc +size 6776 diff --git a/checkpoint-1050/zero_to_fp32.py b/checkpoint-1050/zero_to_fp32.py new file mode 100755 index 0000000000000000000000000000000000000000..24cc342e78d1a006c782b3a4cd68d9ce786d8fd8 --- /dev/null +++ b/checkpoint-1050/zero_to_fp32.py @@ -0,0 +1,604 @@ +#!/usr/bin/env python + +# Copyright (c) Microsoft Corporation. +# SPDX-License-Identifier: Apache-2.0 + +# DeepSpeed Team + +# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets +# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in +# the future. Once extracted, the weights don't require DeepSpeed and can be used in any +# application. +# +# example: python zero_to_fp32.py . pytorch_model.bin + +import argparse +import torch +import glob +import math +import os +import re +from collections import OrderedDict +from dataclasses import dataclass + +# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with +# DeepSpeed data structures it has to be available in the current python environment. +from deepspeed.utils import logger +from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS, + FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES, + FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS) + + +@dataclass +class zero_model_state: + buffers: dict() + param_shapes: dict() + shared_params: list + ds_version: int + frozen_param_shapes: dict() + frozen_param_fragments: dict() + + +debug = 0 + +# load to cpu +device = torch.device('cpu') + + +def atoi(text): + return int(text) if text.isdigit() else text + + +def natural_keys(text): + ''' + alist.sort(key=natural_keys) sorts in human order + http://nedbatchelder.com/blog/200712/human_sorting.html + (See Toothy's implementation in the comments) + ''' + return [atoi(c) for c in re.split(r'(\d+)', text)] + + +def get_model_state_file(checkpoint_dir, zero_stage): + if not os.path.isdir(checkpoint_dir): + raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist") + + # there should be only one file + if zero_stage <= 2: + file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt") + elif zero_stage == 3: + file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt") + + if not os.path.exists(file): + raise FileNotFoundError(f"can't find model states file at '{file}'") + + return file + + +def get_checkpoint_files(checkpoint_dir, glob_pattern): + # XXX: need to test that this simple glob rule works for multi-node setup too + ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys) + + if len(ckpt_files) == 0: + raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'") + + return ckpt_files + + +def get_optim_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt") + + +def get_model_state_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_model_states.pt") + + +def parse_model_states(files): + zero_model_states = [] + for file in files: + state_dict = torch.load(file, map_location=device) + + if BUFFER_NAMES not in state_dict: + raise ValueError(f"{file} is not a model state checkpoint") + buffer_names = state_dict[BUFFER_NAMES] + if debug: + print("Found buffers:", buffer_names) + + # recover just the buffers while restoring them to fp32 if they were saved in fp16 + buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names} + param_shapes = state_dict[PARAM_SHAPES] + + # collect parameters that are included in param_shapes + param_names = [] + for s in param_shapes: + for name in s.keys(): + param_names.append(name) + + # update with frozen parameters + frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None) + if frozen_param_shapes is not None: + if debug: + print(f"Found frozen_param_shapes: {frozen_param_shapes}") + param_names += list(frozen_param_shapes.keys()) + + # handle shared params + shared_params = [[k, v] for k, v in state_dict["shared_params"].items()] + + ds_version = state_dict.get(DS_VERSION, None) + + frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None) + + z_model_state = zero_model_state(buffers=buffers, + param_shapes=param_shapes, + shared_params=shared_params, + ds_version=ds_version, + frozen_param_shapes=frozen_param_shapes, + frozen_param_fragments=frozen_param_fragments) + zero_model_states.append(z_model_state) + + return zero_model_states + + +def parse_optim_states(files, ds_checkpoint_dir): + + total_files = len(files) + state_dicts = [] + for f in files: + state_dict = torch.load(f, map_location=device) + # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights + # and also handle the case where it was already removed by another helper script + state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None) + state_dicts.append(state_dict) + + if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]: + raise ValueError(f"{files[0]} is not a zero checkpoint") + zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE] + world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT] + + # For ZeRO-2 each param group can have different partition_count as data parallelism for expert + # parameters can be different from data parallelism for non-expert parameters. So we can just + # use the max of the partition_count to get the dp world_size. + + if type(world_size) is list: + world_size = max(world_size) + + if world_size != total_files: + raise ValueError( + f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. " + "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes." + ) + + # the groups are named differently in each stage + if zero_stage <= 2: + fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS + elif zero_stage == 3: + fp32_groups_key = FP32_FLAT_GROUPS + else: + raise ValueError(f"unknown zero stage {zero_stage}") + + if zero_stage <= 2: + fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))] + elif zero_stage == 3: + # if there is more than one param group, there will be multiple flattened tensors - one + # flattened tensor per group - for simplicity merge them into a single tensor + # + # XXX: could make the script more memory efficient for when there are multiple groups - it + # will require matching the sub-lists of param_shapes for each param group flattened tensor + + fp32_flat_groups = [ + torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts)) + ] + + return zero_stage, world_size, fp32_flat_groups + + +def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters): + """ + Returns fp32 state_dict reconstructed from ds checkpoint + + Args: + - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are) + + """ + print(f"Processing zero checkpoint '{ds_checkpoint_dir}'") + + optim_files = get_optim_files(ds_checkpoint_dir) + zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir) + print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}") + + model_files = get_model_state_files(ds_checkpoint_dir) + + zero_model_states = parse_model_states(model_files) + print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}') + + if zero_stage <= 2: + return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + elif zero_stage == 3: + return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + + +def _zero2_merge_frozen_params(state_dict, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + frozen_param_fragments = zero_model_states[0].frozen_param_fragments + + if debug: + num_elem = sum(s.numel() for s in frozen_param_shapes.values()) + print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in frozen_param_fragments.values()]) + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + state_dict[name] = frozen_param_fragments[name] + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _has_callable(obj, fn): + attr = getattr(obj, fn, None) + return callable(attr) + + +def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + + # Reconstruction protocol: + # + # XXX: document this + + if debug: + for i in range(world_size): + for j in range(len(fp32_flat_groups[0])): + print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}") + + # XXX: memory usage doubles here (zero2) + num_param_groups = len(fp32_flat_groups[0]) + merged_single_partition_of_fp32_groups = [] + for i in range(num_param_groups): + merged_partitions = [sd[i] for sd in fp32_flat_groups] + full_single_fp32_vector = torch.cat(merged_partitions, 0) + merged_single_partition_of_fp32_groups.append(full_single_fp32_vector) + avail_numel = sum( + [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups]) + + if debug: + wanted_params = sum([len(shapes) for shapes in param_shapes]) + wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes]) + # not asserting if there is a mismatch due to possible padding + print(f"Have {avail_numel} numels to process.") + print(f"Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + total_numel = 0 + total_params = 0 + for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups): + offset = 0 + avail_numel = full_single_fp32_vector.numel() + for name, shape in shapes.items(): + + unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape) + total_numel += unpartitioned_numel + total_params += 1 + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape) + offset += unpartitioned_numel + + # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and + # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex + # paddings performed in the code it's almost impossible to predict the exact numbers w/o the + # live optimizer object, so we are checking that the numbers are within the right range + align_to = 2 * world_size + + def zero2_align(x): + return align_to * math.ceil(x / align_to) + + if debug: + print(f"original offset={offset}, avail_numel={avail_numel}") + + offset = zero2_align(offset) + avail_numel = zero2_align(avail_numel) + + if debug: + print(f"aligned offset={offset}, avail_numel={avail_numel}") + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero2_merge_frozen_params(state_dict, zero_model_states) + + _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def zero3_partitioned_param_info(unpartitioned_numel, world_size): + remainder = unpartitioned_numel % world_size + padding_numel = (world_size - remainder) if remainder else 0 + partitioned_numel = math.ceil(unpartitioned_numel / world_size) + return partitioned_numel, padding_numel + + +def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + if debug: + for i in range(world_size): + num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values()) + print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in zero_model_states[0].frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states) + state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape) + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + avail_numel = fp32_flat_groups[0].numel() * world_size + # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each + # param, re-consolidating each param, while dealing with padding if any + + # merge list of dicts, preserving order + param_shapes = {k: v for d in param_shapes for k, v in d.items()} + + if debug: + for i in range(world_size): + print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}") + + wanted_params = len(param_shapes) + wanted_numel = sum(shape.numel() for shape in param_shapes.values()) + # not asserting if there is a mismatch due to possible padding + avail_numel = fp32_flat_groups[0].numel() * world_size + print(f"Trainable params: Have {avail_numel} numels to process.") + print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + offset = 0 + total_numel = 0 + total_params = 0 + for name, shape in param_shapes.items(): + + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + total_params += 1 + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + # XXX: memory usage doubles here + state_dict[name] = torch.cat( + tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)), + 0).narrow(0, 0, unpartitioned_numel).view(shape) + offset += partitioned_numel + + offset *= world_size + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero3_merge_frozen_params(state_dict, world_size, zero_model_states) + + _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with + ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example + via a model hub. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + + Returns: + - pytorch ``state_dict`` + + Note: this approach may not work if your application doesn't have sufficient free CPU memory and + you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with + the checkpoint. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + # do the training and checkpoint saving + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu + model = model.cpu() # move to cpu + model.load_state_dict(state_dict) + # submit to model hub or save the model to share with others + + In this example the ``model`` will no longer be usable in the deepspeed context of the same + application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead. + + """ + if tag is None: + latest_path = os.path.join(checkpoint_dir, 'latest') + if os.path.isfile(latest_path): + with open(latest_path, 'r') as fd: + tag = fd.read().strip() + else: + raise ValueError(f"Unable to find 'latest' file at {latest_path}") + + ds_checkpoint_dir = os.path.join(checkpoint_dir, tag) + + if not os.path.isdir(ds_checkpoint_dir): + raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist") + + return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters) + + +def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be + loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + """ + + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters) + print(f"Saving fp32 state dict to {output_file}") + torch.save(state_dict, output_file) + + +def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None): + """ + 1. Put the provided model to cpu + 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` + 3. Load it into the provided model + + Args: + - ``model``: the model object to update + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + + Returns: + - ``model`: modified model + + Make sure you have plenty of CPU memory available before you call this function. If you don't + have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it + conveniently placed for you in the checkpoint folder. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint + model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir) + # submit to model hub or save the model to share with others + + Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context + of the same application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + """ + logger.info(f"Extracting fp32 weights") + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag) + + logger.info(f"Overwriting model with fp32 weights") + model = model.cpu() + model.load_state_dict(state_dict, strict=False) + + return model + + +if __name__ == "__main__": + + parser = argparse.ArgumentParser() + parser.add_argument("checkpoint_dir", + type=str, + help="path to the desired checkpoint folder, e.g., path/checkpoint-12") + parser.add_argument( + "output_file", + type=str, + help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)") + parser.add_argument("-t", + "--tag", + type=str, + default=None, + help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1") + parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters") + parser.add_argument("-d", "--debug", action='store_true', help="enable debug") + args = parser.parse_args() + + debug = args.debug + + convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, + args.output_file, + tag=args.tag, + exclude_frozen_parameters=args.exclude_frozen_parameters) diff --git a/checkpoint-1100/README.md b/checkpoint-1100/README.md new file mode 100644 index 0000000000000000000000000000000000000000..16b1eacdd9353dec380a08ee77ce6ed5ab50f12e --- /dev/null +++ b/checkpoint-1100/README.md @@ -0,0 +1,202 @@ +--- +library_name: peft +base_model: gotzmann/uni +--- + +# Model Card for Model ID + + + + + +## Model Details + +### Model Description + + + + + +- **Developed by:** [More Information Needed] +- **Funded by [optional]:** [More Information Needed] +- **Shared by [optional]:** [More Information Needed] +- **Model type:** [More Information Needed] +- **Language(s) (NLP):** [More Information Needed] +- **License:** [More Information Needed] +- **Finetuned from model [optional]:** [More Information Needed] + +### Model Sources [optional] + + + +- **Repository:** [More Information Needed] +- **Paper [optional]:** [More Information Needed] +- **Demo [optional]:** [More Information Needed] + +## Uses + + + +### Direct Use + + + +[More Information Needed] + +### Downstream Use [optional] + + + +[More Information Needed] + +### Out-of-Scope Use + + + +[More Information Needed] + +## Bias, Risks, and Limitations + + + +[More Information Needed] + +### Recommendations + + + +Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations. + +## How to Get Started with the Model + +Use the code below to get started with the model. + +[More Information Needed] + +## Training Details + +### Training Data + + + +[More Information Needed] + +### Training Procedure + + + +#### Preprocessing [optional] + +[More Information Needed] + + +#### Training Hyperparameters + +- **Training regime:** [More Information Needed] + +#### Speeds, Sizes, Times [optional] + + + +[More Information Needed] + +## Evaluation + + + +### Testing Data, Factors & Metrics + +#### Testing Data + + + +[More Information Needed] + +#### Factors + + + +[More Information Needed] + +#### Metrics + + + +[More Information Needed] + +### Results + +[More Information Needed] + +#### Summary + + + +## Model Examination [optional] + + + +[More Information Needed] + +## Environmental Impact + + + +Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700). + +- **Hardware Type:** [More Information Needed] +- **Hours used:** [More Information Needed] +- **Cloud Provider:** [More Information Needed] +- **Compute Region:** [More Information Needed] +- **Carbon Emitted:** [More Information Needed] + +## Technical Specifications [optional] + +### Model Architecture and Objective + +[More Information Needed] + +### Compute Infrastructure + +[More Information Needed] + +#### Hardware + +[More Information Needed] + +#### Software + +[More Information Needed] + +## Citation [optional] + + + +**BibTeX:** + +[More Information Needed] + +**APA:** + +[More Information Needed] + +## Glossary [optional] + + + +[More Information Needed] + +## More Information [optional] + +[More Information Needed] + +## Model Card Authors [optional] + +[More Information Needed] + +## Model Card Contact + +[More Information Needed] +### Framework versions + +- PEFT 0.10.0 \ No newline at end of file diff --git a/checkpoint-1100/adapter_config.json b/checkpoint-1100/adapter_config.json new file mode 100644 index 0000000000000000000000000000000000000000..17e258368eee5905260715114fa1cc21fde3fe1c --- /dev/null +++ b/checkpoint-1100/adapter_config.json @@ -0,0 +1,31 @@ +{ + "alpha_pattern": {}, + "auto_mapping": null, + "base_model_name_or_path": "gotzmann/uni", + "bias": "none", + "fan_in_fan_out": false, + "inference_mode": true, + "init_lora_weights": true, + "layer_replication": null, + "layers_pattern": null, + "layers_to_transform": null, + "loftq_config": {}, + "lora_alpha": 128, + "lora_dropout": 0.0, + "megatron_config": null, + "megatron_core": "megatron.core", + "modules_to_save": null, + "peft_type": "LORA", + "r": 128, + "rank_pattern": {}, + "revision": null, + "target_modules": [ + "k_proj", + "q_proj", + "o_proj", + "v_proj" + ], + "task_type": "CAUSAL_LM", + "use_dora": false, + "use_rslora": true +} \ No newline at end of file diff --git a/checkpoint-1100/adapter_model.safetensors b/checkpoint-1100/adapter_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..9d022d74bf3613c8f84c170d4658e83535a3e026 --- /dev/null +++ b/checkpoint-1100/adapter_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:316ac193cc5592bd8d6e50a1701ef5c28729e999a88c35dc9173f4c9569c2273 +size 1048664848 diff --git a/checkpoint-1100/global_step1100/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt b/checkpoint-1100/global_step1100/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..910725281a3c4439a0731165371ac049d1b49e96 --- /dev/null +++ b/checkpoint-1100/global_step1100/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9e9e593f2ba9cb2b98bec0b40c7eb880c3bd54c0a3d63a0cdfd32f94305bd695 +size 787270042 diff --git a/checkpoint-1100/global_step1100/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt b/checkpoint-1100/global_step1100/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..ecbce101f5ff38ab78f0e433e18405528ec4fa1c --- /dev/null +++ b/checkpoint-1100/global_step1100/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:a4d424ba42629a4cd118ea7b3a16fee22cadbd2dd1766d07d8714f033c1c1636 +size 787270042 diff --git a/checkpoint-1100/global_step1100/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt b/checkpoint-1100/global_step1100/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..d0129ad9a5a9a6f8880f0c80422f9e422108635d --- /dev/null +++ b/checkpoint-1100/global_step1100/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:b48c1d0ad0c55442d28d51d18e27a901d5a514306d58f92a9e4c42b6559e36ee +size 787270042 diff --git a/checkpoint-1100/global_step1100/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt b/checkpoint-1100/global_step1100/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..1933e20cd73be2ba6e600f64cd1223e911acb777 --- /dev/null +++ b/checkpoint-1100/global_step1100/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:fc96e99992a02ff0d586173a7e789722264da08bb8c3af1d6a2fa8a7a53a5c47 +size 787270042 diff --git a/checkpoint-1100/global_step1100/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt b/checkpoint-1100/global_step1100/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..ccdd831979de4f3107539660333fc854a69d80d0 --- /dev/null +++ b/checkpoint-1100/global_step1100/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:984e073e59af9c46df4c004cceda03b60089ac6401f3e66ede86cdb574808f19 +size 787270042 diff --git a/checkpoint-1100/global_step1100/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt b/checkpoint-1100/global_step1100/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..de7b2c23e961cddd677e9485668d222360dbd49e --- /dev/null +++ b/checkpoint-1100/global_step1100/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0b08f0fd2c469f967a09b8848ddf371c0e29ad9c940ae18682c6d15846a629af +size 787270042 diff --git a/checkpoint-1100/global_step1100/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt b/checkpoint-1100/global_step1100/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..5aa8452e224e919bf9c47e59876009b4133c0e07 --- /dev/null +++ b/checkpoint-1100/global_step1100/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:99744a67797dd2ddd3291f10d48463a00c17a1ebdfbc6629a3de50a73feb53b6 +size 787270042 diff --git a/checkpoint-1100/global_step1100/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt b/checkpoint-1100/global_step1100/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..872a8a8ea67d65fed422f7abb75e6ec18e835d06 --- /dev/null +++ b/checkpoint-1100/global_step1100/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7cc97a2cbfe7f66897b4eece4afe406cff5351261db7fb7b7c4b444a4a7fcbb4 +size 787270042 diff --git a/checkpoint-1100/global_step1100/zero_pp_rank_0_mp_rank_00_model_states.pt b/checkpoint-1100/global_step1100/zero_pp_rank_0_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..56b1b9a47bbd6bdc455849e47649a03dccb32248 --- /dev/null +++ b/checkpoint-1100/global_step1100/zero_pp_rank_0_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9c769cbfb9bb02c99ba367fb77ae4600142c10796a5f2981693d9217e84ec52a +size 653742 diff --git a/checkpoint-1100/global_step1100/zero_pp_rank_1_mp_rank_00_model_states.pt b/checkpoint-1100/global_step1100/zero_pp_rank_1_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..26ea63fae69e18c1fbeb1640836919e8a9290e13 --- /dev/null +++ b/checkpoint-1100/global_step1100/zero_pp_rank_1_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5285eb24bfe43180e25a0aed0d6135893889228b40d88c09d87172badb155d02 +size 653742 diff --git a/checkpoint-1100/global_step1100/zero_pp_rank_2_mp_rank_00_model_states.pt b/checkpoint-1100/global_step1100/zero_pp_rank_2_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..7c5c53379e0cb424446e8424be9cca7081c51134 --- /dev/null +++ b/checkpoint-1100/global_step1100/zero_pp_rank_2_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:45d17f3b92f5f0897599a89c9ca82ccd4bcf980ac05427ec47b37774dd0d7836 +size 653742 diff --git a/checkpoint-1100/global_step1100/zero_pp_rank_3_mp_rank_00_model_states.pt b/checkpoint-1100/global_step1100/zero_pp_rank_3_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..e2ed0042a11be1fd9dfddd799b76d5c7a9d5e35a --- /dev/null +++ b/checkpoint-1100/global_step1100/zero_pp_rank_3_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e1005b64cba9f84408a432a5465c0821012797187041b73dd62245150d699401 +size 653742 diff --git a/checkpoint-1100/global_step1100/zero_pp_rank_4_mp_rank_00_model_states.pt b/checkpoint-1100/global_step1100/zero_pp_rank_4_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..6abe8550bbc218a84119ae98c6fa9d182ffd3179 --- /dev/null +++ b/checkpoint-1100/global_step1100/zero_pp_rank_4_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:33fea851aea5540e1e14ef2e33ba8014b1a1f69e8c22311c0cedfd7e359f979d +size 653742 diff --git a/checkpoint-1100/global_step1100/zero_pp_rank_5_mp_rank_00_model_states.pt b/checkpoint-1100/global_step1100/zero_pp_rank_5_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..0f1fa6c671027baf62c12a8618cf9a646cb0dc21 --- /dev/null +++ b/checkpoint-1100/global_step1100/zero_pp_rank_5_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e48f64d145ae4602275a2877ecc448efc9ec58f93f8364811a30a93ced0b76ad +size 653742 diff --git a/checkpoint-1100/global_step1100/zero_pp_rank_6_mp_rank_00_model_states.pt b/checkpoint-1100/global_step1100/zero_pp_rank_6_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..20720bdb5e208aa3129f7ee9ed0da54850f1a41b --- /dev/null +++ b/checkpoint-1100/global_step1100/zero_pp_rank_6_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:60582efe7474aa55d4745b11de94f5e294ef1c61df59dc79360753e051374e71 +size 653742 diff --git a/checkpoint-1100/global_step1100/zero_pp_rank_7_mp_rank_00_model_states.pt b/checkpoint-1100/global_step1100/zero_pp_rank_7_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..800a772e5b331ddeb871df43a2c5b72366b05364 --- /dev/null +++ b/checkpoint-1100/global_step1100/zero_pp_rank_7_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d1f58731d732acd2c0b1bdda0b00be0412c32ccdac00d654324bbb3ad53c323a +size 653742 diff --git a/checkpoint-1100/latest b/checkpoint-1100/latest new file mode 100644 index 0000000000000000000000000000000000000000..22cd5c3402316b70299aed2025d7943595f5d495 --- /dev/null +++ b/checkpoint-1100/latest @@ -0,0 +1 @@ +global_step1100 \ No newline at end of file diff --git a/checkpoint-1100/rng_state_0.pth b/checkpoint-1100/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..d4ade713ef57d0535c32a9251c786bc57de03d06 --- /dev/null +++ b/checkpoint-1100/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fb1165242405b17b3d6a8186ae61b13dcb1faa5a54320bebd74ef8d71b964bf7 +size 15984 diff --git a/checkpoint-1100/rng_state_1.pth b/checkpoint-1100/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..d91c511b147b4dd17988903c57adcefb6c1f20b0 --- /dev/null +++ b/checkpoint-1100/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:562c262916c9997ec644c42fed9655ab28706b74fca20290ca921c4761d6a4b0 +size 15984 diff --git a/checkpoint-1100/rng_state_2.pth b/checkpoint-1100/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..f71e829b3e3570a540263d07783c4e906a78a803 --- /dev/null +++ b/checkpoint-1100/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a8d40f8118f513299624ded0a9bcf09778b961635615090409394d4f96f928f6 +size 15984 diff --git a/checkpoint-1100/rng_state_3.pth b/checkpoint-1100/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..be7f0176676a7c526bb10cbb336b2afa89d8841c --- /dev/null +++ b/checkpoint-1100/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b4391f924238a4cb855c4cbdc6d1a14954f785431c75997d05c7a4ee6615dae7 +size 15984 diff --git a/checkpoint-1100/rng_state_4.pth b/checkpoint-1100/rng_state_4.pth new file mode 100644 index 0000000000000000000000000000000000000000..8dd1a877dd1f03799067fd08739e82b9f2cd2ad3 --- /dev/null +++ b/checkpoint-1100/rng_state_4.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:be7b19bb9543a16bf9f4cd96466ac581436f63070f5815f3a7ba57980608994f +size 15984 diff --git a/checkpoint-1100/rng_state_5.pth b/checkpoint-1100/rng_state_5.pth new file mode 100644 index 0000000000000000000000000000000000000000..dcf1b720014f72a27a09ab9ef8570430a8e3c96d --- /dev/null +++ b/checkpoint-1100/rng_state_5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:97da4a1ede0a3e0f96411cacd5bfdf84d9355198f7aadc9bcb8be41122043f63 +size 15984 diff --git a/checkpoint-1100/rng_state_6.pth b/checkpoint-1100/rng_state_6.pth new file mode 100644 index 0000000000000000000000000000000000000000..2b58cbeed7b25ef61c6439aced60df473cbaf6d4 --- /dev/null +++ b/checkpoint-1100/rng_state_6.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:544cb6421b975bd5d2b2360a4e666003794e6197ae654d2ad963cd6572a86ede +size 15984 diff --git a/checkpoint-1100/rng_state_7.pth b/checkpoint-1100/rng_state_7.pth new file mode 100644 index 0000000000000000000000000000000000000000..36a7dcefe0e0264868d40586546699306878a454 --- /dev/null +++ b/checkpoint-1100/rng_state_7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f8d6eb32a23f3bef6262bbcb2eda724b2fd6f5e579969aa27c71a5971331722b +size 15984 diff --git a/checkpoint-1100/scheduler.pt b/checkpoint-1100/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..82133a41263df3772d03bb81dbdf70b97ffb13d9 --- /dev/null +++ b/checkpoint-1100/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c08441fe4d6081d0dce5306ab4664d738c567d94c90b431beb90d45c64769f66 +size 1064 diff --git a/checkpoint-1100/special_tokens_map.json b/checkpoint-1100/special_tokens_map.json new file mode 100644 index 0000000000000000000000000000000000000000..72ecfeeb7e14d244c936169d2ed139eeae235ef1 --- /dev/null +++ b/checkpoint-1100/special_tokens_map.json @@ -0,0 +1,24 @@ +{ + "bos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "eos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "pad_token": "", + "unk_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + } +} diff --git a/checkpoint-1100/tokenizer.model b/checkpoint-1100/tokenizer.model new file mode 100644 index 0000000000000000000000000000000000000000..6c00c742ce03c627d6cd5b795984876fa49fa899 --- /dev/null +++ b/checkpoint-1100/tokenizer.model @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347 +size 499723 diff --git a/checkpoint-1100/tokenizer_config.json b/checkpoint-1100/tokenizer_config.json new file mode 100644 index 0000000000000000000000000000000000000000..bb5a9f09d8c0f3c32c66fc6118fe5c76c5c6fd90 --- /dev/null +++ b/checkpoint-1100/tokenizer_config.json @@ -0,0 +1,45 @@ +{ + "add_bos_token": true, + "add_eos_token": false, + "add_prefix_space": false, + "added_tokens_decoder": { + "0": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "1": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "2": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + } + }, + "bos_token": "", + "chat_template": "{% if messages[0]['role'] == 'system' %}{% set system_message = messages[0]['content'] %}{% endif %}{% if system_message is defined %}{{ '' + '### System:\\n\\n' + system_message }}{% endif %}{% for message in messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ '\\n\\n### Human:\\n\\n' + content }}{% elif message['role'] == 'assistant' %}{{ '\\n\\n### Assistant:\\n\\n' + content + '' }}{% endif %}{% endfor %}", + "clean_up_tokenization_spaces": false, + "eos_token": "", + "legacy": false, + "model_max_length": 1000000000000000019884624838656, + "pad_token": "", + "padding_side": "right", + "sp_model_kwargs": {}, + "spaces_between_special_tokens": false, + "split_special_tokens": false, + "tokenizer_class": "LlamaTokenizer", + "unk_token": "", + "use_default_system_prompt": false +} diff --git a/checkpoint-1100/trainer_state.json b/checkpoint-1100/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..956ff5b8c9b05ef2d9af28dd21efc1598843c2e6 --- /dev/null +++ b/checkpoint-1100/trainer_state.json @@ -0,0 +1,7721 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 1.0914494741655236, + "eval_steps": 500, + "global_step": 1100, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "grad_norm": 0.849355824164473, + "learning_rate": 4.878048780487805e-07, + "loss": 1.3655, + "step": 1 + }, + { + "epoch": 0.0, + "grad_norm": 10.01567518957158, + "learning_rate": 9.75609756097561e-07, + "loss": 1.5767, + "step": 2 + }, + { + "epoch": 0.01, + "grad_norm": 0.6466000875559635, + "learning_rate": 1.4634146341463414e-06, + "loss": 1.3913, + "step": 3 + }, + { + "epoch": 0.01, + "grad_norm": 0.6644565932010504, + "learning_rate": 1.951219512195122e-06, + "loss": 1.3218, + "step": 4 + }, + { + "epoch": 0.01, + "grad_norm": 0.571354207588475, + "learning_rate": 2.4390243902439027e-06, + "loss": 1.3597, + "step": 5 + }, + { + "epoch": 0.01, + "grad_norm": 0.31036262839244955, + "learning_rate": 2.926829268292683e-06, + "loss": 1.2832, + "step": 6 + }, + { + "epoch": 0.01, + "grad_norm": 0.2622135027188184, + "learning_rate": 3.414634146341464e-06, + "loss": 1.2161, + "step": 7 + }, + { + "epoch": 0.01, + "grad_norm": 0.296824630261661, + "learning_rate": 3.902439024390244e-06, + "loss": 1.2985, + "step": 8 + }, + { + "epoch": 0.02, + "grad_norm": 0.2557267467361569, + "learning_rate": 4.390243902439025e-06, + "loss": 1.3175, + "step": 9 + }, + { + "epoch": 0.02, + "grad_norm": 0.23418939513890769, + "learning_rate": 4.8780487804878055e-06, + "loss": 1.2617, + "step": 10 + }, + { + "epoch": 0.02, + "grad_norm": 0.2364760983285843, + "learning_rate": 5.365853658536586e-06, + "loss": 1.3103, + "step": 11 + }, + { + "epoch": 0.02, + "grad_norm": 0.23893034721889, + "learning_rate": 5.853658536585366e-06, + "loss": 1.2405, + "step": 12 + }, + { + "epoch": 0.02, + "grad_norm": 0.25563593295485887, + "learning_rate": 6.341463414634147e-06, + "loss": 1.2831, + "step": 13 + }, + { + "epoch": 0.03, + "grad_norm": 0.23239975352661665, + "learning_rate": 6.829268292682928e-06, + "loss": 1.3125, + "step": 14 + }, + { + "epoch": 0.03, + "grad_norm": 0.3092813858209507, + "learning_rate": 7.317073170731707e-06, + "loss": 1.2422, + "step": 15 + }, + { + "epoch": 0.03, + "grad_norm": 0.282563380367434, + "learning_rate": 7.804878048780489e-06, + "loss": 1.2453, + "step": 16 + }, + { + "epoch": 0.03, + "grad_norm": 0.22065680088315018, + "learning_rate": 8.292682926829268e-06, + "loss": 1.2491, + "step": 17 + }, + { + "epoch": 0.03, + "grad_norm": 0.22777800877980184, + "learning_rate": 8.78048780487805e-06, + "loss": 1.2655, + "step": 18 + }, + { + "epoch": 0.03, + "grad_norm": 0.22145212540177928, + "learning_rate": 9.268292682926831e-06, + "loss": 1.2413, + "step": 19 + }, + { + "epoch": 0.04, + "grad_norm": 0.22482351883112714, + "learning_rate": 9.756097560975611e-06, + "loss": 1.2653, + "step": 20 + }, + { + "epoch": 0.04, + "grad_norm": 0.20823080508385733, + "learning_rate": 1.024390243902439e-05, + "loss": 1.2374, + "step": 21 + }, + { + "epoch": 0.04, + "grad_norm": 0.26025492562935737, + "learning_rate": 1.0731707317073172e-05, + "loss": 1.2065, + "step": 22 + }, + { + "epoch": 0.04, + "grad_norm": 0.2150252124176173, + "learning_rate": 1.1219512195121953e-05, + "loss": 1.2782, + "step": 23 + }, + { + "epoch": 0.04, + "grad_norm": 0.2505915177425618, + "learning_rate": 1.1707317073170731e-05, + "loss": 1.2742, + "step": 24 + }, + { + "epoch": 0.05, + "grad_norm": 0.20129223044786942, + "learning_rate": 1.2195121951219513e-05, + "loss": 1.3366, + "step": 25 + }, + { + "epoch": 0.05, + "grad_norm": 0.1973508510397107, + "learning_rate": 1.2682926829268294e-05, + "loss": 1.2476, + "step": 26 + }, + { + "epoch": 0.05, + "grad_norm": 0.27103325392437194, + "learning_rate": 1.3170731707317076e-05, + "loss": 1.2325, + "step": 27 + }, + { + "epoch": 0.05, + "grad_norm": 0.17954976411006285, + "learning_rate": 1.3658536585365855e-05, + "loss": 1.2523, + "step": 28 + }, + { + "epoch": 0.05, + "grad_norm": 0.22216997851088888, + "learning_rate": 1.4146341463414635e-05, + "loss": 1.3297, + "step": 29 + }, + { + "epoch": 0.05, + "grad_norm": 0.2071458864548587, + "learning_rate": 1.4634146341463415e-05, + "loss": 1.2127, + "step": 30 + }, + { + "epoch": 0.06, + "grad_norm": 0.18039422081622164, + "learning_rate": 1.5121951219512196e-05, + "loss": 1.2509, + "step": 31 + }, + { + "epoch": 0.06, + "grad_norm": 0.18631254372974412, + "learning_rate": 1.5609756097560978e-05, + "loss": 1.2247, + "step": 32 + }, + { + "epoch": 0.06, + "grad_norm": 0.18843872523649827, + "learning_rate": 1.6097560975609757e-05, + "loss": 1.195, + "step": 33 + }, + { + "epoch": 0.06, + "grad_norm": 0.2163847267778325, + "learning_rate": 1.6585365853658537e-05, + "loss": 1.2179, + "step": 34 + }, + { + "epoch": 0.06, + "grad_norm": 0.19687688475496104, + "learning_rate": 1.7073170731707317e-05, + "loss": 1.2763, + "step": 35 + }, + { + "epoch": 0.07, + "grad_norm": 0.20409643064887947, + "learning_rate": 1.75609756097561e-05, + "loss": 1.253, + "step": 36 + }, + { + "epoch": 0.07, + "grad_norm": 0.1879182661759335, + "learning_rate": 1.804878048780488e-05, + "loss": 1.2586, + "step": 37 + }, + { + "epoch": 0.07, + "grad_norm": 0.19400648948514373, + "learning_rate": 1.8536585365853663e-05, + "loss": 1.2154, + "step": 38 + }, + { + "epoch": 0.07, + "grad_norm": 0.1878879343148452, + "learning_rate": 1.902439024390244e-05, + "loss": 1.2304, + "step": 39 + }, + { + "epoch": 0.07, + "grad_norm": 0.17687475469924052, + "learning_rate": 1.9512195121951222e-05, + "loss": 1.2351, + "step": 40 + }, + { + "epoch": 0.07, + "grad_norm": 0.18223935625384885, + "learning_rate": 2e-05, + "loss": 1.2222, + "step": 41 + }, + { + "epoch": 0.08, + "grad_norm": 0.1943061629408338, + "learning_rate": 2.048780487804878e-05, + "loss": 1.2044, + "step": 42 + }, + { + "epoch": 0.08, + "grad_norm": 0.17027514338700078, + "learning_rate": 2.0975609756097564e-05, + "loss": 1.1548, + "step": 43 + }, + { + "epoch": 0.08, + "grad_norm": 0.18553769630586192, + "learning_rate": 2.1463414634146344e-05, + "loss": 1.2721, + "step": 44 + }, + { + "epoch": 0.08, + "grad_norm": 0.19732826914228765, + "learning_rate": 2.1951219512195124e-05, + "loss": 1.3097, + "step": 45 + }, + { + "epoch": 0.08, + "grad_norm": 0.18714230986631472, + "learning_rate": 2.2439024390243907e-05, + "loss": 1.2662, + "step": 46 + }, + { + "epoch": 0.09, + "grad_norm": 0.19988987568002223, + "learning_rate": 2.2926829268292683e-05, + "loss": 1.2904, + "step": 47 + }, + { + "epoch": 0.09, + "grad_norm": 0.17744650133390918, + "learning_rate": 2.3414634146341463e-05, + "loss": 1.1825, + "step": 48 + }, + { + "epoch": 0.09, + "grad_norm": 0.16576734763834533, + "learning_rate": 2.3902439024390246e-05, + "loss": 1.1858, + "step": 49 + }, + { + "epoch": 0.09, + "grad_norm": 0.179591794065527, + "learning_rate": 2.4390243902439026e-05, + "loss": 1.2711, + "step": 50 + }, + { + "epoch": 0.09, + "grad_norm": 0.17923464471176911, + "learning_rate": 2.4878048780487805e-05, + "loss": 1.2289, + "step": 51 + }, + { + "epoch": 0.1, + "grad_norm": 0.18991742907836837, + "learning_rate": 2.536585365853659e-05, + "loss": 1.3097, + "step": 52 + }, + { + "epoch": 0.1, + "grad_norm": 0.19849796137254636, + "learning_rate": 2.5853658536585368e-05, + "loss": 1.2489, + "step": 53 + }, + { + "epoch": 0.1, + "grad_norm": 0.17452371110976383, + "learning_rate": 2.634146341463415e-05, + "loss": 1.2461, + "step": 54 + }, + { + "epoch": 0.1, + "grad_norm": 0.17671022353085036, + "learning_rate": 2.682926829268293e-05, + "loss": 1.153, + "step": 55 + }, + { + "epoch": 0.1, + "grad_norm": 0.36820559192096686, + "learning_rate": 2.731707317073171e-05, + "loss": 1.2431, + "step": 56 + }, + { + "epoch": 0.1, + "grad_norm": 0.20331468526494198, + "learning_rate": 2.7804878048780487e-05, + "loss": 1.2575, + "step": 57 + }, + { + "epoch": 0.11, + "grad_norm": 0.2402486598118377, + "learning_rate": 2.829268292682927e-05, + "loss": 1.2538, + "step": 58 + }, + { + "epoch": 0.11, + "grad_norm": 0.2549409484173144, + "learning_rate": 2.878048780487805e-05, + "loss": 1.2065, + "step": 59 + }, + { + "epoch": 0.11, + "grad_norm": 0.2053105349872685, + "learning_rate": 2.926829268292683e-05, + "loss": 1.2094, + "step": 60 + }, + { + "epoch": 0.11, + "grad_norm": 0.17971910872957886, + "learning_rate": 2.9756097560975613e-05, + "loss": 1.228, + "step": 61 + }, + { + "epoch": 0.11, + "grad_norm": 0.1885853654992973, + "learning_rate": 3.0243902439024392e-05, + "loss": 1.2286, + "step": 62 + }, + { + "epoch": 0.12, + "grad_norm": 0.1848524571968613, + "learning_rate": 3.073170731707317e-05, + "loss": 1.2718, + "step": 63 + }, + { + "epoch": 0.12, + "grad_norm": 0.18734105883548513, + "learning_rate": 3.1219512195121955e-05, + "loss": 1.2357, + "step": 64 + }, + { + "epoch": 0.12, + "grad_norm": 0.17774668052121825, + "learning_rate": 3.170731707317074e-05, + "loss": 1.1509, + "step": 65 + }, + { + "epoch": 0.12, + "grad_norm": 0.17890968008080646, + "learning_rate": 3.2195121951219514e-05, + "loss": 1.1924, + "step": 66 + }, + { + "epoch": 0.12, + "grad_norm": 0.18249273371332375, + "learning_rate": 3.268292682926829e-05, + "loss": 1.2545, + "step": 67 + }, + { + "epoch": 0.12, + "grad_norm": 0.21064122671902577, + "learning_rate": 3.3170731707317074e-05, + "loss": 1.2832, + "step": 68 + }, + { + "epoch": 0.13, + "grad_norm": 0.1820064171955093, + "learning_rate": 3.365853658536586e-05, + "loss": 1.2071, + "step": 69 + }, + { + "epoch": 0.13, + "grad_norm": 0.16996662800553433, + "learning_rate": 3.414634146341463e-05, + "loss": 1.2073, + "step": 70 + }, + { + "epoch": 0.13, + "grad_norm": 0.1618669302922445, + "learning_rate": 3.4634146341463416e-05, + "loss": 1.1289, + "step": 71 + }, + { + "epoch": 0.13, + "grad_norm": 0.18948744950985544, + "learning_rate": 3.51219512195122e-05, + "loss": 1.2915, + "step": 72 + }, + { + "epoch": 0.13, + "grad_norm": 0.18326143691603383, + "learning_rate": 3.5609756097560976e-05, + "loss": 1.2238, + "step": 73 + }, + { + "epoch": 0.14, + "grad_norm": 0.17410704510700503, + "learning_rate": 3.609756097560976e-05, + "loss": 1.1784, + "step": 74 + }, + { + "epoch": 0.14, + "grad_norm": 0.1983667344995625, + "learning_rate": 3.658536585365854e-05, + "loss": 1.2452, + "step": 75 + }, + { + "epoch": 0.14, + "grad_norm": 0.3416310763369357, + "learning_rate": 3.7073170731707325e-05, + "loss": 1.1972, + "step": 76 + }, + { + "epoch": 0.14, + "grad_norm": 0.2776466983511955, + "learning_rate": 3.75609756097561e-05, + "loss": 1.3121, + "step": 77 + }, + { + "epoch": 0.14, + "grad_norm": 0.20026129636576834, + "learning_rate": 3.804878048780488e-05, + "loss": 1.2436, + "step": 78 + }, + { + "epoch": 0.14, + "grad_norm": 0.21064549243917835, + "learning_rate": 3.853658536585366e-05, + "loss": 1.2064, + "step": 79 + }, + { + "epoch": 0.15, + "grad_norm": 0.22119482175714267, + "learning_rate": 3.9024390243902444e-05, + "loss": 1.2715, + "step": 80 + }, + { + "epoch": 0.15, + "grad_norm": 0.23047133748844142, + "learning_rate": 3.951219512195122e-05, + "loss": 1.2888, + "step": 81 + }, + { + "epoch": 0.15, + "grad_norm": 0.18741863156973176, + "learning_rate": 4e-05, + "loss": 1.248, + "step": 82 + }, + { + "epoch": 0.15, + "grad_norm": 0.1747859810629604, + "learning_rate": 4.0487804878048786e-05, + "loss": 1.1683, + "step": 83 + }, + { + "epoch": 0.15, + "grad_norm": 0.1896944798413341, + "learning_rate": 4.097560975609756e-05, + "loss": 1.2155, + "step": 84 + }, + { + "epoch": 0.16, + "grad_norm": 0.18724128114363303, + "learning_rate": 4.1463414634146346e-05, + "loss": 1.2273, + "step": 85 + }, + { + "epoch": 0.16, + "grad_norm": 0.17368125504855478, + "learning_rate": 4.195121951219513e-05, + "loss": 1.224, + "step": 86 + }, + { + "epoch": 0.16, + "grad_norm": 0.18371141013625703, + "learning_rate": 4.2439024390243905e-05, + "loss": 1.2294, + "step": 87 + }, + { + "epoch": 0.16, + "grad_norm": 0.1791029365673714, + "learning_rate": 4.292682926829269e-05, + "loss": 1.2895, + "step": 88 + }, + { + "epoch": 0.16, + "grad_norm": 0.20259974283859655, + "learning_rate": 4.341463414634147e-05, + "loss": 1.1841, + "step": 89 + }, + { + "epoch": 0.16, + "grad_norm": 0.17457456183272174, + "learning_rate": 4.390243902439025e-05, + "loss": 1.2357, + "step": 90 + }, + { + "epoch": 0.17, + "grad_norm": 0.1815824380789748, + "learning_rate": 4.439024390243903e-05, + "loss": 1.2304, + "step": 91 + }, + { + "epoch": 0.17, + "grad_norm": 0.17566480599583392, + "learning_rate": 4.4878048780487814e-05, + "loss": 1.242, + "step": 92 + }, + { + "epoch": 0.17, + "grad_norm": 0.18422975005984474, + "learning_rate": 4.536585365853658e-05, + "loss": 1.2177, + "step": 93 + }, + { + "epoch": 0.17, + "grad_norm": 0.16796781877940678, + "learning_rate": 4.5853658536585366e-05, + "loss": 1.1482, + "step": 94 + }, + { + "epoch": 0.17, + "grad_norm": 0.18636131653783305, + "learning_rate": 4.634146341463415e-05, + "loss": 1.1758, + "step": 95 + }, + { + "epoch": 0.18, + "grad_norm": 0.1823665700289814, + "learning_rate": 4.6829268292682926e-05, + "loss": 1.289, + "step": 96 + }, + { + "epoch": 0.18, + "grad_norm": 0.1719900691262439, + "learning_rate": 4.731707317073171e-05, + "loss": 1.1626, + "step": 97 + }, + { + "epoch": 0.18, + "grad_norm": 0.17937994168039778, + "learning_rate": 4.780487804878049e-05, + "loss": 1.175, + "step": 98 + }, + { + "epoch": 0.18, + "grad_norm": 0.16631851422106986, + "learning_rate": 4.829268292682927e-05, + "loss": 1.2177, + "step": 99 + }, + { + "epoch": 0.18, + "grad_norm": 0.19143696232800309, + "learning_rate": 4.878048780487805e-05, + "loss": 1.3071, + "step": 100 + }, + { + "epoch": 0.18, + "grad_norm": 0.17859506638780318, + "learning_rate": 4.9268292682926835e-05, + "loss": 1.2351, + "step": 101 + }, + { + "epoch": 0.19, + "grad_norm": 0.18381520321248196, + "learning_rate": 4.975609756097561e-05, + "loss": 1.2342, + "step": 102 + }, + { + "epoch": 0.19, + "grad_norm": 0.17968218683773912, + "learning_rate": 5.0243902439024394e-05, + "loss": 1.2074, + "step": 103 + }, + { + "epoch": 0.19, + "grad_norm": 0.18139489969339018, + "learning_rate": 5.073170731707318e-05, + "loss": 1.1558, + "step": 104 + }, + { + "epoch": 0.19, + "grad_norm": 0.17366624842514394, + "learning_rate": 5.121951219512195e-05, + "loss": 1.1897, + "step": 105 + }, + { + "epoch": 0.19, + "grad_norm": 0.16034845455223745, + "learning_rate": 5.1707317073170736e-05, + "loss": 1.179, + "step": 106 + }, + { + "epoch": 0.2, + "grad_norm": 0.17583069577827776, + "learning_rate": 5.219512195121952e-05, + "loss": 1.1856, + "step": 107 + }, + { + "epoch": 0.2, + "grad_norm": 0.1853758076989552, + "learning_rate": 5.26829268292683e-05, + "loss": 1.2072, + "step": 108 + }, + { + "epoch": 0.2, + "grad_norm": 0.19597443965936462, + "learning_rate": 5.317073170731708e-05, + "loss": 1.2271, + "step": 109 + }, + { + "epoch": 0.2, + "grad_norm": 0.1899206334098331, + "learning_rate": 5.365853658536586e-05, + "loss": 1.1961, + "step": 110 + }, + { + "epoch": 0.2, + "grad_norm": 0.17463763837757018, + "learning_rate": 5.4146341463414645e-05, + "loss": 1.2049, + "step": 111 + }, + { + "epoch": 0.2, + "grad_norm": 0.20431371701229986, + "learning_rate": 5.463414634146342e-05, + "loss": 1.2891, + "step": 112 + }, + { + "epoch": 0.21, + "grad_norm": 0.1814475107638498, + "learning_rate": 5.51219512195122e-05, + "loss": 1.2346, + "step": 113 + }, + { + "epoch": 0.21, + "grad_norm": 0.1883849423207823, + "learning_rate": 5.5609756097560974e-05, + "loss": 1.244, + "step": 114 + }, + { + "epoch": 0.21, + "grad_norm": 0.1857258128640568, + "learning_rate": 5.609756097560976e-05, + "loss": 1.2669, + "step": 115 + }, + { + "epoch": 0.21, + "grad_norm": 0.1740768514118401, + "learning_rate": 5.658536585365854e-05, + "loss": 1.2414, + "step": 116 + }, + { + "epoch": 0.21, + "grad_norm": 0.1919320335584178, + "learning_rate": 5.7073170731707317e-05, + "loss": 1.2886, + "step": 117 + }, + { + "epoch": 0.22, + "grad_norm": 0.18288775167828136, + "learning_rate": 5.75609756097561e-05, + "loss": 1.1875, + "step": 118 + }, + { + "epoch": 0.22, + "grad_norm": 0.18208588867750863, + "learning_rate": 5.804878048780488e-05, + "loss": 1.2388, + "step": 119 + }, + { + "epoch": 0.22, + "grad_norm": 0.1743260015658331, + "learning_rate": 5.853658536585366e-05, + "loss": 1.1762, + "step": 120 + }, + { + "epoch": 0.22, + "grad_norm": 0.17856046291517946, + "learning_rate": 5.902439024390244e-05, + "loss": 1.2888, + "step": 121 + }, + { + "epoch": 0.22, + "grad_norm": 0.17493794870966536, + "learning_rate": 5.9512195121951225e-05, + "loss": 1.2222, + "step": 122 + }, + { + "epoch": 0.22, + "grad_norm": 0.1909202655203384, + "learning_rate": 6.000000000000001e-05, + "loss": 1.2414, + "step": 123 + }, + { + "epoch": 0.23, + "grad_norm": 0.18345819482834988, + "learning_rate": 6.0487804878048785e-05, + "loss": 1.2756, + "step": 124 + }, + { + "epoch": 0.23, + "grad_norm": 0.2057069352956621, + "learning_rate": 6.097560975609757e-05, + "loss": 1.261, + "step": 125 + }, + { + "epoch": 0.23, + "grad_norm": 0.299775882469108, + "learning_rate": 6.146341463414634e-05, + "loss": 1.2566, + "step": 126 + }, + { + "epoch": 0.23, + "grad_norm": 0.1869687633018095, + "learning_rate": 6.195121951219513e-05, + "loss": 1.3039, + "step": 127 + }, + { + "epoch": 0.23, + "grad_norm": 0.17747149926197442, + "learning_rate": 6.243902439024391e-05, + "loss": 1.2524, + "step": 128 + }, + { + "epoch": 0.24, + "grad_norm": 0.17885157788044242, + "learning_rate": 6.29268292682927e-05, + "loss": 1.2455, + "step": 129 + }, + { + "epoch": 0.24, + "grad_norm": 0.17617298187845123, + "learning_rate": 6.341463414634148e-05, + "loss": 1.2009, + "step": 130 + }, + { + "epoch": 0.24, + "grad_norm": 0.20164176323497066, + "learning_rate": 6.390243902439025e-05, + "loss": 1.2634, + "step": 131 + }, + { + "epoch": 0.24, + "grad_norm": 0.20459903417307612, + "learning_rate": 6.439024390243903e-05, + "loss": 1.1963, + "step": 132 + }, + { + "epoch": 0.24, + "grad_norm": 0.1863755486334296, + "learning_rate": 6.487804878048781e-05, + "loss": 1.2387, + "step": 133 + }, + { + "epoch": 0.25, + "grad_norm": 0.19265866140295207, + "learning_rate": 6.536585365853658e-05, + "loss": 1.2688, + "step": 134 + }, + { + "epoch": 0.25, + "grad_norm": 0.1823425868969493, + "learning_rate": 6.585365853658536e-05, + "loss": 1.2041, + "step": 135 + }, + { + "epoch": 0.25, + "grad_norm": 0.2016853266472781, + "learning_rate": 6.634146341463415e-05, + "loss": 1.1223, + "step": 136 + }, + { + "epoch": 0.25, + "grad_norm": 0.17282675192463448, + "learning_rate": 6.682926829268293e-05, + "loss": 1.1879, + "step": 137 + }, + { + "epoch": 0.25, + "grad_norm": 0.17398811693399288, + "learning_rate": 6.731707317073171e-05, + "loss": 1.2682, + "step": 138 + }, + { + "epoch": 0.25, + "grad_norm": 0.18516916965434696, + "learning_rate": 6.78048780487805e-05, + "loss": 1.1666, + "step": 139 + }, + { + "epoch": 0.26, + "grad_norm": 0.1852213129647933, + "learning_rate": 6.829268292682927e-05, + "loss": 1.2501, + "step": 140 + }, + { + "epoch": 0.26, + "grad_norm": 0.17915948766591883, + "learning_rate": 6.878048780487805e-05, + "loss": 1.2264, + "step": 141 + }, + { + "epoch": 0.26, + "grad_norm": 0.21599939417233183, + "learning_rate": 6.926829268292683e-05, + "loss": 1.2376, + "step": 142 + }, + { + "epoch": 0.26, + "grad_norm": 0.17839304459521851, + "learning_rate": 6.975609756097562e-05, + "loss": 1.2353, + "step": 143 + }, + { + "epoch": 0.26, + "grad_norm": 0.20826913231380875, + "learning_rate": 7.02439024390244e-05, + "loss": 1.1901, + "step": 144 + }, + { + "epoch": 0.27, + "grad_norm": 0.20788894913361589, + "learning_rate": 7.073170731707318e-05, + "loss": 1.2577, + "step": 145 + }, + { + "epoch": 0.27, + "grad_norm": 0.18420055842301297, + "learning_rate": 7.121951219512195e-05, + "loss": 1.1393, + "step": 146 + }, + { + "epoch": 0.27, + "grad_norm": 0.19903048468685589, + "learning_rate": 7.170731707317073e-05, + "loss": 1.2321, + "step": 147 + }, + { + "epoch": 0.27, + "grad_norm": 0.19074116314985748, + "learning_rate": 7.219512195121952e-05, + "loss": 1.1912, + "step": 148 + }, + { + "epoch": 0.27, + "grad_norm": 0.2353816469403903, + "learning_rate": 7.26829268292683e-05, + "loss": 1.28, + "step": 149 + }, + { + "epoch": 0.27, + "grad_norm": 0.21634875684769345, + "learning_rate": 7.317073170731708e-05, + "loss": 1.3312, + "step": 150 + }, + { + "epoch": 0.28, + "grad_norm": 0.18290969006743918, + "learning_rate": 7.365853658536587e-05, + "loss": 1.2214, + "step": 151 + }, + { + "epoch": 0.28, + "grad_norm": 0.18484243897545208, + "learning_rate": 7.414634146341465e-05, + "loss": 1.1895, + "step": 152 + }, + { + "epoch": 0.28, + "grad_norm": 0.21882343112978872, + "learning_rate": 7.463414634146342e-05, + "loss": 1.2219, + "step": 153 + }, + { + "epoch": 0.28, + "grad_norm": 0.19868284379241205, + "learning_rate": 7.51219512195122e-05, + "loss": 1.2176, + "step": 154 + }, + { + "epoch": 0.28, + "grad_norm": 0.20912516312950613, + "learning_rate": 7.560975609756097e-05, + "loss": 1.242, + "step": 155 + }, + { + "epoch": 0.29, + "grad_norm": 0.23811880045549916, + "learning_rate": 7.609756097560976e-05, + "loss": 1.2838, + "step": 156 + }, + { + "epoch": 0.29, + "grad_norm": 0.19511077122033713, + "learning_rate": 7.658536585365854e-05, + "loss": 1.1594, + "step": 157 + }, + { + "epoch": 0.29, + "grad_norm": 0.20094129399534238, + "learning_rate": 7.707317073170732e-05, + "loss": 1.2966, + "step": 158 + }, + { + "epoch": 0.29, + "grad_norm": 0.19366245038292418, + "learning_rate": 7.75609756097561e-05, + "loss": 1.2246, + "step": 159 + }, + { + "epoch": 0.29, + "grad_norm": 0.19409570223867306, + "learning_rate": 7.804878048780489e-05, + "loss": 1.2312, + "step": 160 + }, + { + "epoch": 0.29, + "grad_norm": 0.2087258457033805, + "learning_rate": 7.853658536585366e-05, + "loss": 1.2169, + "step": 161 + }, + { + "epoch": 0.3, + "grad_norm": 0.18765223996270428, + "learning_rate": 7.902439024390244e-05, + "loss": 1.2383, + "step": 162 + }, + { + "epoch": 0.3, + "grad_norm": 0.20734180224147242, + "learning_rate": 7.951219512195122e-05, + "loss": 1.2587, + "step": 163 + }, + { + "epoch": 0.3, + "grad_norm": 0.24690929540287834, + "learning_rate": 8e-05, + "loss": 1.1951, + "step": 164 + }, + { + "epoch": 0.3, + "grad_norm": 0.2003538797619543, + "learning_rate": 7.999990914797545e-05, + "loss": 1.1982, + "step": 165 + }, + { + "epoch": 0.3, + "grad_norm": 0.22469075613510484, + "learning_rate": 7.99996365923145e-05, + "loss": 1.2355, + "step": 166 + }, + { + "epoch": 0.31, + "grad_norm": 0.21870100788336058, + "learning_rate": 7.999918233425526e-05, + "loss": 1.1103, + "step": 167 + }, + { + "epoch": 0.31, + "grad_norm": 0.20939989594131886, + "learning_rate": 7.999854637586122e-05, + "loss": 1.1966, + "step": 168 + }, + { + "epoch": 0.31, + "grad_norm": 0.43108211416237796, + "learning_rate": 7.999772872002132e-05, + "loss": 1.2882, + "step": 169 + }, + { + "epoch": 0.31, + "grad_norm": 0.27045413432174487, + "learning_rate": 7.999672937044984e-05, + "loss": 1.2399, + "step": 170 + }, + { + "epoch": 0.31, + "grad_norm": 0.19700483036740515, + "learning_rate": 7.999554833168642e-05, + "loss": 1.202, + "step": 171 + }, + { + "epoch": 0.31, + "grad_norm": 0.3335979493370708, + "learning_rate": 7.999418560909604e-05, + "loss": 1.1995, + "step": 172 + }, + { + "epoch": 0.32, + "grad_norm": 0.3165803974474567, + "learning_rate": 7.999264120886902e-05, + "loss": 1.1569, + "step": 173 + }, + { + "epoch": 0.32, + "grad_norm": 0.1951699080346223, + "learning_rate": 7.999091513802093e-05, + "loss": 1.1778, + "step": 174 + }, + { + "epoch": 0.32, + "grad_norm": 0.2087559121749787, + "learning_rate": 7.998900740439265e-05, + "loss": 1.1736, + "step": 175 + }, + { + "epoch": 0.32, + "grad_norm": 0.20345180977460478, + "learning_rate": 7.998691801665024e-05, + "loss": 1.2281, + "step": 176 + }, + { + "epoch": 0.32, + "grad_norm": 0.24617644827252333, + "learning_rate": 7.998464698428495e-05, + "loss": 1.2072, + "step": 177 + }, + { + "epoch": 0.33, + "grad_norm": 0.2469050959356265, + "learning_rate": 7.998219431761318e-05, + "loss": 1.2242, + "step": 178 + }, + { + "epoch": 0.33, + "grad_norm": 0.19529317748460623, + "learning_rate": 7.997956002777642e-05, + "loss": 1.2567, + "step": 179 + }, + { + "epoch": 0.33, + "grad_norm": 0.19048389491381376, + "learning_rate": 7.99767441267412e-05, + "loss": 1.2982, + "step": 180 + }, + { + "epoch": 0.33, + "grad_norm": 0.2085799116493225, + "learning_rate": 7.997374662729904e-05, + "loss": 1.1254, + "step": 181 + }, + { + "epoch": 0.33, + "grad_norm": 0.20636853256378995, + "learning_rate": 7.997056754306636e-05, + "loss": 1.2435, + "step": 182 + }, + { + "epoch": 0.33, + "grad_norm": 0.20590016382290252, + "learning_rate": 7.99672068884845e-05, + "loss": 1.2658, + "step": 183 + }, + { + "epoch": 0.34, + "grad_norm": 0.1931166169764433, + "learning_rate": 7.996366467881955e-05, + "loss": 1.1637, + "step": 184 + }, + { + "epoch": 0.34, + "grad_norm": 0.18873318157988098, + "learning_rate": 7.995994093016237e-05, + "loss": 1.1335, + "step": 185 + }, + { + "epoch": 0.34, + "grad_norm": 0.19210254625199108, + "learning_rate": 7.995603565942846e-05, + "loss": 1.1928, + "step": 186 + }, + { + "epoch": 0.34, + "grad_norm": 0.2130986479765664, + "learning_rate": 7.995194888435792e-05, + "loss": 1.2158, + "step": 187 + }, + { + "epoch": 0.34, + "grad_norm": 0.22003854501814088, + "learning_rate": 7.994768062351532e-05, + "loss": 1.2288, + "step": 188 + }, + { + "epoch": 0.35, + "grad_norm": 0.20330803191993058, + "learning_rate": 7.994323089628968e-05, + "loss": 1.2426, + "step": 189 + }, + { + "epoch": 0.35, + "grad_norm": 0.20567314642208634, + "learning_rate": 7.993859972289434e-05, + "loss": 1.2649, + "step": 190 + }, + { + "epoch": 0.35, + "grad_norm": 0.21556663727342962, + "learning_rate": 7.993378712436686e-05, + "loss": 1.2545, + "step": 191 + }, + { + "epoch": 0.35, + "grad_norm": 0.20309165469109888, + "learning_rate": 7.992879312256897e-05, + "loss": 1.3338, + "step": 192 + }, + { + "epoch": 0.35, + "grad_norm": 0.19574356669421325, + "learning_rate": 7.992361774018641e-05, + "loss": 1.278, + "step": 193 + }, + { + "epoch": 0.35, + "grad_norm": 0.2763613746722313, + "learning_rate": 7.991826100072891e-05, + "loss": 1.2571, + "step": 194 + }, + { + "epoch": 0.36, + "grad_norm": 0.19346552479915102, + "learning_rate": 7.991272292852996e-05, + "loss": 1.2027, + "step": 195 + }, + { + "epoch": 0.36, + "grad_norm": 0.2281167812123908, + "learning_rate": 7.990700354874683e-05, + "loss": 1.2586, + "step": 196 + }, + { + "epoch": 0.36, + "grad_norm": 0.19699013712137542, + "learning_rate": 7.990110288736042e-05, + "loss": 1.1371, + "step": 197 + }, + { + "epoch": 0.36, + "grad_norm": 0.21768209981475933, + "learning_rate": 7.989502097117503e-05, + "loss": 1.2522, + "step": 198 + }, + { + "epoch": 0.36, + "grad_norm": 0.21335427847754582, + "learning_rate": 7.988875782781838e-05, + "loss": 1.2437, + "step": 199 + }, + { + "epoch": 0.37, + "grad_norm": 0.21856710629066897, + "learning_rate": 7.988231348574147e-05, + "loss": 1.2135, + "step": 200 + }, + { + "epoch": 0.37, + "grad_norm": 0.20482062658774797, + "learning_rate": 7.987568797421836e-05, + "loss": 1.1755, + "step": 201 + }, + { + "epoch": 0.37, + "grad_norm": 0.2017756813960897, + "learning_rate": 7.986888132334608e-05, + "loss": 1.1699, + "step": 202 + }, + { + "epoch": 0.37, + "grad_norm": 0.20496443848153809, + "learning_rate": 7.986189356404458e-05, + "loss": 1.2125, + "step": 203 + }, + { + "epoch": 0.37, + "grad_norm": 0.2134603800558358, + "learning_rate": 7.985472472805643e-05, + "loss": 1.2391, + "step": 204 + }, + { + "epoch": 0.37, + "grad_norm": 0.2364175573420861, + "learning_rate": 7.98473748479468e-05, + "loss": 1.2384, + "step": 205 + }, + { + "epoch": 0.38, + "grad_norm": 0.1872419861598724, + "learning_rate": 7.983984395710326e-05, + "loss": 1.1457, + "step": 206 + }, + { + "epoch": 0.38, + "grad_norm": 0.28222194007095774, + "learning_rate": 7.983213208973566e-05, + "loss": 1.2952, + "step": 207 + }, + { + "epoch": 0.38, + "grad_norm": 0.1916094851162064, + "learning_rate": 7.982423928087593e-05, + "loss": 1.1763, + "step": 208 + }, + { + "epoch": 0.38, + "grad_norm": 0.18446245256166657, + "learning_rate": 7.981616556637795e-05, + "loss": 1.1863, + "step": 209 + }, + { + "epoch": 0.38, + "grad_norm": 0.195191961022491, + "learning_rate": 7.980791098291737e-05, + "loss": 1.2036, + "step": 210 + }, + { + "epoch": 0.39, + "grad_norm": 0.2652439657825496, + "learning_rate": 7.979947556799151e-05, + "loss": 1.2834, + "step": 211 + }, + { + "epoch": 0.39, + "grad_norm": 0.24308438957843412, + "learning_rate": 7.979085935991906e-05, + "loss": 1.234, + "step": 212 + }, + { + "epoch": 0.39, + "grad_norm": 0.21294701043622016, + "learning_rate": 7.978206239784004e-05, + "loss": 1.3006, + "step": 213 + }, + { + "epoch": 0.39, + "grad_norm": 0.25809277041859524, + "learning_rate": 7.977308472171553e-05, + "loss": 1.2272, + "step": 214 + }, + { + "epoch": 0.39, + "grad_norm": 0.193463860107294, + "learning_rate": 7.976392637232754e-05, + "loss": 1.2295, + "step": 215 + }, + { + "epoch": 0.4, + "grad_norm": 0.2150023760609626, + "learning_rate": 7.975458739127877e-05, + "loss": 1.2135, + "step": 216 + }, + { + "epoch": 0.4, + "grad_norm": 0.22590495955605894, + "learning_rate": 7.974506782099253e-05, + "loss": 1.2532, + "step": 217 + }, + { + "epoch": 0.4, + "grad_norm": 0.21023744668403702, + "learning_rate": 7.973536770471242e-05, + "loss": 1.2472, + "step": 218 + }, + { + "epoch": 0.4, + "grad_norm": 0.2345749799511543, + "learning_rate": 7.972548708650218e-05, + "loss": 1.1791, + "step": 219 + }, + { + "epoch": 0.4, + "grad_norm": 0.2158876734005217, + "learning_rate": 7.971542601124553e-05, + "loss": 1.2483, + "step": 220 + }, + { + "epoch": 0.4, + "grad_norm": 0.29455339949432446, + "learning_rate": 7.970518452464593e-05, + "loss": 1.2894, + "step": 221 + }, + { + "epoch": 0.41, + "grad_norm": 0.23983708730626851, + "learning_rate": 7.969476267322636e-05, + "loss": 1.271, + "step": 222 + }, + { + "epoch": 0.41, + "grad_norm": 0.1922400905426158, + "learning_rate": 7.968416050432912e-05, + "loss": 1.2139, + "step": 223 + }, + { + "epoch": 0.41, + "grad_norm": 0.2238136844422931, + "learning_rate": 7.967337806611568e-05, + "loss": 1.2655, + "step": 224 + }, + { + "epoch": 0.41, + "grad_norm": 0.21230292828267672, + "learning_rate": 7.966241540756631e-05, + "loss": 1.2406, + "step": 225 + }, + { + "epoch": 0.41, + "grad_norm": 0.26656119419070456, + "learning_rate": 7.965127257848004e-05, + "loss": 1.2595, + "step": 226 + }, + { + "epoch": 0.42, + "grad_norm": 0.22381385502992684, + "learning_rate": 7.963994962947426e-05, + "loss": 1.1737, + "step": 227 + }, + { + "epoch": 0.42, + "grad_norm": 0.20056702203994298, + "learning_rate": 7.962844661198462e-05, + "loss": 1.1969, + "step": 228 + }, + { + "epoch": 0.42, + "grad_norm": 0.20148701321526885, + "learning_rate": 7.961676357826478e-05, + "loss": 1.2151, + "step": 229 + }, + { + "epoch": 0.42, + "grad_norm": 0.20034834807028637, + "learning_rate": 7.960490058138604e-05, + "loss": 1.1455, + "step": 230 + }, + { + "epoch": 0.42, + "grad_norm": 0.21050838521846033, + "learning_rate": 7.959285767523732e-05, + "loss": 1.2223, + "step": 231 + }, + { + "epoch": 0.42, + "grad_norm": 0.20904772138969777, + "learning_rate": 7.95806349145247e-05, + "loss": 1.2534, + "step": 232 + }, + { + "epoch": 0.43, + "grad_norm": 0.20307877304792957, + "learning_rate": 7.956823235477134e-05, + "loss": 1.1352, + "step": 233 + }, + { + "epoch": 0.43, + "grad_norm": 0.20501105270897094, + "learning_rate": 7.95556500523171e-05, + "loss": 1.2031, + "step": 234 + }, + { + "epoch": 0.43, + "grad_norm": 0.19800586972038586, + "learning_rate": 7.954288806431838e-05, + "loss": 1.2567, + "step": 235 + }, + { + "epoch": 0.43, + "grad_norm": 0.2175102450594135, + "learning_rate": 7.952994644874777e-05, + "loss": 1.2538, + "step": 236 + }, + { + "epoch": 0.43, + "grad_norm": 0.22698189300067595, + "learning_rate": 7.951682526439391e-05, + "loss": 1.3088, + "step": 237 + }, + { + "epoch": 0.44, + "grad_norm": 0.19208392014975315, + "learning_rate": 7.950352457086109e-05, + "loss": 1.2336, + "step": 238 + }, + { + "epoch": 0.44, + "grad_norm": 0.27004086334319655, + "learning_rate": 7.949004442856905e-05, + "loss": 1.2012, + "step": 239 + }, + { + "epoch": 0.44, + "grad_norm": 0.23420974954538043, + "learning_rate": 7.947638489875272e-05, + "loss": 1.2244, + "step": 240 + }, + { + "epoch": 0.44, + "grad_norm": 0.20514399124802024, + "learning_rate": 7.946254604346186e-05, + "loss": 1.2548, + "step": 241 + }, + { + "epoch": 0.44, + "grad_norm": 0.19334973602372896, + "learning_rate": 7.944852792556092e-05, + "loss": 1.2104, + "step": 242 + }, + { + "epoch": 0.44, + "grad_norm": 0.1992640714537956, + "learning_rate": 7.943433060872858e-05, + "loss": 1.2628, + "step": 243 + }, + { + "epoch": 0.45, + "grad_norm": 0.203284617090413, + "learning_rate": 7.941995415745761e-05, + "loss": 1.2002, + "step": 244 + }, + { + "epoch": 0.45, + "grad_norm": 0.22795306969682058, + "learning_rate": 7.94053986370545e-05, + "loss": 1.2215, + "step": 245 + }, + { + "epoch": 0.45, + "grad_norm": 0.20789041346838505, + "learning_rate": 7.939066411363915e-05, + "loss": 1.0998, + "step": 246 + }, + { + "epoch": 0.45, + "grad_norm": 0.22354868884742066, + "learning_rate": 7.937575065414464e-05, + "loss": 1.2564, + "step": 247 + }, + { + "epoch": 0.45, + "grad_norm": 0.21176392726647736, + "learning_rate": 7.936065832631687e-05, + "loss": 1.2816, + "step": 248 + }, + { + "epoch": 0.46, + "grad_norm": 0.19967179557235587, + "learning_rate": 7.934538719871427e-05, + "loss": 1.1961, + "step": 249 + }, + { + "epoch": 0.46, + "grad_norm": 0.210819577350627, + "learning_rate": 7.932993734070747e-05, + "loss": 1.2167, + "step": 250 + }, + { + "epoch": 0.46, + "grad_norm": 0.21537794551756187, + "learning_rate": 7.931430882247903e-05, + "loss": 1.2341, + "step": 251 + }, + { + "epoch": 0.46, + "grad_norm": 0.22850872387256574, + "learning_rate": 7.929850171502304e-05, + "loss": 1.1686, + "step": 252 + }, + { + "epoch": 0.46, + "grad_norm": 0.22380366415076383, + "learning_rate": 7.928251609014493e-05, + "loss": 1.1462, + "step": 253 + }, + { + "epoch": 0.46, + "grad_norm": 0.22426923149036065, + "learning_rate": 7.926635202046102e-05, + "loss": 1.1792, + "step": 254 + }, + { + "epoch": 0.47, + "grad_norm": 0.42082703321103965, + "learning_rate": 7.925000957939822e-05, + "loss": 1.2718, + "step": 255 + }, + { + "epoch": 0.47, + "grad_norm": 0.2235432774854074, + "learning_rate": 7.92334888411937e-05, + "loss": 1.2598, + "step": 256 + }, + { + "epoch": 0.47, + "grad_norm": 0.281644028934108, + "learning_rate": 7.92167898808946e-05, + "loss": 1.2205, + "step": 257 + }, + { + "epoch": 0.47, + "grad_norm": 0.2037705143888748, + "learning_rate": 7.919991277435763e-05, + "loss": 1.1737, + "step": 258 + }, + { + "epoch": 0.47, + "grad_norm": 0.20917419230028977, + "learning_rate": 7.918285759824879e-05, + "loss": 1.2035, + "step": 259 + }, + { + "epoch": 0.48, + "grad_norm": 0.20510847570635518, + "learning_rate": 7.916562443004292e-05, + "loss": 1.2135, + "step": 260 + }, + { + "epoch": 0.48, + "grad_norm": 0.25172483071092466, + "learning_rate": 7.914821334802342e-05, + "loss": 1.2218, + "step": 261 + }, + { + "epoch": 0.48, + "grad_norm": 0.21102706700634313, + "learning_rate": 7.91306244312819e-05, + "loss": 1.1738, + "step": 262 + }, + { + "epoch": 0.48, + "grad_norm": 0.22626060872645815, + "learning_rate": 7.911285775971781e-05, + "loss": 1.238, + "step": 263 + }, + { + "epoch": 0.48, + "grad_norm": 0.22448567539778486, + "learning_rate": 7.909491341403805e-05, + "loss": 1.2404, + "step": 264 + }, + { + "epoch": 0.48, + "grad_norm": 0.2019099786139193, + "learning_rate": 7.907679147575661e-05, + "loss": 1.213, + "step": 265 + }, + { + "epoch": 0.49, + "grad_norm": 0.24307234839096267, + "learning_rate": 7.905849202719422e-05, + "loss": 1.2322, + "step": 266 + }, + { + "epoch": 0.49, + "grad_norm": 0.19801890521743487, + "learning_rate": 7.904001515147802e-05, + "loss": 1.2448, + "step": 267 + }, + { + "epoch": 0.49, + "grad_norm": 0.2102742273575385, + "learning_rate": 7.902136093254106e-05, + "loss": 1.1657, + "step": 268 + }, + { + "epoch": 0.49, + "grad_norm": 0.2173464476815016, + "learning_rate": 7.900252945512201e-05, + "loss": 1.2549, + "step": 269 + }, + { + "epoch": 0.49, + "grad_norm": 0.20957275458699595, + "learning_rate": 7.898352080476479e-05, + "loss": 1.2536, + "step": 270 + }, + { + "epoch": 0.5, + "grad_norm": 0.20691966388952363, + "learning_rate": 7.896433506781811e-05, + "loss": 1.2661, + "step": 271 + }, + { + "epoch": 0.5, + "grad_norm": 0.2276662275112648, + "learning_rate": 7.894497233143509e-05, + "loss": 1.2409, + "step": 272 + }, + { + "epoch": 0.5, + "grad_norm": 0.23854109569301263, + "learning_rate": 7.892543268357297e-05, + "loss": 1.2681, + "step": 273 + }, + { + "epoch": 0.5, + "grad_norm": 0.2233864156677627, + "learning_rate": 7.890571621299252e-05, + "loss": 1.1687, + "step": 274 + }, + { + "epoch": 0.5, + "grad_norm": 0.20114129147925475, + "learning_rate": 7.888582300925787e-05, + "loss": 1.2184, + "step": 275 + }, + { + "epoch": 0.5, + "grad_norm": 0.2154654670569462, + "learning_rate": 7.886575316273586e-05, + "loss": 1.1982, + "step": 276 + }, + { + "epoch": 0.51, + "grad_norm": 0.2292982209343639, + "learning_rate": 7.884550676459583e-05, + "loss": 1.2129, + "step": 277 + }, + { + "epoch": 0.51, + "grad_norm": 0.21302713135229548, + "learning_rate": 7.882508390680908e-05, + "loss": 1.1605, + "step": 278 + }, + { + "epoch": 0.51, + "grad_norm": 0.2123661020671048, + "learning_rate": 7.88044846821485e-05, + "loss": 1.2308, + "step": 279 + }, + { + "epoch": 0.51, + "grad_norm": 0.2080577410800404, + "learning_rate": 7.878370918418818e-05, + "loss": 1.2195, + "step": 280 + }, + { + "epoch": 0.51, + "grad_norm": 0.19663901881127385, + "learning_rate": 7.876275750730289e-05, + "loss": 1.1591, + "step": 281 + }, + { + "epoch": 0.52, + "grad_norm": 0.20534502031312163, + "learning_rate": 7.874162974666776e-05, + "loss": 1.2664, + "step": 282 + }, + { + "epoch": 0.52, + "grad_norm": 0.23240445399513837, + "learning_rate": 7.872032599825779e-05, + "loss": 1.2151, + "step": 283 + }, + { + "epoch": 0.52, + "grad_norm": 0.2672527316717507, + "learning_rate": 7.86988463588474e-05, + "loss": 1.2406, + "step": 284 + }, + { + "epoch": 0.52, + "grad_norm": 0.19893903058743695, + "learning_rate": 7.867719092601003e-05, + "loss": 1.1291, + "step": 285 + }, + { + "epoch": 0.52, + "grad_norm": 0.33275268109930917, + "learning_rate": 7.865535979811768e-05, + "loss": 1.1406, + "step": 286 + }, + { + "epoch": 0.52, + "grad_norm": 0.2373619455690358, + "learning_rate": 7.863335307434045e-05, + "loss": 1.2799, + "step": 287 + }, + { + "epoch": 0.53, + "grad_norm": 0.263235735390858, + "learning_rate": 7.861117085464612e-05, + "loss": 1.2415, + "step": 288 + }, + { + "epoch": 0.53, + "grad_norm": 0.25884281780784324, + "learning_rate": 7.858881323979965e-05, + "loss": 1.3919, + "step": 289 + }, + { + "epoch": 0.53, + "grad_norm": 0.25426288332255736, + "learning_rate": 7.85662803313628e-05, + "loss": 1.174, + "step": 290 + }, + { + "epoch": 0.53, + "grad_norm": 0.26655405527881243, + "learning_rate": 7.854357223169356e-05, + "loss": 1.2806, + "step": 291 + }, + { + "epoch": 0.53, + "grad_norm": 0.20909844432349833, + "learning_rate": 7.852068904394579e-05, + "loss": 1.2627, + "step": 292 + }, + { + "epoch": 0.54, + "grad_norm": 0.21307115068935759, + "learning_rate": 7.849763087206866e-05, + "loss": 1.1879, + "step": 293 + }, + { + "epoch": 0.54, + "grad_norm": 0.25009949471398946, + "learning_rate": 7.847439782080628e-05, + "loss": 1.2881, + "step": 294 + }, + { + "epoch": 0.54, + "grad_norm": 0.20960783418679174, + "learning_rate": 7.845098999569712e-05, + "loss": 1.2723, + "step": 295 + }, + { + "epoch": 0.54, + "grad_norm": 0.24968832437925104, + "learning_rate": 7.842740750307362e-05, + "loss": 1.2029, + "step": 296 + }, + { + "epoch": 0.54, + "grad_norm": 0.22981196585125677, + "learning_rate": 7.84036504500616e-05, + "loss": 1.1695, + "step": 297 + }, + { + "epoch": 0.55, + "grad_norm": 0.2320606844751365, + "learning_rate": 7.837971894457991e-05, + "loss": 1.2317, + "step": 298 + }, + { + "epoch": 0.55, + "grad_norm": 0.23051459673906124, + "learning_rate": 7.835561309533981e-05, + "loss": 1.2046, + "step": 299 + }, + { + "epoch": 0.55, + "grad_norm": 0.2510027231060586, + "learning_rate": 7.833133301184457e-05, + "loss": 1.199, + "step": 300 + }, + { + "epoch": 0.55, + "grad_norm": 0.23601180466018787, + "learning_rate": 7.830687880438895e-05, + "loss": 1.1755, + "step": 301 + }, + { + "epoch": 0.55, + "grad_norm": 0.24740820934385369, + "learning_rate": 7.828225058405864e-05, + "loss": 1.2054, + "step": 302 + }, + { + "epoch": 0.55, + "grad_norm": 0.23065372979111173, + "learning_rate": 7.825744846272984e-05, + "loss": 1.2066, + "step": 303 + }, + { + "epoch": 0.56, + "grad_norm": 0.22385077334838213, + "learning_rate": 7.823247255306866e-05, + "loss": 1.2147, + "step": 304 + }, + { + "epoch": 0.56, + "grad_norm": 0.42981213948386104, + "learning_rate": 7.820732296853074e-05, + "loss": 1.2314, + "step": 305 + }, + { + "epoch": 0.56, + "grad_norm": 0.21122844902751076, + "learning_rate": 7.818199982336058e-05, + "loss": 1.1462, + "step": 306 + }, + { + "epoch": 0.56, + "grad_norm": 0.23374869692118933, + "learning_rate": 7.815650323259117e-05, + "loss": 1.2051, + "step": 307 + }, + { + "epoch": 0.56, + "grad_norm": 0.21662363795962128, + "learning_rate": 7.813083331204332e-05, + "loss": 1.1575, + "step": 308 + }, + { + "epoch": 0.57, + "grad_norm": 0.2088315773384112, + "learning_rate": 7.810499017832526e-05, + "loss": 1.1316, + "step": 309 + }, + { + "epoch": 0.57, + "grad_norm": 0.2095238410730976, + "learning_rate": 7.807897394883203e-05, + "loss": 1.2087, + "step": 310 + }, + { + "epoch": 0.57, + "grad_norm": 0.22672932127256515, + "learning_rate": 7.805278474174499e-05, + "loss": 1.2512, + "step": 311 + }, + { + "epoch": 0.57, + "grad_norm": 0.21873052340922736, + "learning_rate": 7.802642267603126e-05, + "loss": 1.1909, + "step": 312 + }, + { + "epoch": 0.57, + "grad_norm": 0.219814521916342, + "learning_rate": 7.79998878714432e-05, + "loss": 1.1669, + "step": 313 + }, + { + "epoch": 0.57, + "grad_norm": 0.3049426027257317, + "learning_rate": 7.797318044851786e-05, + "loss": 1.1797, + "step": 314 + }, + { + "epoch": 0.58, + "grad_norm": 0.22309435690065985, + "learning_rate": 7.794630052857638e-05, + "loss": 1.1417, + "step": 315 + }, + { + "epoch": 0.58, + "grad_norm": 0.3891885169154885, + "learning_rate": 7.791924823372354e-05, + "loss": 1.2369, + "step": 316 + }, + { + "epoch": 0.58, + "grad_norm": 0.24780269452456372, + "learning_rate": 7.789202368684711e-05, + "loss": 1.2521, + "step": 317 + }, + { + "epoch": 0.58, + "grad_norm": 0.21660460720269362, + "learning_rate": 7.786462701161738e-05, + "loss": 1.2151, + "step": 318 + }, + { + "epoch": 0.58, + "grad_norm": 0.23635409466561857, + "learning_rate": 7.783705833248649e-05, + "loss": 1.2363, + "step": 319 + }, + { + "epoch": 0.59, + "grad_norm": 0.2616135839903218, + "learning_rate": 7.780931777468797e-05, + "loss": 1.2428, + "step": 320 + }, + { + "epoch": 0.59, + "grad_norm": 0.21461059159245083, + "learning_rate": 7.77814054642361e-05, + "loss": 1.1434, + "step": 321 + }, + { + "epoch": 0.59, + "grad_norm": 0.25348824286656163, + "learning_rate": 7.775332152792539e-05, + "loss": 1.2368, + "step": 322 + }, + { + "epoch": 0.59, + "grad_norm": 0.22275034726331247, + "learning_rate": 7.772506609332995e-05, + "loss": 1.1827, + "step": 323 + }, + { + "epoch": 0.59, + "grad_norm": 0.25030821228147526, + "learning_rate": 7.769663928880298e-05, + "loss": 1.2428, + "step": 324 + }, + { + "epoch": 0.59, + "grad_norm": 0.22251804398745534, + "learning_rate": 7.766804124347608e-05, + "loss": 1.1889, + "step": 325 + }, + { + "epoch": 0.6, + "grad_norm": 0.23381455520411995, + "learning_rate": 7.763927208725879e-05, + "loss": 1.2115, + "step": 326 + }, + { + "epoch": 0.6, + "grad_norm": 0.27341902651946226, + "learning_rate": 7.761033195083791e-05, + "loss": 1.2535, + "step": 327 + }, + { + "epoch": 0.6, + "grad_norm": 0.24862471659814522, + "learning_rate": 7.758122096567694e-05, + "loss": 1.2128, + "step": 328 + }, + { + "epoch": 0.6, + "grad_norm": 0.2251357082045494, + "learning_rate": 7.755193926401547e-05, + "loss": 1.2334, + "step": 329 + }, + { + "epoch": 0.6, + "grad_norm": 0.3173274941622932, + "learning_rate": 7.752248697886857e-05, + "loss": 1.226, + "step": 330 + }, + { + "epoch": 0.61, + "grad_norm": 0.23056440717672175, + "learning_rate": 7.74928642440263e-05, + "loss": 1.2339, + "step": 331 + }, + { + "epoch": 0.61, + "grad_norm": 0.2801507500859342, + "learning_rate": 7.746307119405286e-05, + "loss": 1.287, + "step": 332 + }, + { + "epoch": 0.61, + "grad_norm": 0.2267818430426272, + "learning_rate": 7.743310796428622e-05, + "loss": 1.1916, + "step": 333 + }, + { + "epoch": 0.61, + "grad_norm": 0.2777329160365585, + "learning_rate": 7.74029746908374e-05, + "loss": 1.252, + "step": 334 + }, + { + "epoch": 0.61, + "grad_norm": 0.25289169762353, + "learning_rate": 7.737267151058983e-05, + "loss": 1.2153, + "step": 335 + }, + { + "epoch": 0.61, + "grad_norm": 0.2424670686901653, + "learning_rate": 7.734219856119875e-05, + "loss": 1.2227, + "step": 336 + }, + { + "epoch": 0.62, + "grad_norm": 0.22747092217441645, + "learning_rate": 7.731155598109067e-05, + "loss": 1.19, + "step": 337 + }, + { + "epoch": 0.62, + "grad_norm": 0.2307810940100189, + "learning_rate": 7.728074390946257e-05, + "loss": 1.1818, + "step": 338 + }, + { + "epoch": 0.62, + "grad_norm": 0.2583402574655623, + "learning_rate": 7.724976248628142e-05, + "loss": 1.1608, + "step": 339 + }, + { + "epoch": 0.62, + "grad_norm": 0.22140209760890694, + "learning_rate": 7.721861185228347e-05, + "loss": 1.1245, + "step": 340 + }, + { + "epoch": 0.62, + "grad_norm": 0.25859310758244686, + "learning_rate": 7.718729214897362e-05, + "loss": 1.2247, + "step": 341 + }, + { + "epoch": 0.63, + "grad_norm": 0.26371179531372124, + "learning_rate": 7.715580351862482e-05, + "loss": 1.2128, + "step": 342 + }, + { + "epoch": 0.63, + "grad_norm": 0.26575541302851047, + "learning_rate": 7.712414610427733e-05, + "loss": 1.2443, + "step": 343 + }, + { + "epoch": 0.63, + "grad_norm": 0.269978305197599, + "learning_rate": 7.709232004973816e-05, + "loss": 1.2231, + "step": 344 + }, + { + "epoch": 0.63, + "grad_norm": 0.26583998705977047, + "learning_rate": 7.70603254995804e-05, + "loss": 1.2476, + "step": 345 + }, + { + "epoch": 0.63, + "grad_norm": 0.24256062164066097, + "learning_rate": 7.702816259914253e-05, + "loss": 1.2901, + "step": 346 + }, + { + "epoch": 0.63, + "grad_norm": 0.3463123472658915, + "learning_rate": 7.699583149452779e-05, + "loss": 1.3277, + "step": 347 + }, + { + "epoch": 0.64, + "grad_norm": 0.2269096590531878, + "learning_rate": 7.696333233260345e-05, + "loss": 1.2047, + "step": 348 + }, + { + "epoch": 0.64, + "grad_norm": 0.25136883001050025, + "learning_rate": 7.693066526100031e-05, + "loss": 1.1619, + "step": 349 + }, + { + "epoch": 0.64, + "grad_norm": 0.2565112571116145, + "learning_rate": 7.68978304281118e-05, + "loss": 1.2389, + "step": 350 + }, + { + "epoch": 0.64, + "grad_norm": 0.22175779550828703, + "learning_rate": 7.686482798309349e-05, + "loss": 1.2238, + "step": 351 + }, + { + "epoch": 0.64, + "grad_norm": 0.22588304332216555, + "learning_rate": 7.683165807586234e-05, + "loss": 1.174, + "step": 352 + }, + { + "epoch": 0.65, + "grad_norm": 0.24889474296529737, + "learning_rate": 7.6798320857096e-05, + "loss": 1.2366, + "step": 353 + }, + { + "epoch": 0.65, + "grad_norm": 0.27339703806525034, + "learning_rate": 7.676481647823214e-05, + "loss": 1.2356, + "step": 354 + }, + { + "epoch": 0.65, + "grad_norm": 0.23424666722888365, + "learning_rate": 7.673114509146782e-05, + "loss": 1.2089, + "step": 355 + }, + { + "epoch": 0.65, + "grad_norm": 0.27978285392461766, + "learning_rate": 7.66973068497587e-05, + "loss": 1.2609, + "step": 356 + }, + { + "epoch": 0.65, + "grad_norm": 0.2509423350138824, + "learning_rate": 7.666330190681844e-05, + "loss": 1.1777, + "step": 357 + }, + { + "epoch": 0.65, + "grad_norm": 0.23007730927468031, + "learning_rate": 7.662913041711793e-05, + "loss": 1.154, + "step": 358 + }, + { + "epoch": 0.66, + "grad_norm": 0.2438648674953112, + "learning_rate": 7.659479253588462e-05, + "loss": 1.2257, + "step": 359 + }, + { + "epoch": 0.66, + "grad_norm": 0.28816093242092233, + "learning_rate": 7.65602884191018e-05, + "loss": 1.2558, + "step": 360 + }, + { + "epoch": 0.66, + "grad_norm": 0.24972815300596035, + "learning_rate": 7.652561822350793e-05, + "loss": 1.2837, + "step": 361 + }, + { + "epoch": 0.66, + "grad_norm": 0.2543189139697063, + "learning_rate": 7.649078210659587e-05, + "loss": 1.2193, + "step": 362 + }, + { + "epoch": 0.66, + "grad_norm": 0.2237937956718952, + "learning_rate": 7.645578022661224e-05, + "loss": 1.2237, + "step": 363 + }, + { + "epoch": 0.67, + "grad_norm": 0.29742029408787396, + "learning_rate": 7.642061274255657e-05, + "loss": 1.2116, + "step": 364 + }, + { + "epoch": 0.67, + "grad_norm": 0.2462883147335493, + "learning_rate": 7.638527981418075e-05, + "loss": 1.1827, + "step": 365 + }, + { + "epoch": 0.67, + "grad_norm": 0.2647802498907096, + "learning_rate": 7.634978160198817e-05, + "loss": 1.2739, + "step": 366 + }, + { + "epoch": 0.67, + "grad_norm": 0.22360398779217264, + "learning_rate": 7.631411826723306e-05, + "loss": 1.2185, + "step": 367 + }, + { + "epoch": 0.67, + "grad_norm": 0.2635048004593543, + "learning_rate": 7.627828997191973e-05, + "loss": 1.2317, + "step": 368 + }, + { + "epoch": 0.67, + "grad_norm": 0.2764803449917684, + "learning_rate": 7.624229687880184e-05, + "loss": 1.1923, + "step": 369 + }, + { + "epoch": 0.68, + "grad_norm": 0.25724943233414527, + "learning_rate": 7.620613915138166e-05, + "loss": 1.2218, + "step": 370 + }, + { + "epoch": 0.68, + "grad_norm": 0.2858318045794755, + "learning_rate": 7.61698169539093e-05, + "loss": 1.1496, + "step": 371 + }, + { + "epoch": 0.68, + "grad_norm": 0.23547216647460364, + "learning_rate": 7.613333045138206e-05, + "loss": 1.1905, + "step": 372 + }, + { + "epoch": 0.68, + "grad_norm": 0.22984814903684375, + "learning_rate": 7.609667980954355e-05, + "loss": 1.2009, + "step": 373 + }, + { + "epoch": 0.68, + "grad_norm": 0.2551903754079084, + "learning_rate": 7.605986519488301e-05, + "loss": 1.2042, + "step": 374 + }, + { + "epoch": 0.69, + "grad_norm": 0.2508257410125616, + "learning_rate": 7.602288677463457e-05, + "loss": 1.2468, + "step": 375 + }, + { + "epoch": 0.69, + "grad_norm": 0.25324577774935964, + "learning_rate": 7.598574471677644e-05, + "loss": 1.2603, + "step": 376 + }, + { + "epoch": 0.69, + "grad_norm": 0.35888776531769967, + "learning_rate": 7.59484391900302e-05, + "loss": 1.1929, + "step": 377 + }, + { + "epoch": 0.69, + "grad_norm": 0.22048517191014724, + "learning_rate": 7.591097036385994e-05, + "loss": 1.1783, + "step": 378 + }, + { + "epoch": 0.69, + "grad_norm": 0.2781160412746083, + "learning_rate": 7.587333840847162e-05, + "loss": 1.3397, + "step": 379 + }, + { + "epoch": 0.7, + "grad_norm": 0.24033046830332258, + "learning_rate": 7.583554349481222e-05, + "loss": 1.2436, + "step": 380 + }, + { + "epoch": 0.7, + "grad_norm": 0.26413762380260003, + "learning_rate": 7.579758579456893e-05, + "loss": 1.1917, + "step": 381 + }, + { + "epoch": 0.7, + "grad_norm": 0.2390937887338632, + "learning_rate": 7.575946548016847e-05, + "loss": 1.2186, + "step": 382 + }, + { + "epoch": 0.7, + "grad_norm": 0.25131263043429275, + "learning_rate": 7.572118272477622e-05, + "loss": 1.2538, + "step": 383 + }, + { + "epoch": 0.7, + "grad_norm": 0.223974104870702, + "learning_rate": 7.568273770229546e-05, + "loss": 1.2165, + "step": 384 + }, + { + "epoch": 0.7, + "grad_norm": 0.25840356830252875, + "learning_rate": 7.564413058736663e-05, + "loss": 1.1848, + "step": 385 + }, + { + "epoch": 0.71, + "grad_norm": 0.2723156683076603, + "learning_rate": 7.560536155536641e-05, + "loss": 1.1982, + "step": 386 + }, + { + "epoch": 0.71, + "grad_norm": 0.265687427976889, + "learning_rate": 7.556643078240708e-05, + "loss": 1.231, + "step": 387 + }, + { + "epoch": 0.71, + "grad_norm": 0.25152762080976077, + "learning_rate": 7.552733844533562e-05, + "loss": 1.1974, + "step": 388 + }, + { + "epoch": 0.71, + "grad_norm": 0.2366049485053541, + "learning_rate": 7.548808472173292e-05, + "loss": 1.3119, + "step": 389 + }, + { + "epoch": 0.71, + "grad_norm": 0.22092196577077122, + "learning_rate": 7.5448669789913e-05, + "loss": 1.195, + "step": 390 + }, + { + "epoch": 0.72, + "grad_norm": 0.22667521540462374, + "learning_rate": 7.540909382892217e-05, + "loss": 1.1431, + "step": 391 + }, + { + "epoch": 0.72, + "grad_norm": 0.25432207282646513, + "learning_rate": 7.536935701853823e-05, + "loss": 1.2173, + "step": 392 + }, + { + "epoch": 0.72, + "grad_norm": 0.29950506457923864, + "learning_rate": 7.53294595392697e-05, + "loss": 1.1962, + "step": 393 + }, + { + "epoch": 0.72, + "grad_norm": 0.24735689607229913, + "learning_rate": 7.528940157235487e-05, + "loss": 1.2053, + "step": 394 + }, + { + "epoch": 0.72, + "grad_norm": 0.24394198607459663, + "learning_rate": 7.524918329976114e-05, + "loss": 1.1979, + "step": 395 + }, + { + "epoch": 0.72, + "grad_norm": 0.2630369372689188, + "learning_rate": 7.520880490418409e-05, + "loss": 1.2111, + "step": 396 + }, + { + "epoch": 0.73, + "grad_norm": 0.26275028416291457, + "learning_rate": 7.516826656904664e-05, + "loss": 1.2133, + "step": 397 + }, + { + "epoch": 0.73, + "grad_norm": 0.23938074620956928, + "learning_rate": 7.512756847849831e-05, + "loss": 1.1355, + "step": 398 + }, + { + "epoch": 0.73, + "grad_norm": 0.3724960610098138, + "learning_rate": 7.508671081741428e-05, + "loss": 1.2572, + "step": 399 + }, + { + "epoch": 0.73, + "grad_norm": 0.24161685847894723, + "learning_rate": 7.504569377139462e-05, + "loss": 1.1706, + "step": 400 + }, + { + "epoch": 0.73, + "grad_norm": 0.26121591322670523, + "learning_rate": 7.50045175267634e-05, + "loss": 1.2135, + "step": 401 + }, + { + "epoch": 0.74, + "grad_norm": 0.2465579498164775, + "learning_rate": 7.496318227056788e-05, + "loss": 1.1641, + "step": 402 + }, + { + "epoch": 0.74, + "grad_norm": 0.2556288696122787, + "learning_rate": 7.492168819057767e-05, + "loss": 1.2939, + "step": 403 + }, + { + "epoch": 0.74, + "grad_norm": 0.261481216336303, + "learning_rate": 7.488003547528382e-05, + "loss": 1.2026, + "step": 404 + }, + { + "epoch": 0.74, + "grad_norm": 0.2389415135676362, + "learning_rate": 7.483822431389799e-05, + "loss": 1.2131, + "step": 405 + }, + { + "epoch": 0.74, + "grad_norm": 0.2559201956627192, + "learning_rate": 7.479625489635162e-05, + "loss": 1.1246, + "step": 406 + }, + { + "epoch": 0.74, + "grad_norm": 0.27127932491822604, + "learning_rate": 7.475412741329504e-05, + "loss": 1.2429, + "step": 407 + }, + { + "epoch": 0.75, + "grad_norm": 0.27006004008695594, + "learning_rate": 7.47118420560966e-05, + "loss": 1.2388, + "step": 408 + }, + { + "epoch": 0.75, + "grad_norm": 0.23716823297200537, + "learning_rate": 7.466939901684182e-05, + "loss": 1.1264, + "step": 409 + }, + { + "epoch": 0.75, + "grad_norm": 0.2885373898669248, + "learning_rate": 7.462679848833252e-05, + "loss": 1.2786, + "step": 410 + }, + { + "epoch": 0.75, + "grad_norm": 0.49215227598639927, + "learning_rate": 7.458404066408588e-05, + "loss": 1.2386, + "step": 411 + }, + { + "epoch": 0.75, + "grad_norm": 0.24235735604947403, + "learning_rate": 7.454112573833368e-05, + "loss": 1.1423, + "step": 412 + }, + { + "epoch": 0.76, + "grad_norm": 0.2584614748054343, + "learning_rate": 7.449805390602127e-05, + "loss": 1.2669, + "step": 413 + }, + { + "epoch": 0.76, + "grad_norm": 0.23806123085998873, + "learning_rate": 7.445482536280684e-05, + "loss": 1.1763, + "step": 414 + }, + { + "epoch": 0.76, + "grad_norm": 0.24459517607786851, + "learning_rate": 7.441144030506043e-05, + "loss": 1.198, + "step": 415 + }, + { + "epoch": 0.76, + "grad_norm": 0.25801616402700395, + "learning_rate": 7.436789892986304e-05, + "loss": 1.2136, + "step": 416 + }, + { + "epoch": 0.76, + "grad_norm": 0.2814819942392514, + "learning_rate": 7.432420143500578e-05, + "loss": 1.2398, + "step": 417 + }, + { + "epoch": 0.76, + "grad_norm": 0.22134709322606153, + "learning_rate": 7.428034801898893e-05, + "loss": 1.1592, + "step": 418 + }, + { + "epoch": 0.77, + "grad_norm": 0.2899677536995633, + "learning_rate": 7.42363388810211e-05, + "loss": 1.2296, + "step": 419 + }, + { + "epoch": 0.77, + "grad_norm": 0.24005943230262294, + "learning_rate": 7.419217422101822e-05, + "loss": 1.2223, + "step": 420 + }, + { + "epoch": 0.77, + "grad_norm": 0.26417562369496167, + "learning_rate": 7.414785423960275e-05, + "loss": 1.2261, + "step": 421 + }, + { + "epoch": 0.77, + "grad_norm": 0.2580815883535521, + "learning_rate": 7.410337913810271e-05, + "loss": 1.2021, + "step": 422 + }, + { + "epoch": 0.77, + "grad_norm": 0.25242217589496435, + "learning_rate": 7.405874911855071e-05, + "loss": 1.239, + "step": 423 + }, + { + "epoch": 0.78, + "grad_norm": 0.21991733999839932, + "learning_rate": 7.401396438368315e-05, + "loss": 1.1716, + "step": 424 + }, + { + "epoch": 0.78, + "grad_norm": 0.40116538322720213, + "learning_rate": 7.396902513693924e-05, + "loss": 1.2773, + "step": 425 + }, + { + "epoch": 0.78, + "grad_norm": 0.277333939455099, + "learning_rate": 7.392393158246002e-05, + "loss": 1.2574, + "step": 426 + }, + { + "epoch": 0.78, + "grad_norm": 0.27146087746385755, + "learning_rate": 7.387868392508756e-05, + "loss": 1.2243, + "step": 427 + }, + { + "epoch": 0.78, + "grad_norm": 0.255881055620786, + "learning_rate": 7.38332823703639e-05, + "loss": 1.223, + "step": 428 + }, + { + "epoch": 0.78, + "grad_norm": 0.24807364856677255, + "learning_rate": 7.378772712453021e-05, + "loss": 1.1985, + "step": 429 + }, + { + "epoch": 0.79, + "grad_norm": 0.25746257617764423, + "learning_rate": 7.37420183945258e-05, + "loss": 1.2502, + "step": 430 + }, + { + "epoch": 0.79, + "grad_norm": 0.28851991049982234, + "learning_rate": 7.369615638798722e-05, + "loss": 1.2535, + "step": 431 + }, + { + "epoch": 0.79, + "grad_norm": 0.24113389811604363, + "learning_rate": 7.365014131324725e-05, + "loss": 1.2227, + "step": 432 + }, + { + "epoch": 0.79, + "grad_norm": 0.2414465151257969, + "learning_rate": 7.360397337933405e-05, + "loss": 1.1884, + "step": 433 + }, + { + "epoch": 0.79, + "grad_norm": 0.2735463134699831, + "learning_rate": 7.355765279597011e-05, + "loss": 1.2756, + "step": 434 + }, + { + "epoch": 0.8, + "grad_norm": 0.2588437452987293, + "learning_rate": 7.351117977357139e-05, + "loss": 1.2108, + "step": 435 + }, + { + "epoch": 0.8, + "grad_norm": 0.26573294117796553, + "learning_rate": 7.346455452324629e-05, + "loss": 1.1821, + "step": 436 + }, + { + "epoch": 0.8, + "grad_norm": 0.2555476577827304, + "learning_rate": 7.341777725679473e-05, + "loss": 1.1937, + "step": 437 + }, + { + "epoch": 0.8, + "grad_norm": 0.2867704132108098, + "learning_rate": 7.337084818670716e-05, + "loss": 1.2272, + "step": 438 + }, + { + "epoch": 0.8, + "grad_norm": 0.27726678115981157, + "learning_rate": 7.332376752616367e-05, + "loss": 1.2331, + "step": 439 + }, + { + "epoch": 0.8, + "grad_norm": 0.26955338021079955, + "learning_rate": 7.32765354890329e-05, + "loss": 1.1731, + "step": 440 + }, + { + "epoch": 0.81, + "grad_norm": 0.25250321202536524, + "learning_rate": 7.322915228987116e-05, + "loss": 1.2653, + "step": 441 + }, + { + "epoch": 0.81, + "grad_norm": 0.24748844179765395, + "learning_rate": 7.318161814392143e-05, + "loss": 1.24, + "step": 442 + }, + { + "epoch": 0.81, + "grad_norm": 0.28177805247356325, + "learning_rate": 7.313393326711239e-05, + "loss": 1.185, + "step": 443 + }, + { + "epoch": 0.81, + "grad_norm": 0.24093242000396312, + "learning_rate": 7.30860978760574e-05, + "loss": 1.1994, + "step": 444 + }, + { + "epoch": 0.81, + "grad_norm": 0.26277803901457075, + "learning_rate": 7.30381121880536e-05, + "loss": 1.212, + "step": 445 + }, + { + "epoch": 0.82, + "grad_norm": 0.2506524258682433, + "learning_rate": 7.298997642108079e-05, + "loss": 1.2421, + "step": 446 + }, + { + "epoch": 0.82, + "grad_norm": 0.2840599700015824, + "learning_rate": 7.294169079380061e-05, + "loss": 1.1818, + "step": 447 + }, + { + "epoch": 0.82, + "grad_norm": 0.24892184038117549, + "learning_rate": 7.289325552555538e-05, + "loss": 1.1916, + "step": 448 + }, + { + "epoch": 0.82, + "grad_norm": 0.2700898428541357, + "learning_rate": 7.284467083636722e-05, + "loss": 1.2517, + "step": 449 + }, + { + "epoch": 0.82, + "grad_norm": 0.2617848546539419, + "learning_rate": 7.279593694693698e-05, + "loss": 1.2063, + "step": 450 + }, + { + "epoch": 0.82, + "grad_norm": 0.2698278585334131, + "learning_rate": 7.274705407864332e-05, + "loss": 1.194, + "step": 451 + }, + { + "epoch": 0.83, + "grad_norm": 0.23678313024953834, + "learning_rate": 7.26980224535416e-05, + "loss": 1.2349, + "step": 452 + }, + { + "epoch": 0.83, + "grad_norm": 0.24851875792002978, + "learning_rate": 7.264884229436293e-05, + "loss": 1.1758, + "step": 453 + }, + { + "epoch": 0.83, + "grad_norm": 0.24122080121681125, + "learning_rate": 7.259951382451318e-05, + "loss": 1.1962, + "step": 454 + }, + { + "epoch": 0.83, + "grad_norm": 0.22741322959884405, + "learning_rate": 7.25500372680719e-05, + "loss": 1.1702, + "step": 455 + }, + { + "epoch": 0.83, + "grad_norm": 0.2297475610861458, + "learning_rate": 7.250041284979137e-05, + "loss": 1.1466, + "step": 456 + }, + { + "epoch": 0.84, + "grad_norm": 0.3057605989721467, + "learning_rate": 7.245064079509553e-05, + "loss": 1.246, + "step": 457 + }, + { + "epoch": 0.84, + "grad_norm": 0.2719638501597136, + "learning_rate": 7.240072133007899e-05, + "loss": 1.2184, + "step": 458 + }, + { + "epoch": 0.84, + "grad_norm": 0.2436807816414479, + "learning_rate": 7.235065468150593e-05, + "loss": 1.2324, + "step": 459 + }, + { + "epoch": 0.84, + "grad_norm": 0.23436349430255515, + "learning_rate": 7.23004410768092e-05, + "loss": 1.1813, + "step": 460 + }, + { + "epoch": 0.84, + "grad_norm": 0.2398940990211377, + "learning_rate": 7.22500807440892e-05, + "loss": 1.1924, + "step": 461 + }, + { + "epoch": 0.84, + "grad_norm": 0.2605716625062531, + "learning_rate": 7.219957391211281e-05, + "loss": 1.182, + "step": 462 + }, + { + "epoch": 0.85, + "grad_norm": 0.260462524570941, + "learning_rate": 7.214892081031244e-05, + "loss": 1.2136, + "step": 463 + }, + { + "epoch": 0.85, + "grad_norm": 0.21979766512306334, + "learning_rate": 7.209812166878491e-05, + "loss": 1.2066, + "step": 464 + }, + { + "epoch": 0.85, + "grad_norm": 0.23324453647530663, + "learning_rate": 7.204717671829051e-05, + "loss": 1.1657, + "step": 465 + }, + { + "epoch": 0.85, + "grad_norm": 0.2529434935507481, + "learning_rate": 7.199608619025177e-05, + "loss": 1.2093, + "step": 466 + }, + { + "epoch": 0.85, + "grad_norm": 0.25371701891720116, + "learning_rate": 7.194485031675265e-05, + "loss": 1.2225, + "step": 467 + }, + { + "epoch": 0.86, + "grad_norm": 0.23272423066292103, + "learning_rate": 7.189346933053725e-05, + "loss": 1.1721, + "step": 468 + }, + { + "epoch": 0.86, + "grad_norm": 0.25122928735587546, + "learning_rate": 7.184194346500892e-05, + "loss": 1.2537, + "step": 469 + }, + { + "epoch": 0.86, + "grad_norm": 0.2159270875490409, + "learning_rate": 7.179027295422913e-05, + "loss": 1.197, + "step": 470 + }, + { + "epoch": 0.86, + "grad_norm": 0.2633111059076544, + "learning_rate": 7.173845803291636e-05, + "loss": 1.1721, + "step": 471 + }, + { + "epoch": 0.86, + "grad_norm": 0.30555936322098703, + "learning_rate": 7.168649893644517e-05, + "loss": 1.3011, + "step": 472 + }, + { + "epoch": 0.87, + "grad_norm": 0.23492670111453726, + "learning_rate": 7.163439590084502e-05, + "loss": 1.1601, + "step": 473 + }, + { + "epoch": 0.87, + "grad_norm": 0.26602734263721806, + "learning_rate": 7.158214916279923e-05, + "loss": 1.2808, + "step": 474 + }, + { + "epoch": 0.87, + "grad_norm": 0.3182695007856262, + "learning_rate": 7.152975895964386e-05, + "loss": 1.2967, + "step": 475 + }, + { + "epoch": 0.87, + "grad_norm": 0.2785021674736721, + "learning_rate": 7.147722552936673e-05, + "loss": 1.1789, + "step": 476 + }, + { + "epoch": 0.87, + "grad_norm": 0.279474303138652, + "learning_rate": 7.142454911060627e-05, + "loss": 1.2596, + "step": 477 + }, + { + "epoch": 0.87, + "grad_norm": 0.2556980144910755, + "learning_rate": 7.137172994265044e-05, + "loss": 1.2426, + "step": 478 + }, + { + "epoch": 0.88, + "grad_norm": 0.3311256331993533, + "learning_rate": 7.131876826543565e-05, + "loss": 1.2059, + "step": 479 + }, + { + "epoch": 0.88, + "grad_norm": 0.26467296197775253, + "learning_rate": 7.12656643195457e-05, + "loss": 1.2482, + "step": 480 + }, + { + "epoch": 0.88, + "grad_norm": 0.27444885274652553, + "learning_rate": 7.121241834621064e-05, + "loss": 1.2528, + "step": 481 + }, + { + "epoch": 0.88, + "grad_norm": 0.2572283861115396, + "learning_rate": 7.115903058730567e-05, + "loss": 1.1849, + "step": 482 + }, + { + "epoch": 0.88, + "grad_norm": 0.2677065778235683, + "learning_rate": 7.11055012853501e-05, + "loss": 1.2011, + "step": 483 + }, + { + "epoch": 0.89, + "grad_norm": 0.29470622036742816, + "learning_rate": 7.105183068350619e-05, + "loss": 1.2398, + "step": 484 + }, + { + "epoch": 0.89, + "grad_norm": 0.27609230248969197, + "learning_rate": 7.099801902557811e-05, + "loss": 1.2259, + "step": 485 + }, + { + "epoch": 0.89, + "grad_norm": 0.24248634168099284, + "learning_rate": 7.094406655601073e-05, + "loss": 1.2282, + "step": 486 + }, + { + "epoch": 0.89, + "grad_norm": 0.2765941767688746, + "learning_rate": 7.088997351988865e-05, + "loss": 1.2319, + "step": 487 + }, + { + "epoch": 0.89, + "grad_norm": 0.29347776909858947, + "learning_rate": 7.083574016293493e-05, + "loss": 1.1765, + "step": 488 + }, + { + "epoch": 0.89, + "grad_norm": 0.285370295424537, + "learning_rate": 7.078136673151008e-05, + "loss": 1.26, + "step": 489 + }, + { + "epoch": 0.9, + "grad_norm": 0.29408734903836536, + "learning_rate": 7.072685347261093e-05, + "loss": 1.226, + "step": 490 + }, + { + "epoch": 0.9, + "grad_norm": 0.27437470239205813, + "learning_rate": 7.067220063386947e-05, + "loss": 1.1976, + "step": 491 + }, + { + "epoch": 0.9, + "grad_norm": 0.2680770258777871, + "learning_rate": 7.061740846355176e-05, + "loss": 1.1915, + "step": 492 + }, + { + "epoch": 0.9, + "grad_norm": 0.27200362879502954, + "learning_rate": 7.056247721055678e-05, + "loss": 1.2002, + "step": 493 + }, + { + "epoch": 0.9, + "grad_norm": 0.2637811092577037, + "learning_rate": 7.050740712441528e-05, + "loss": 1.287, + "step": 494 + }, + { + "epoch": 0.91, + "grad_norm": 0.24657959209271266, + "learning_rate": 7.045219845528875e-05, + "loss": 1.2284, + "step": 495 + }, + { + "epoch": 0.91, + "grad_norm": 0.25311992110358666, + "learning_rate": 7.039685145396812e-05, + "loss": 1.1616, + "step": 496 + }, + { + "epoch": 0.91, + "grad_norm": 0.2564633694193358, + "learning_rate": 7.034136637187275e-05, + "loss": 1.2067, + "step": 497 + }, + { + "epoch": 0.91, + "grad_norm": 0.2446797651174144, + "learning_rate": 7.028574346104926e-05, + "loss": 1.2284, + "step": 498 + }, + { + "epoch": 0.91, + "grad_norm": 0.2592751463399255, + "learning_rate": 7.022998297417034e-05, + "loss": 1.2371, + "step": 499 + }, + { + "epoch": 0.91, + "grad_norm": 0.2500713943206808, + "learning_rate": 7.017408516453365e-05, + "loss": 1.1061, + "step": 500 + }, + { + "epoch": 0.92, + "grad_norm": 0.2812266276040743, + "learning_rate": 7.011805028606064e-05, + "loss": 1.1949, + "step": 501 + }, + { + "epoch": 0.92, + "grad_norm": 0.298829667668083, + "learning_rate": 7.006187859329544e-05, + "loss": 1.2313, + "step": 502 + }, + { + "epoch": 0.92, + "grad_norm": 0.26518768159745104, + "learning_rate": 7.000557034140361e-05, + "loss": 1.2246, + "step": 503 + }, + { + "epoch": 0.92, + "grad_norm": 0.3037280360760458, + "learning_rate": 6.994912578617113e-05, + "loss": 1.1617, + "step": 504 + }, + { + "epoch": 0.92, + "grad_norm": 0.2726903109255714, + "learning_rate": 6.989254518400309e-05, + "loss": 1.2415, + "step": 505 + }, + { + "epoch": 0.93, + "grad_norm": 0.25568082003046966, + "learning_rate": 6.98358287919226e-05, + "loss": 1.1817, + "step": 506 + }, + { + "epoch": 0.93, + "grad_norm": 0.25633294893705044, + "learning_rate": 6.97789768675696e-05, + "loss": 1.2149, + "step": 507 + }, + { + "epoch": 0.93, + "grad_norm": 0.28291439435087123, + "learning_rate": 6.972198966919972e-05, + "loss": 1.1578, + "step": 508 + }, + { + "epoch": 0.93, + "grad_norm": 0.27195184756655516, + "learning_rate": 6.966486745568308e-05, + "loss": 1.2355, + "step": 509 + }, + { + "epoch": 0.93, + "grad_norm": 0.239159568376005, + "learning_rate": 6.960761048650312e-05, + "loss": 1.1688, + "step": 510 + }, + { + "epoch": 0.93, + "grad_norm": 0.22961475425949177, + "learning_rate": 6.955021902175543e-05, + "loss": 1.2094, + "step": 511 + }, + { + "epoch": 0.94, + "grad_norm": 0.27443773600741117, + "learning_rate": 6.949269332214651e-05, + "loss": 1.2559, + "step": 512 + }, + { + "epoch": 0.94, + "grad_norm": 0.26230551832002097, + "learning_rate": 6.94350336489927e-05, + "loss": 1.2121, + "step": 513 + }, + { + "epoch": 0.94, + "grad_norm": 0.2716742985303849, + "learning_rate": 6.937724026421892e-05, + "loss": 1.2444, + "step": 514 + }, + { + "epoch": 0.94, + "grad_norm": 0.2537850139439542, + "learning_rate": 6.931931343035742e-05, + "loss": 1.1327, + "step": 515 + }, + { + "epoch": 0.94, + "grad_norm": 0.28599587967496826, + "learning_rate": 6.926125341054676e-05, + "loss": 1.2236, + "step": 516 + }, + { + "epoch": 0.95, + "grad_norm": 0.26780654378470103, + "learning_rate": 6.920306046853043e-05, + "loss": 1.2295, + "step": 517 + }, + { + "epoch": 0.95, + "grad_norm": 0.23606296888412015, + "learning_rate": 6.914473486865577e-05, + "loss": 1.1543, + "step": 518 + }, + { + "epoch": 0.95, + "grad_norm": 0.34976881174240837, + "learning_rate": 6.90862768758727e-05, + "loss": 1.2067, + "step": 519 + }, + { + "epoch": 0.95, + "grad_norm": 0.2481257873494882, + "learning_rate": 6.902768675573258e-05, + "loss": 1.2188, + "step": 520 + }, + { + "epoch": 0.95, + "grad_norm": 0.2996395778117021, + "learning_rate": 6.896896477438699e-05, + "loss": 1.2326, + "step": 521 + }, + { + "epoch": 0.95, + "grad_norm": 0.8839768816333193, + "learning_rate": 6.891011119858643e-05, + "loss": 1.2435, + "step": 522 + }, + { + "epoch": 0.96, + "grad_norm": 0.2851882482058998, + "learning_rate": 6.885112629567927e-05, + "loss": 1.2644, + "step": 523 + }, + { + "epoch": 0.96, + "grad_norm": 0.2813663482913699, + "learning_rate": 6.879201033361035e-05, + "loss": 1.2309, + "step": 524 + }, + { + "epoch": 0.96, + "grad_norm": 0.3257551560135454, + "learning_rate": 6.873276358091996e-05, + "loss": 1.2755, + "step": 525 + }, + { + "epoch": 0.96, + "grad_norm": 0.28930479952494365, + "learning_rate": 6.867338630674247e-05, + "loss": 1.1962, + "step": 526 + }, + { + "epoch": 0.96, + "grad_norm": 0.3077462996938649, + "learning_rate": 6.861387878080511e-05, + "loss": 1.2402, + "step": 527 + }, + { + "epoch": 0.97, + "grad_norm": 0.2848900193452761, + "learning_rate": 6.855424127342688e-05, + "loss": 1.2748, + "step": 528 + }, + { + "epoch": 0.97, + "grad_norm": 0.4765938812802202, + "learning_rate": 6.849447405551718e-05, + "loss": 1.2226, + "step": 529 + }, + { + "epoch": 0.97, + "grad_norm": 0.53184473292579, + "learning_rate": 6.843457739857467e-05, + "loss": 1.2347, + "step": 530 + }, + { + "epoch": 0.97, + "grad_norm": 0.6416239346492343, + "learning_rate": 6.837455157468596e-05, + "loss": 1.2429, + "step": 531 + }, + { + "epoch": 0.97, + "grad_norm": 0.3188092712502773, + "learning_rate": 6.831439685652442e-05, + "loss": 1.216, + "step": 532 + }, + { + "epoch": 0.97, + "grad_norm": 0.3527495731006385, + "learning_rate": 6.825411351734895e-05, + "loss": 1.1682, + "step": 533 + }, + { + "epoch": 0.98, + "grad_norm": 0.29603753744741856, + "learning_rate": 6.819370183100274e-05, + "loss": 1.1434, + "step": 534 + }, + { + "epoch": 0.98, + "grad_norm": 0.5252450389976622, + "learning_rate": 6.813316207191198e-05, + "loss": 1.1943, + "step": 535 + }, + { + "epoch": 0.98, + "grad_norm": 0.32999419558659937, + "learning_rate": 6.807249451508466e-05, + "loss": 1.192, + "step": 536 + }, + { + "epoch": 0.98, + "grad_norm": 0.3650175469778724, + "learning_rate": 6.801169943610929e-05, + "loss": 1.2141, + "step": 537 + }, + { + "epoch": 0.98, + "grad_norm": 1.0643532150783557, + "learning_rate": 6.795077711115368e-05, + "loss": 1.2253, + "step": 538 + }, + { + "epoch": 0.99, + "grad_norm": 0.5041310609130145, + "learning_rate": 6.788972781696363e-05, + "loss": 1.278, + "step": 539 + }, + { + "epoch": 0.99, + "grad_norm": 0.5123058164360991, + "learning_rate": 6.782855183086177e-05, + "loss": 1.2231, + "step": 540 + }, + { + "epoch": 0.99, + "grad_norm": 0.3533015702394419, + "learning_rate": 6.776724943074619e-05, + "loss": 1.2072, + "step": 541 + }, + { + "epoch": 0.99, + "grad_norm": 0.30253964625417207, + "learning_rate": 6.770582089508927e-05, + "loss": 1.1382, + "step": 542 + }, + { + "epoch": 0.99, + "grad_norm": 0.348991618828202, + "learning_rate": 6.764426650293633e-05, + "loss": 1.2079, + "step": 543 + }, + { + "epoch": 0.99, + "grad_norm": 0.46017440578788743, + "learning_rate": 6.758258653390444e-05, + "loss": 1.1813, + "step": 544 + }, + { + "epoch": 1.0, + "grad_norm": 0.31962101755594885, + "learning_rate": 6.75207812681811e-05, + "loss": 1.1339, + "step": 545 + }, + { + "epoch": 1.0, + "grad_norm": 0.37092024548285923, + "learning_rate": 6.745885098652298e-05, + "loss": 1.2591, + "step": 546 + }, + { + "epoch": 1.0, + "grad_norm": 0.32347106450715835, + "learning_rate": 6.739679597025466e-05, + "loss": 1.2017, + "step": 547 + }, + { + "epoch": 1.0, + "grad_norm": 0.39250187112342494, + "learning_rate": 6.733461650126733e-05, + "loss": 1.0933, + "step": 548 + }, + { + "epoch": 1.0, + "grad_norm": 0.473522452217324, + "learning_rate": 6.727231286201752e-05, + "loss": 1.1124, + "step": 549 + }, + { + "epoch": 1.01, + "grad_norm": 0.4809062179622052, + "learning_rate": 6.720988533552582e-05, + "loss": 1.1585, + "step": 550 + }, + { + "epoch": 1.01, + "grad_norm": 0.3529662801059162, + "learning_rate": 6.714733420537559e-05, + "loss": 1.0501, + "step": 551 + }, + { + "epoch": 1.01, + "grad_norm": 0.5958247214391118, + "learning_rate": 6.708465975571168e-05, + "loss": 1.1086, + "step": 552 + }, + { + "epoch": 1.01, + "grad_norm": 0.5341364205022454, + "learning_rate": 6.70218622712391e-05, + "loss": 1.0518, + "step": 553 + }, + { + "epoch": 1.01, + "grad_norm": 0.3601805724462006, + "learning_rate": 6.695894203722181e-05, + "loss": 1.1779, + "step": 554 + }, + { + "epoch": 1.02, + "grad_norm": 0.43410190338280613, + "learning_rate": 6.68958993394813e-05, + "loss": 1.093, + "step": 555 + }, + { + "epoch": 1.02, + "grad_norm": 0.46217742572873594, + "learning_rate": 6.683273446439546e-05, + "loss": 1.0117, + "step": 556 + }, + { + "epoch": 1.02, + "grad_norm": 0.8591682373623357, + "learning_rate": 6.676944769889708e-05, + "loss": 1.1002, + "step": 557 + }, + { + "epoch": 1.02, + "grad_norm": 0.7383229487622726, + "learning_rate": 6.670603933047272e-05, + "loss": 1.0779, + "step": 558 + }, + { + "epoch": 1.02, + "grad_norm": 0.5965305891207813, + "learning_rate": 6.664250964716131e-05, + "loss": 1.0889, + "step": 559 + }, + { + "epoch": 1.02, + "grad_norm": 0.6030858606684543, + "learning_rate": 6.657885893755288e-05, + "loss": 1.0982, + "step": 560 + }, + { + "epoch": 1.03, + "grad_norm": 0.4644510682398409, + "learning_rate": 6.65150874907872e-05, + "loss": 1.1004, + "step": 561 + }, + { + "epoch": 1.03, + "grad_norm": 0.43943285132452564, + "learning_rate": 6.645119559655254e-05, + "loss": 1.0536, + "step": 562 + }, + { + "epoch": 1.03, + "grad_norm": 0.4456395978600012, + "learning_rate": 6.638718354508427e-05, + "loss": 1.0733, + "step": 563 + }, + { + "epoch": 1.03, + "grad_norm": 0.3303824433217466, + "learning_rate": 6.632305162716365e-05, + "loss": 1.0552, + "step": 564 + }, + { + "epoch": 1.03, + "grad_norm": 0.3617704823170143, + "learning_rate": 6.62588001341164e-05, + "loss": 1.1092, + "step": 565 + }, + { + "epoch": 1.04, + "grad_norm": 0.4465013349903427, + "learning_rate": 6.619442935781141e-05, + "loss": 1.0781, + "step": 566 + }, + { + "epoch": 1.04, + "grad_norm": 0.48516780613791277, + "learning_rate": 6.612993959065947e-05, + "loss": 1.0686, + "step": 567 + }, + { + "epoch": 1.04, + "grad_norm": 0.38867820318536633, + "learning_rate": 6.606533112561186e-05, + "loss": 1.1215, + "step": 568 + }, + { + "epoch": 1.04, + "grad_norm": 0.38566119820378336, + "learning_rate": 6.600060425615907e-05, + "loss": 1.1213, + "step": 569 + }, + { + "epoch": 1.04, + "grad_norm": 0.35534855445058544, + "learning_rate": 6.593575927632947e-05, + "loss": 1.0955, + "step": 570 + }, + { + "epoch": 1.04, + "grad_norm": 0.38124406233349717, + "learning_rate": 6.587079648068795e-05, + "loss": 1.0659, + "step": 571 + }, + { + "epoch": 1.05, + "grad_norm": 0.454750160923548, + "learning_rate": 6.580571616433457e-05, + "loss": 1.1149, + "step": 572 + }, + { + "epoch": 1.05, + "grad_norm": 0.35353190088025255, + "learning_rate": 6.574051862290325e-05, + "loss": 1.0388, + "step": 573 + }, + { + "epoch": 1.05, + "grad_norm": 0.3249395594793626, + "learning_rate": 6.567520415256045e-05, + "loss": 1.0784, + "step": 574 + }, + { + "epoch": 1.05, + "grad_norm": 0.40078898818247227, + "learning_rate": 6.560977305000375e-05, + "loss": 1.0859, + "step": 575 + }, + { + "epoch": 1.05, + "grad_norm": 0.4115264795060035, + "learning_rate": 6.554422561246054e-05, + "loss": 1.1828, + "step": 576 + }, + { + "epoch": 1.06, + "grad_norm": 0.30090229228069215, + "learning_rate": 6.54785621376867e-05, + "loss": 1.0901, + "step": 577 + }, + { + "epoch": 1.06, + "grad_norm": 0.28827860350299206, + "learning_rate": 6.541278292396523e-05, + "loss": 1.0277, + "step": 578 + }, + { + "epoch": 1.06, + "grad_norm": 0.34690404488996757, + "learning_rate": 6.534688827010484e-05, + "loss": 1.048, + "step": 579 + }, + { + "epoch": 1.06, + "grad_norm": 0.29943113556644785, + "learning_rate": 6.528087847543867e-05, + "loss": 1.0646, + "step": 580 + }, + { + "epoch": 1.06, + "grad_norm": 0.37318202575874415, + "learning_rate": 6.521475383982291e-05, + "loss": 1.1091, + "step": 581 + }, + { + "epoch": 1.06, + "grad_norm": 0.3049663659203959, + "learning_rate": 6.51485146636354e-05, + "loss": 1.0552, + "step": 582 + }, + { + "epoch": 1.07, + "grad_norm": 0.3342407867509692, + "learning_rate": 6.508216124777431e-05, + "loss": 1.2227, + "step": 583 + }, + { + "epoch": 1.07, + "grad_norm": 0.3348396047855952, + "learning_rate": 6.501569389365674e-05, + "loss": 1.0861, + "step": 584 + }, + { + "epoch": 1.07, + "grad_norm": 0.30951429367513383, + "learning_rate": 6.494911290321737e-05, + "loss": 1.0461, + "step": 585 + }, + { + "epoch": 1.07, + "grad_norm": 0.33898401361064606, + "learning_rate": 6.488241857890711e-05, + "loss": 1.0854, + "step": 586 + }, + { + "epoch": 1.07, + "grad_norm": 0.4901462068263497, + "learning_rate": 6.481561122369164e-05, + "loss": 1.1012, + "step": 587 + }, + { + "epoch": 1.08, + "grad_norm": 0.3179574879809652, + "learning_rate": 6.474869114105018e-05, + "loss": 1.0451, + "step": 588 + }, + { + "epoch": 1.08, + "grad_norm": 0.32159328915060714, + "learning_rate": 6.468165863497395e-05, + "loss": 1.0458, + "step": 589 + }, + { + "epoch": 1.08, + "grad_norm": 0.36462235008537297, + "learning_rate": 6.461451400996491e-05, + "loss": 1.1247, + "step": 590 + }, + { + "epoch": 1.08, + "grad_norm": 0.5373862753611778, + "learning_rate": 6.454725757103432e-05, + "loss": 1.0542, + "step": 591 + }, + { + "epoch": 1.08, + "grad_norm": 0.3160409270291303, + "learning_rate": 6.447988962370133e-05, + "loss": 1.0829, + "step": 592 + }, + { + "epoch": 1.08, + "grad_norm": 0.390452102978435, + "learning_rate": 6.441241047399169e-05, + "loss": 1.192, + "step": 593 + }, + { + "epoch": 1.09, + "grad_norm": 0.3802122712014928, + "learning_rate": 6.434482042843627e-05, + "loss": 1.1153, + "step": 594 + }, + { + "epoch": 1.09, + "grad_norm": 0.4081584328242501, + "learning_rate": 6.427711979406966e-05, + "loss": 1.1635, + "step": 595 + }, + { + "epoch": 1.09, + "grad_norm": 0.3791962989638633, + "learning_rate": 6.420930887842889e-05, + "loss": 1.1581, + "step": 596 + }, + { + "epoch": 1.09, + "grad_norm": 0.33239440056484193, + "learning_rate": 6.414138798955189e-05, + "loss": 1.0926, + "step": 597 + }, + { + "epoch": 1.09, + "grad_norm": 0.3279881540815014, + "learning_rate": 6.407335743597616e-05, + "loss": 1.1386, + "step": 598 + }, + { + "epoch": 1.1, + "grad_norm": 0.30309644763750837, + "learning_rate": 6.40052175267374e-05, + "loss": 1.0523, + "step": 599 + }, + { + "epoch": 1.1, + "grad_norm": 0.3349097308403333, + "learning_rate": 6.393696857136801e-05, + "loss": 1.0815, + "step": 600 + }, + { + "epoch": 1.1, + "grad_norm": 0.3288227593556618, + "learning_rate": 6.386861087989581e-05, + "loss": 1.015, + "step": 601 + }, + { + "epoch": 1.1, + "grad_norm": 0.36685586740843157, + "learning_rate": 6.380014476284255e-05, + "loss": 1.1232, + "step": 602 + }, + { + "epoch": 1.1, + "grad_norm": 0.3620977714204643, + "learning_rate": 6.373157053122243e-05, + "loss": 1.1138, + "step": 603 + }, + { + "epoch": 1.1, + "grad_norm": 0.3130587018197183, + "learning_rate": 6.366288849654091e-05, + "loss": 1.1255, + "step": 604 + }, + { + "epoch": 1.11, + "grad_norm": 0.3602737087072766, + "learning_rate": 6.359409897079303e-05, + "loss": 1.0282, + "step": 605 + }, + { + "epoch": 1.11, + "grad_norm": 0.31168852571991945, + "learning_rate": 6.352520226646222e-05, + "loss": 1.0779, + "step": 606 + }, + { + "epoch": 1.11, + "grad_norm": 0.3516045580189353, + "learning_rate": 6.345619869651871e-05, + "loss": 1.1028, + "step": 607 + }, + { + "epoch": 1.11, + "grad_norm": 0.3231857927563657, + "learning_rate": 6.33870885744182e-05, + "loss": 1.1202, + "step": 608 + }, + { + "epoch": 1.11, + "grad_norm": 0.30205205129701157, + "learning_rate": 6.331787221410041e-05, + "loss": 1.1369, + "step": 609 + }, + { + "epoch": 1.12, + "grad_norm": 0.3198359813888166, + "learning_rate": 6.32485499299877e-05, + "loss": 1.1763, + "step": 610 + }, + { + "epoch": 1.12, + "grad_norm": 0.3128641370321787, + "learning_rate": 6.31791220369835e-05, + "loss": 1.0223, + "step": 611 + }, + { + "epoch": 1.12, + "grad_norm": 0.2989105616213649, + "learning_rate": 6.31095888504711e-05, + "loss": 1.0358, + "step": 612 + }, + { + "epoch": 1.12, + "grad_norm": 0.3103537906853337, + "learning_rate": 6.303995068631203e-05, + "loss": 1.1261, + "step": 613 + }, + { + "epoch": 1.12, + "grad_norm": 0.28598715532508207, + "learning_rate": 6.297020786084467e-05, + "loss": 1.0629, + "step": 614 + }, + { + "epoch": 1.12, + "grad_norm": 0.29809789918093255, + "learning_rate": 6.290036069088288e-05, + "loss": 1.035, + "step": 615 + }, + { + "epoch": 1.13, + "grad_norm": 0.33765270252261453, + "learning_rate": 6.283040949371451e-05, + "loss": 1.1221, + "step": 616 + }, + { + "epoch": 1.13, + "grad_norm": 0.3424617501293415, + "learning_rate": 6.276035458709993e-05, + "loss": 1.155, + "step": 617 + }, + { + "epoch": 1.13, + "grad_norm": 0.3799189737987811, + "learning_rate": 6.269019628927067e-05, + "loss": 1.0701, + "step": 618 + }, + { + "epoch": 1.13, + "grad_norm": 0.3358898935253196, + "learning_rate": 6.261993491892791e-05, + "loss": 1.1649, + "step": 619 + }, + { + "epoch": 1.13, + "grad_norm": 0.31569979424117356, + "learning_rate": 6.254957079524099e-05, + "loss": 1.0633, + "step": 620 + }, + { + "epoch": 1.14, + "grad_norm": 0.3002168156888237, + "learning_rate": 6.247910423784609e-05, + "loss": 1.0846, + "step": 621 + }, + { + "epoch": 1.14, + "grad_norm": 0.3097238823450595, + "learning_rate": 6.24085355668447e-05, + "loss": 1.0808, + "step": 622 + }, + { + "epoch": 1.14, + "grad_norm": 0.3120312761417578, + "learning_rate": 6.233786510280212e-05, + "loss": 1.0142, + "step": 623 + }, + { + "epoch": 1.14, + "grad_norm": 0.3335343015064923, + "learning_rate": 6.22670931667461e-05, + "loss": 1.0674, + "step": 624 + }, + { + "epoch": 1.14, + "grad_norm": 0.3234062304634526, + "learning_rate": 6.219622008016533e-05, + "loss": 1.0981, + "step": 625 + }, + { + "epoch": 1.14, + "grad_norm": 0.32152678786547273, + "learning_rate": 6.212524616500798e-05, + "loss": 1.0244, + "step": 626 + }, + { + "epoch": 1.15, + "grad_norm": 0.39031977608147594, + "learning_rate": 6.205417174368023e-05, + "loss": 1.1205, + "step": 627 + }, + { + "epoch": 1.15, + "grad_norm": 0.3806189090017157, + "learning_rate": 6.198299713904485e-05, + "loss": 1.1134, + "step": 628 + }, + { + "epoch": 1.15, + "grad_norm": 0.2978349276971668, + "learning_rate": 6.191172267441967e-05, + "loss": 1.0088, + "step": 629 + }, + { + "epoch": 1.15, + "grad_norm": 0.3190354077382501, + "learning_rate": 6.184034867357617e-05, + "loss": 1.108, + "step": 630 + }, + { + "epoch": 1.15, + "grad_norm": 0.32633048665038994, + "learning_rate": 6.176887546073797e-05, + "loss": 1.0825, + "step": 631 + }, + { + "epoch": 1.16, + "grad_norm": 0.3428026413020903, + "learning_rate": 6.169730336057939e-05, + "loss": 1.0765, + "step": 632 + }, + { + "epoch": 1.16, + "grad_norm": 0.3475737151929015, + "learning_rate": 6.162563269822391e-05, + "loss": 1.0693, + "step": 633 + }, + { + "epoch": 1.16, + "grad_norm": 0.3870252154591392, + "learning_rate": 6.15538637992428e-05, + "loss": 1.1081, + "step": 634 + }, + { + "epoch": 1.16, + "grad_norm": 0.33597355193652834, + "learning_rate": 6.148199698965352e-05, + "loss": 1.0893, + "step": 635 + }, + { + "epoch": 1.16, + "grad_norm": 0.30805894179787247, + "learning_rate": 6.141003259591834e-05, + "loss": 1.0995, + "step": 636 + }, + { + "epoch": 1.17, + "grad_norm": 0.3025073882734066, + "learning_rate": 6.133797094494281e-05, + "loss": 1.0388, + "step": 637 + }, + { + "epoch": 1.17, + "grad_norm": 0.3524395196391662, + "learning_rate": 6.126581236407429e-05, + "loss": 1.1196, + "step": 638 + }, + { + "epoch": 1.17, + "grad_norm": 0.3377646188130345, + "learning_rate": 6.119355718110039e-05, + "loss": 1.0382, + "step": 639 + }, + { + "epoch": 1.17, + "grad_norm": 0.35508400659785483, + "learning_rate": 6.112120572424763e-05, + "loss": 1.1402, + "step": 640 + }, + { + "epoch": 1.17, + "grad_norm": 0.3454418793700457, + "learning_rate": 6.104875832217982e-05, + "loss": 1.1032, + "step": 641 + }, + { + "epoch": 1.17, + "grad_norm": 0.32629806837059866, + "learning_rate": 6.097621530399661e-05, + "loss": 1.0959, + "step": 642 + }, + { + "epoch": 1.18, + "grad_norm": 0.3329536837751315, + "learning_rate": 6.090357699923202e-05, + "loss": 1.0467, + "step": 643 + }, + { + "epoch": 1.18, + "grad_norm": 0.32302233828349475, + "learning_rate": 6.083084373785287e-05, + "loss": 1.0858, + "step": 644 + }, + { + "epoch": 1.18, + "grad_norm": 0.3310358826507611, + "learning_rate": 6.075801585025739e-05, + "loss": 1.0715, + "step": 645 + }, + { + "epoch": 1.18, + "grad_norm": 0.319322035854079, + "learning_rate": 6.068509366727362e-05, + "loss": 1.177, + "step": 646 + }, + { + "epoch": 1.18, + "grad_norm": 0.3065230667302707, + "learning_rate": 6.061207752015797e-05, + "loss": 1.0649, + "step": 647 + }, + { + "epoch": 1.19, + "grad_norm": 0.29926795565748227, + "learning_rate": 6.053896774059368e-05, + "loss": 1.1325, + "step": 648 + }, + { + "epoch": 1.19, + "grad_norm": 0.3556069634279046, + "learning_rate": 6.046576466068931e-05, + "loss": 1.1366, + "step": 649 + }, + { + "epoch": 1.19, + "grad_norm": 0.3189191131461966, + "learning_rate": 6.039246861297727e-05, + "loss": 1.0693, + "step": 650 + }, + { + "epoch": 1.19, + "grad_norm": 0.3347197156648834, + "learning_rate": 6.031907993041227e-05, + "loss": 1.1009, + "step": 651 + }, + { + "epoch": 1.19, + "grad_norm": 0.32274156348185445, + "learning_rate": 6.0245598946369826e-05, + "loss": 1.1675, + "step": 652 + }, + { + "epoch": 1.19, + "grad_norm": 0.35534089035455224, + "learning_rate": 6.017202599464476e-05, + "loss": 1.1723, + "step": 653 + }, + { + "epoch": 1.2, + "grad_norm": 0.3106026578570133, + "learning_rate": 6.009836140944965e-05, + "loss": 1.0954, + "step": 654 + }, + { + "epoch": 1.2, + "grad_norm": 0.3309144454564729, + "learning_rate": 6.002460552541331e-05, + "loss": 1.0209, + "step": 655 + }, + { + "epoch": 1.2, + "grad_norm": 0.3023619281400003, + "learning_rate": 5.9950758677579345e-05, + "loss": 1.0363, + "step": 656 + }, + { + "epoch": 1.2, + "grad_norm": 0.3311182880219704, + "learning_rate": 5.987682120140451e-05, + "loss": 1.0515, + "step": 657 + }, + { + "epoch": 1.2, + "grad_norm": 0.33396486010030413, + "learning_rate": 5.980279343275729e-05, + "loss": 1.1251, + "step": 658 + }, + { + "epoch": 1.21, + "grad_norm": 0.3465764556678002, + "learning_rate": 5.97286757079163e-05, + "loss": 1.165, + "step": 659 + }, + { + "epoch": 1.21, + "grad_norm": 0.304193441363374, + "learning_rate": 5.965446836356882e-05, + "loss": 1.0228, + "step": 660 + }, + { + "epoch": 1.21, + "grad_norm": 0.3415149030413082, + "learning_rate": 5.9580171736809224e-05, + "loss": 1.0742, + "step": 661 + }, + { + "epoch": 1.21, + "grad_norm": 0.33138658321132064, + "learning_rate": 5.950578616513746e-05, + "loss": 1.0843, + "step": 662 + }, + { + "epoch": 1.21, + "grad_norm": 0.30774403421162994, + "learning_rate": 5.943131198645752e-05, + "loss": 1.065, + "step": 663 + }, + { + "epoch": 1.21, + "grad_norm": 0.3428877492183819, + "learning_rate": 5.9356749539075885e-05, + "loss": 1.1101, + "step": 664 + }, + { + "epoch": 1.22, + "grad_norm": 0.3621290546130101, + "learning_rate": 5.928209916170003e-05, + "loss": 1.1372, + "step": 665 + }, + { + "epoch": 1.22, + "grad_norm": 0.3482375945469884, + "learning_rate": 5.9207361193436865e-05, + "loss": 1.132, + "step": 666 + }, + { + "epoch": 1.22, + "grad_norm": 0.31754384974068384, + "learning_rate": 5.9132535973791156e-05, + "loss": 1.148, + "step": 667 + }, + { + "epoch": 1.22, + "grad_norm": 0.36003834782050365, + "learning_rate": 5.9057623842664044e-05, + "loss": 1.1099, + "step": 668 + }, + { + "epoch": 1.22, + "grad_norm": 0.2963701622969662, + "learning_rate": 5.8982625140351464e-05, + "loss": 1.0755, + "step": 669 + }, + { + "epoch": 1.23, + "grad_norm": 0.32579569606066516, + "learning_rate": 5.8907540207542616e-05, + "loss": 1.0809, + "step": 670 + }, + { + "epoch": 1.23, + "grad_norm": 0.4247563451753457, + "learning_rate": 5.8832369385318416e-05, + "loss": 1.097, + "step": 671 + }, + { + "epoch": 1.23, + "grad_norm": 0.33076932102169776, + "learning_rate": 5.875711301514992e-05, + "loss": 1.1078, + "step": 672 + }, + { + "epoch": 1.23, + "grad_norm": 0.3609238032332309, + "learning_rate": 5.8681771438896815e-05, + "loss": 1.1031, + "step": 673 + }, + { + "epoch": 1.23, + "grad_norm": 0.325159585649425, + "learning_rate": 5.860634499880583e-05, + "loss": 1.0707, + "step": 674 + }, + { + "epoch": 1.23, + "grad_norm": 0.4620687271068983, + "learning_rate": 5.853083403750922e-05, + "loss": 1.1017, + "step": 675 + }, + { + "epoch": 1.24, + "grad_norm": 0.33485279064365936, + "learning_rate": 5.845523889802316e-05, + "loss": 1.0989, + "step": 676 + }, + { + "epoch": 1.24, + "grad_norm": 0.30952573170841513, + "learning_rate": 5.8379559923746214e-05, + "loss": 1.0393, + "step": 677 + }, + { + "epoch": 1.24, + "grad_norm": 0.33498605810588283, + "learning_rate": 5.830379745845781e-05, + "loss": 1.1259, + "step": 678 + }, + { + "epoch": 1.24, + "grad_norm": 0.35771921163037307, + "learning_rate": 5.822795184631659e-05, + "loss": 1.0815, + "step": 679 + }, + { + "epoch": 1.24, + "grad_norm": 0.3329650192347647, + "learning_rate": 5.815202343185894e-05, + "loss": 1.1344, + "step": 680 + }, + { + "epoch": 1.25, + "grad_norm": 0.3356634465845771, + "learning_rate": 5.807601255999736e-05, + "loss": 1.1297, + "step": 681 + }, + { + "epoch": 1.25, + "grad_norm": 0.3289442034151235, + "learning_rate": 5.7999919576018934e-05, + "loss": 1.022, + "step": 682 + }, + { + "epoch": 1.25, + "grad_norm": 0.3207007334784113, + "learning_rate": 5.7923744825583745e-05, + "loss": 1.0571, + "step": 683 + }, + { + "epoch": 1.25, + "grad_norm": 0.3582460325329284, + "learning_rate": 5.7847488654723304e-05, + "loss": 1.0778, + "step": 684 + }, + { + "epoch": 1.25, + "grad_norm": 0.3563317666176927, + "learning_rate": 5.777115140983899e-05, + "loss": 1.1003, + "step": 685 + }, + { + "epoch": 1.25, + "grad_norm": 3.4694912945702105, + "learning_rate": 5.769473343770047e-05, + "loss": 1.121, + "step": 686 + }, + { + "epoch": 1.26, + "grad_norm": 0.43002349520483113, + "learning_rate": 5.761823508544411e-05, + "loss": 1.0765, + "step": 687 + }, + { + "epoch": 1.26, + "grad_norm": 0.39467783104839754, + "learning_rate": 5.754165670057142e-05, + "loss": 1.0788, + "step": 688 + }, + { + "epoch": 1.26, + "grad_norm": 0.39629029674867916, + "learning_rate": 5.7464998630947464e-05, + "loss": 1.0812, + "step": 689 + }, + { + "epoch": 1.26, + "grad_norm": 0.3880152093965208, + "learning_rate": 5.738826122479929e-05, + "loss": 1.1228, + "step": 690 + }, + { + "epoch": 1.26, + "grad_norm": 0.3777874121959188, + "learning_rate": 5.7311444830714324e-05, + "loss": 1.0907, + "step": 691 + }, + { + "epoch": 1.27, + "grad_norm": 0.38004041653523696, + "learning_rate": 5.723454979763882e-05, + "loss": 1.1263, + "step": 692 + }, + { + "epoch": 1.27, + "grad_norm": 0.37049672627797636, + "learning_rate": 5.7157576474876246e-05, + "loss": 1.1438, + "step": 693 + }, + { + "epoch": 1.27, + "grad_norm": 0.32973606103437614, + "learning_rate": 5.7080525212085725e-05, + "loss": 1.0553, + "step": 694 + }, + { + "epoch": 1.27, + "grad_norm": 0.31674639252070325, + "learning_rate": 5.700339635928038e-05, + "loss": 1.06, + "step": 695 + }, + { + "epoch": 1.27, + "grad_norm": 0.32282199426553837, + "learning_rate": 5.692619026682588e-05, + "loss": 1.0841, + "step": 696 + }, + { + "epoch": 1.27, + "grad_norm": 0.4810882958061859, + "learning_rate": 5.684890728543869e-05, + "loss": 1.0803, + "step": 697 + }, + { + "epoch": 1.28, + "grad_norm": 0.3995638550178378, + "learning_rate": 5.6771547766184566e-05, + "loss": 1.1187, + "step": 698 + }, + { + "epoch": 1.28, + "grad_norm": 0.35264932960583484, + "learning_rate": 5.669411206047699e-05, + "loss": 1.0641, + "step": 699 + }, + { + "epoch": 1.28, + "grad_norm": 0.35240640524733, + "learning_rate": 5.661660052007547e-05, + "loss": 1.076, + "step": 700 + }, + { + "epoch": 1.28, + "grad_norm": 0.3540694609860389, + "learning_rate": 5.653901349708401e-05, + "loss": 1.1369, + "step": 701 + }, + { + "epoch": 1.28, + "grad_norm": 0.3196055112925304, + "learning_rate": 5.646135134394955e-05, + "loss": 1.0677, + "step": 702 + }, + { + "epoch": 1.29, + "grad_norm": 0.4214141007955914, + "learning_rate": 5.6383614413460266e-05, + "loss": 1.1139, + "step": 703 + }, + { + "epoch": 1.29, + "grad_norm": 0.3625611311798579, + "learning_rate": 5.630580305874402e-05, + "loss": 1.1845, + "step": 704 + }, + { + "epoch": 1.29, + "grad_norm": 0.3425208672181188, + "learning_rate": 5.62279176332668e-05, + "loss": 1.174, + "step": 705 + }, + { + "epoch": 1.29, + "grad_norm": 0.3108419862818321, + "learning_rate": 5.6149958490830996e-05, + "loss": 1.0331, + "step": 706 + }, + { + "epoch": 1.29, + "grad_norm": 0.3274644181571904, + "learning_rate": 5.607192598557394e-05, + "loss": 1.0664, + "step": 707 + }, + { + "epoch": 1.29, + "grad_norm": 0.346218197215145, + "learning_rate": 5.599382047196617e-05, + "loss": 1.2088, + "step": 708 + }, + { + "epoch": 1.3, + "grad_norm": 0.328497632267458, + "learning_rate": 5.591564230480989e-05, + "loss": 1.0287, + "step": 709 + }, + { + "epoch": 1.3, + "grad_norm": 0.3708173720611468, + "learning_rate": 5.583739183923732e-05, + "loss": 1.0883, + "step": 710 + }, + { + "epoch": 1.3, + "grad_norm": 0.3631427403535479, + "learning_rate": 5.575906943070915e-05, + "loss": 1.1155, + "step": 711 + }, + { + "epoch": 1.3, + "grad_norm": 0.3305201458598695, + "learning_rate": 5.5680675435012834e-05, + "loss": 1.0958, + "step": 712 + }, + { + "epoch": 1.3, + "grad_norm": 0.34978833532083714, + "learning_rate": 5.5602210208261036e-05, + "loss": 1.1437, + "step": 713 + }, + { + "epoch": 1.31, + "grad_norm": 0.3510553882510229, + "learning_rate": 5.552367410688999e-05, + "loss": 1.0941, + "step": 714 + }, + { + "epoch": 1.31, + "grad_norm": 0.3523747462465078, + "learning_rate": 5.544506748765789e-05, + "loss": 1.1289, + "step": 715 + }, + { + "epoch": 1.31, + "grad_norm": 0.38262637783927445, + "learning_rate": 5.5366390707643266e-05, + "loss": 1.099, + "step": 716 + }, + { + "epoch": 1.31, + "grad_norm": 0.38620065989073454, + "learning_rate": 5.528764412424334e-05, + "loss": 1.083, + "step": 717 + }, + { + "epoch": 1.31, + "grad_norm": 0.3401355276121096, + "learning_rate": 5.520882809517245e-05, + "loss": 1.028, + "step": 718 + }, + { + "epoch": 1.32, + "grad_norm": 0.3392061008943934, + "learning_rate": 5.512994297846039e-05, + "loss": 1.1083, + "step": 719 + }, + { + "epoch": 1.32, + "grad_norm": 0.34219480421015414, + "learning_rate": 5.505098913245077e-05, + "loss": 1.1108, + "step": 720 + }, + { + "epoch": 1.32, + "grad_norm": 0.3275058061553761, + "learning_rate": 5.497196691579945e-05, + "loss": 1.111, + "step": 721 + }, + { + "epoch": 1.32, + "grad_norm": 0.36800249746509384, + "learning_rate": 5.489287668747283e-05, + "loss": 1.1221, + "step": 722 + }, + { + "epoch": 1.32, + "grad_norm": 0.4129005533101575, + "learning_rate": 5.481371880674628e-05, + "loss": 1.0966, + "step": 723 + }, + { + "epoch": 1.32, + "grad_norm": 0.36563906596251655, + "learning_rate": 5.4734493633202505e-05, + "loss": 1.0927, + "step": 724 + }, + { + "epoch": 1.33, + "grad_norm": 0.3614650536839971, + "learning_rate": 5.465520152672986e-05, + "loss": 1.13, + "step": 725 + }, + { + "epoch": 1.33, + "grad_norm": 0.36419665098633497, + "learning_rate": 5.4575842847520765e-05, + "loss": 1.1183, + "step": 726 + }, + { + "epoch": 1.33, + "grad_norm": 0.34490689807258995, + "learning_rate": 5.449641795607005e-05, + "loss": 1.0919, + "step": 727 + }, + { + "epoch": 1.33, + "grad_norm": 0.3627643746876298, + "learning_rate": 5.441692721317334e-05, + "loss": 1.0411, + "step": 728 + }, + { + "epoch": 1.33, + "grad_norm": 0.323620411949565, + "learning_rate": 5.433737097992537e-05, + "loss": 1.0725, + "step": 729 + }, + { + "epoch": 1.34, + "grad_norm": 0.3521599501824965, + "learning_rate": 5.425774961771838e-05, + "loss": 1.0926, + "step": 730 + }, + { + "epoch": 1.34, + "grad_norm": 0.3302390546764222, + "learning_rate": 5.417806348824047e-05, + "loss": 1.0468, + "step": 731 + }, + { + "epoch": 1.34, + "grad_norm": 0.3833325802616019, + "learning_rate": 5.4098312953473956e-05, + "loss": 1.1291, + "step": 732 + }, + { + "epoch": 1.34, + "grad_norm": 0.3708621126835512, + "learning_rate": 5.401849837569372e-05, + "loss": 1.0887, + "step": 733 + }, + { + "epoch": 1.34, + "grad_norm": 0.3625834373416278, + "learning_rate": 5.393862011746555e-05, + "loss": 1.0981, + "step": 734 + }, + { + "epoch": 1.34, + "grad_norm": 0.3583343965080617, + "learning_rate": 5.385867854164451e-05, + "loss": 1.1021, + "step": 735 + }, + { + "epoch": 1.35, + "grad_norm": 0.34598320594096066, + "learning_rate": 5.377867401137332e-05, + "loss": 1.1376, + "step": 736 + }, + { + "epoch": 1.35, + "grad_norm": 0.3046382791315433, + "learning_rate": 5.369860689008066e-05, + "loss": 1.0206, + "step": 737 + }, + { + "epoch": 1.35, + "grad_norm": 0.34464948380043725, + "learning_rate": 5.3618477541479505e-05, + "loss": 1.1084, + "step": 738 + }, + { + "epoch": 1.35, + "grad_norm": 0.3203242519627101, + "learning_rate": 5.353828632956557e-05, + "loss": 1.0731, + "step": 739 + }, + { + "epoch": 1.35, + "grad_norm": 0.3431169960355163, + "learning_rate": 5.3458033618615516e-05, + "loss": 1.091, + "step": 740 + }, + { + "epoch": 1.36, + "grad_norm": 0.33492074521678705, + "learning_rate": 5.337771977318543e-05, + "loss": 1.1112, + "step": 741 + }, + { + "epoch": 1.36, + "grad_norm": 0.32576546585541344, + "learning_rate": 5.3297345158109086e-05, + "loss": 1.0993, + "step": 742 + }, + { + "epoch": 1.36, + "grad_norm": 0.3410007245037574, + "learning_rate": 5.3216910138496286e-05, + "loss": 1.094, + "step": 743 + }, + { + "epoch": 1.36, + "grad_norm": 0.34891180680896833, + "learning_rate": 5.313641507973128e-05, + "loss": 1.1331, + "step": 744 + }, + { + "epoch": 1.36, + "grad_norm": 0.37135766946717214, + "learning_rate": 5.3055860347471006e-05, + "loss": 1.1, + "step": 745 + }, + { + "epoch": 1.36, + "grad_norm": 0.3465019415478411, + "learning_rate": 5.297524630764349e-05, + "loss": 1.1256, + "step": 746 + }, + { + "epoch": 1.37, + "grad_norm": 0.37035388481626563, + "learning_rate": 5.289457332644615e-05, + "loss": 1.0366, + "step": 747 + }, + { + "epoch": 1.37, + "grad_norm": 0.33853883270759155, + "learning_rate": 5.281384177034421e-05, + "loss": 1.0547, + "step": 748 + }, + { + "epoch": 1.37, + "grad_norm": 0.364306618627317, + "learning_rate": 5.2733052006068897e-05, + "loss": 1.0768, + "step": 749 + }, + { + "epoch": 1.37, + "grad_norm": 0.4021754315731627, + "learning_rate": 5.2652204400615916e-05, + "loss": 1.1382, + "step": 750 + }, + { + "epoch": 1.37, + "grad_norm": 0.3332185389039008, + "learning_rate": 5.257129932124368e-05, + "loss": 1.0815, + "step": 751 + }, + { + "epoch": 1.38, + "grad_norm": 0.3453105709879854, + "learning_rate": 5.249033713547173e-05, + "loss": 1.1109, + "step": 752 + }, + { + "epoch": 1.38, + "grad_norm": 0.3385397539717797, + "learning_rate": 5.2409318211078966e-05, + "loss": 1.0529, + "step": 753 + }, + { + "epoch": 1.38, + "grad_norm": 0.33197994450130447, + "learning_rate": 5.232824291610206e-05, + "loss": 1.0721, + "step": 754 + }, + { + "epoch": 1.38, + "grad_norm": 0.32836289576124167, + "learning_rate": 5.224711161883375e-05, + "loss": 1.0459, + "step": 755 + }, + { + "epoch": 1.38, + "grad_norm": 0.32491620058831744, + "learning_rate": 5.216592468782117e-05, + "loss": 1.0897, + "step": 756 + }, + { + "epoch": 1.38, + "grad_norm": 0.3137879047811153, + "learning_rate": 5.2084682491864155e-05, + "loss": 1.096, + "step": 757 + }, + { + "epoch": 1.39, + "grad_norm": 0.3356938043023012, + "learning_rate": 5.200338540001364e-05, + "loss": 1.0827, + "step": 758 + }, + { + "epoch": 1.39, + "grad_norm": 0.36044340490819055, + "learning_rate": 5.192203378156984e-05, + "loss": 1.0617, + "step": 759 + }, + { + "epoch": 1.39, + "grad_norm": 0.34674262047888293, + "learning_rate": 5.184062800608077e-05, + "loss": 1.1267, + "step": 760 + }, + { + "epoch": 1.39, + "grad_norm": 0.32469442322149333, + "learning_rate": 5.1759168443340375e-05, + "loss": 1.1483, + "step": 761 + }, + { + "epoch": 1.39, + "grad_norm": 0.3290384307774216, + "learning_rate": 5.167765546338698e-05, + "loss": 1.047, + "step": 762 + }, + { + "epoch": 1.4, + "grad_norm": 0.31637612188770403, + "learning_rate": 5.1596089436501525e-05, + "loss": 1.0311, + "step": 763 + }, + { + "epoch": 1.4, + "grad_norm": 0.3168693829641207, + "learning_rate": 5.151447073320597e-05, + "loss": 1.1405, + "step": 764 + }, + { + "epoch": 1.4, + "grad_norm": 0.34322421571238926, + "learning_rate": 5.143279972426153e-05, + "loss": 1.1428, + "step": 765 + }, + { + "epoch": 1.4, + "grad_norm": 0.3291030435830325, + "learning_rate": 5.1351076780667026e-05, + "loss": 1.0473, + "step": 766 + }, + { + "epoch": 1.4, + "grad_norm": 0.33772039158758044, + "learning_rate": 5.1269302273657195e-05, + "loss": 1.0909, + "step": 767 + }, + { + "epoch": 1.4, + "grad_norm": 0.3802031736890876, + "learning_rate": 5.118747657470102e-05, + "loss": 1.1482, + "step": 768 + }, + { + "epoch": 1.41, + "grad_norm": 0.3296067628997962, + "learning_rate": 5.1105600055500025e-05, + "loss": 1.0085, + "step": 769 + }, + { + "epoch": 1.41, + "grad_norm": 0.3707139982828035, + "learning_rate": 5.102367308798658e-05, + "loss": 1.0746, + "step": 770 + }, + { + "epoch": 1.41, + "grad_norm": 0.3378537316757011, + "learning_rate": 5.094169604432225e-05, + "loss": 1.0482, + "step": 771 + }, + { + "epoch": 1.41, + "grad_norm": 0.4008417246255145, + "learning_rate": 5.085966929689601e-05, + "loss": 1.1065, + "step": 772 + }, + { + "epoch": 1.41, + "grad_norm": 0.3244385106988064, + "learning_rate": 5.077759321832271e-05, + "loss": 1.0827, + "step": 773 + }, + { + "epoch": 1.42, + "grad_norm": 0.37228575732812336, + "learning_rate": 5.0695468181441215e-05, + "loss": 1.1146, + "step": 774 + }, + { + "epoch": 1.42, + "grad_norm": 0.33761714797540276, + "learning_rate": 5.061329455931283e-05, + "loss": 1.092, + "step": 775 + }, + { + "epoch": 1.42, + "grad_norm": 0.3158158390913494, + "learning_rate": 5.053107272521955e-05, + "loss": 1.1058, + "step": 776 + }, + { + "epoch": 1.42, + "grad_norm": 0.3691501929738938, + "learning_rate": 5.044880305266239e-05, + "loss": 1.1599, + "step": 777 + }, + { + "epoch": 1.42, + "grad_norm": 0.33730914019805525, + "learning_rate": 5.0366485915359645e-05, + "loss": 1.0615, + "step": 778 + }, + { + "epoch": 1.42, + "grad_norm": 0.34970059240017, + "learning_rate": 5.0284121687245257e-05, + "loss": 1.1475, + "step": 779 + }, + { + "epoch": 1.43, + "grad_norm": 0.3374028029407197, + "learning_rate": 5.020171074246707e-05, + "loss": 1.0926, + "step": 780 + }, + { + "epoch": 1.43, + "grad_norm": 0.3350020681123992, + "learning_rate": 5.011925345538514e-05, + "loss": 1.1276, + "step": 781 + }, + { + "epoch": 1.43, + "grad_norm": 0.3224228965786606, + "learning_rate": 5.003675020057003e-05, + "loss": 1.0183, + "step": 782 + }, + { + "epoch": 1.43, + "grad_norm": 0.3357310714740298, + "learning_rate": 4.995420135280114e-05, + "loss": 1.1114, + "step": 783 + }, + { + "epoch": 1.43, + "grad_norm": 0.3590203255363759, + "learning_rate": 4.9871607287064966e-05, + "loss": 1.1504, + "step": 784 + }, + { + "epoch": 1.44, + "grad_norm": 0.33011195419611655, + "learning_rate": 4.9788968378553396e-05, + "loss": 1.0826, + "step": 785 + }, + { + "epoch": 1.44, + "grad_norm": 0.31088868195439445, + "learning_rate": 4.970628500266207e-05, + "loss": 1.0704, + "step": 786 + }, + { + "epoch": 1.44, + "grad_norm": 0.3144996103179409, + "learning_rate": 4.962355753498858e-05, + "loss": 1.1403, + "step": 787 + }, + { + "epoch": 1.44, + "grad_norm": 0.3147269555419068, + "learning_rate": 4.954078635133081e-05, + "loss": 1.0898, + "step": 788 + }, + { + "epoch": 1.44, + "grad_norm": 0.3280151747783868, + "learning_rate": 4.945797182768524e-05, + "loss": 1.1115, + "step": 789 + }, + { + "epoch": 1.44, + "grad_norm": 0.3551996569232493, + "learning_rate": 4.937511434024524e-05, + "loss": 1.1731, + "step": 790 + }, + { + "epoch": 1.45, + "grad_norm": 0.343863208057807, + "learning_rate": 4.9292214265399336e-05, + "loss": 1.0866, + "step": 791 + }, + { + "epoch": 1.45, + "grad_norm": 0.37316699385322466, + "learning_rate": 4.920927197972949e-05, + "loss": 1.1083, + "step": 792 + }, + { + "epoch": 1.45, + "grad_norm": 0.3635739774067832, + "learning_rate": 4.9126287860009453e-05, + "loss": 1.1393, + "step": 793 + }, + { + "epoch": 1.45, + "grad_norm": 0.3755910554972886, + "learning_rate": 4.9043262283202974e-05, + "loss": 1.1624, + "step": 794 + }, + { + "epoch": 1.45, + "grad_norm": 0.3635899120146823, + "learning_rate": 4.8960195626462145e-05, + "loss": 1.2095, + "step": 795 + }, + { + "epoch": 1.46, + "grad_norm": 0.3642202684342816, + "learning_rate": 4.8877088267125664e-05, + "loss": 1.1099, + "step": 796 + }, + { + "epoch": 1.46, + "grad_norm": 0.3339946548799316, + "learning_rate": 4.879394058271712e-05, + "loss": 1.1157, + "step": 797 + }, + { + "epoch": 1.46, + "grad_norm": 0.3457189703100475, + "learning_rate": 4.871075295094329e-05, + "loss": 1.129, + "step": 798 + }, + { + "epoch": 1.46, + "grad_norm": 0.3550931839691424, + "learning_rate": 4.862752574969241e-05, + "loss": 1.076, + "step": 799 + }, + { + "epoch": 1.46, + "grad_norm": 0.36139108917966734, + "learning_rate": 4.8544259357032475e-05, + "loss": 1.1577, + "step": 800 + }, + { + "epoch": 1.0, + "grad_norm": 0.39569703665247874, + "learning_rate": 4.8460954151209486e-05, + "loss": 1.0543, + "step": 801 + }, + { + "epoch": 1.0, + "grad_norm": 0.3879033670170866, + "learning_rate": 4.837761051064579e-05, + "loss": 1.0688, + "step": 802 + }, + { + "epoch": 1.01, + "grad_norm": 0.3796846713967255, + "learning_rate": 4.8294228813938285e-05, + "loss": 0.9911, + "step": 803 + }, + { + "epoch": 1.01, + "grad_norm": 0.4007831430409375, + "learning_rate": 4.8210809439856804e-05, + "loss": 1.0126, + "step": 804 + }, + { + "epoch": 1.01, + "grad_norm": 0.37588078665500885, + "learning_rate": 4.8127352767342276e-05, + "loss": 0.9302, + "step": 805 + }, + { + "epoch": 1.01, + "grad_norm": 0.4078509175013281, + "learning_rate": 4.8043859175505095e-05, + "loss": 0.9982, + "step": 806 + }, + { + "epoch": 1.01, + "grad_norm": 0.379096046185539, + "learning_rate": 4.7960329043623344e-05, + "loss": 1.0035, + "step": 807 + }, + { + "epoch": 1.01, + "grad_norm": 0.3813938568133554, + "learning_rate": 4.787676275114111e-05, + "loss": 0.9579, + "step": 808 + }, + { + "epoch": 1.02, + "grad_norm": 0.3686863564511168, + "learning_rate": 4.779316067766673e-05, + "loss": 1.0105, + "step": 809 + }, + { + "epoch": 1.02, + "grad_norm": 0.4263940878847523, + "learning_rate": 4.770952320297109e-05, + "loss": 1.0677, + "step": 810 + }, + { + "epoch": 1.02, + "grad_norm": 0.37178778374665006, + "learning_rate": 4.7625850706985886e-05, + "loss": 1.0019, + "step": 811 + }, + { + "epoch": 1.02, + "grad_norm": 0.36803355429187945, + "learning_rate": 4.7542143569801894e-05, + "loss": 0.9937, + "step": 812 + }, + { + "epoch": 1.02, + "grad_norm": 0.3897072472941179, + "learning_rate": 4.745840217166725e-05, + "loss": 1.0877, + "step": 813 + }, + { + "epoch": 1.03, + "grad_norm": 0.35571833841716255, + "learning_rate": 4.737462689298577e-05, + "loss": 1.0015, + "step": 814 + }, + { + "epoch": 1.03, + "grad_norm": 0.38930229991094323, + "learning_rate": 4.7290818114315086e-05, + "loss": 1.028, + "step": 815 + }, + { + "epoch": 1.03, + "grad_norm": 0.411005007105147, + "learning_rate": 4.72069762163651e-05, + "loss": 1.0068, + "step": 816 + }, + { + "epoch": 1.03, + "grad_norm": 0.3980240190337736, + "learning_rate": 4.7123101579996106e-05, + "loss": 0.9919, + "step": 817 + }, + { + "epoch": 1.03, + "grad_norm": 0.36369517703115467, + "learning_rate": 4.7039194586217136e-05, + "loss": 0.967, + "step": 818 + }, + { + "epoch": 1.03, + "grad_norm": 0.38591148840458894, + "learning_rate": 4.695525561618418e-05, + "loss": 0.9743, + "step": 819 + }, + { + "epoch": 1.04, + "grad_norm": 0.45873135108949337, + "learning_rate": 4.687128505119853e-05, + "loss": 1.0516, + "step": 820 + }, + { + "epoch": 1.04, + "grad_norm": 0.3866330351411308, + "learning_rate": 4.6787283272704966e-05, + "loss": 0.9939, + "step": 821 + }, + { + "epoch": 1.04, + "grad_norm": 0.4620340173291326, + "learning_rate": 4.670325066229009e-05, + "loss": 1.0526, + "step": 822 + }, + { + "epoch": 1.04, + "grad_norm": 0.38877454299870284, + "learning_rate": 4.661918760168052e-05, + "loss": 0.9904, + "step": 823 + }, + { + "epoch": 1.04, + "grad_norm": 0.3880489386116793, + "learning_rate": 4.653509447274121e-05, + "loss": 0.9623, + "step": 824 + }, + { + "epoch": 1.05, + "grad_norm": 0.3827392356186151, + "learning_rate": 4.6450971657473743e-05, + "loss": 1.0772, + "step": 825 + }, + { + "epoch": 1.05, + "grad_norm": 0.4132814641854327, + "learning_rate": 4.63668195380145e-05, + "loss": 1.0533, + "step": 826 + }, + { + "epoch": 1.05, + "grad_norm": 0.3703610182402835, + "learning_rate": 4.628263849663301e-05, + "loss": 0.9336, + "step": 827 + }, + { + "epoch": 1.05, + "grad_norm": 0.4152053683299823, + "learning_rate": 4.619842891573016e-05, + "loss": 0.9801, + "step": 828 + }, + { + "epoch": 1.05, + "grad_norm": 0.41791059043554274, + "learning_rate": 4.6114191177836514e-05, + "loss": 1.0617, + "step": 829 + }, + { + "epoch": 1.05, + "grad_norm": 0.46363896517299136, + "learning_rate": 4.6029925665610524e-05, + "loss": 0.9687, + "step": 830 + }, + { + "epoch": 1.06, + "grad_norm": 0.41141959057512445, + "learning_rate": 4.59456327618368e-05, + "loss": 1.0965, + "step": 831 + }, + { + "epoch": 1.06, + "grad_norm": 0.3789192764519836, + "learning_rate": 4.5861312849424386e-05, + "loss": 0.9793, + "step": 832 + }, + { + "epoch": 1.06, + "grad_norm": 0.4047291581107866, + "learning_rate": 4.5776966311405035e-05, + "loss": 1.0342, + "step": 833 + }, + { + "epoch": 1.06, + "grad_norm": 0.4425157400959256, + "learning_rate": 4.5692593530931416e-05, + "loss": 1.0892, + "step": 834 + }, + { + "epoch": 1.06, + "grad_norm": 0.3707332144806616, + "learning_rate": 4.560819489127545e-05, + "loss": 0.9815, + "step": 835 + }, + { + "epoch": 1.07, + "grad_norm": 0.3897444102572823, + "learning_rate": 4.552377077582646e-05, + "loss": 0.9884, + "step": 836 + }, + { + "epoch": 1.07, + "grad_norm": 0.42725787957019346, + "learning_rate": 4.543932156808959e-05, + "loss": 0.9972, + "step": 837 + }, + { + "epoch": 1.07, + "grad_norm": 0.40615269781820007, + "learning_rate": 4.535484765168386e-05, + "loss": 0.9529, + "step": 838 + }, + { + "epoch": 1.07, + "grad_norm": 0.3505829736050887, + "learning_rate": 4.527034941034063e-05, + "loss": 0.9492, + "step": 839 + }, + { + "epoch": 1.07, + "grad_norm": 0.36688064686440497, + "learning_rate": 4.51858272279017e-05, + "loss": 0.9592, + "step": 840 + }, + { + "epoch": 1.07, + "grad_norm": 0.4043468777955929, + "learning_rate": 4.5101281488317634e-05, + "loss": 1.048, + "step": 841 + }, + { + "epoch": 1.08, + "grad_norm": 0.3811489793242706, + "learning_rate": 4.501671257564602e-05, + "loss": 1.0138, + "step": 842 + }, + { + "epoch": 1.08, + "grad_norm": 0.39813004142325986, + "learning_rate": 4.49321208740497e-05, + "loss": 1.071, + "step": 843 + }, + { + "epoch": 1.08, + "grad_norm": 0.3809751022095503, + "learning_rate": 4.484750676779504e-05, + "loss": 1.0351, + "step": 844 + }, + { + "epoch": 1.08, + "grad_norm": 0.384312178013823, + "learning_rate": 4.4762870641250185e-05, + "loss": 0.9737, + "step": 845 + }, + { + "epoch": 1.08, + "grad_norm": 0.40769404907923557, + "learning_rate": 4.467821287888331e-05, + "loss": 0.9659, + "step": 846 + }, + { + "epoch": 1.09, + "grad_norm": 0.39594136851937817, + "learning_rate": 4.459353386526086e-05, + "loss": 0.9405, + "step": 847 + }, + { + "epoch": 1.09, + "grad_norm": 0.37180161011562185, + "learning_rate": 4.450883398504584e-05, + "loss": 1.0732, + "step": 848 + }, + { + "epoch": 1.09, + "grad_norm": 0.3772603623154663, + "learning_rate": 4.442411362299602e-05, + "loss": 0.9646, + "step": 849 + }, + { + "epoch": 1.09, + "grad_norm": 0.4346142368506476, + "learning_rate": 4.433937316396224e-05, + "loss": 0.9572, + "step": 850 + }, + { + "epoch": 1.09, + "grad_norm": 0.3997258084612474, + "learning_rate": 4.425461299288659e-05, + "loss": 0.9492, + "step": 851 + }, + { + "epoch": 1.1, + "grad_norm": 0.41245476865247155, + "learning_rate": 4.416983349480073e-05, + "loss": 0.8732, + "step": 852 + }, + { + "epoch": 1.1, + "grad_norm": 0.6761499297939195, + "learning_rate": 4.408503505482412e-05, + "loss": 1.0425, + "step": 853 + }, + { + "epoch": 1.1, + "grad_norm": 0.40340979486858985, + "learning_rate": 4.400021805816225e-05, + "loss": 0.9596, + "step": 854 + }, + { + "epoch": 1.1, + "grad_norm": 0.43290732392699666, + "learning_rate": 4.391538289010493e-05, + "loss": 1.0123, + "step": 855 + }, + { + "epoch": 1.1, + "grad_norm": 0.36878054442190156, + "learning_rate": 4.383052993602448e-05, + "loss": 0.9448, + "step": 856 + }, + { + "epoch": 1.1, + "grad_norm": 0.7146145128961262, + "learning_rate": 4.374565958137404e-05, + "loss": 1.0342, + "step": 857 + }, + { + "epoch": 1.11, + "grad_norm": 0.44429357586145607, + "learning_rate": 4.3660772211685775e-05, + "loss": 1.0436, + "step": 858 + }, + { + "epoch": 1.11, + "grad_norm": 0.4565751973640598, + "learning_rate": 4.357586821256918e-05, + "loss": 1.0311, + "step": 859 + }, + { + "epoch": 1.11, + "grad_norm": 0.3919991236654277, + "learning_rate": 4.349094796970925e-05, + "loss": 1.1401, + "step": 860 + }, + { + "epoch": 1.11, + "grad_norm": 0.4347441949284011, + "learning_rate": 4.3406011868864795e-05, + "loss": 1.0252, + "step": 861 + }, + { + "epoch": 1.11, + "grad_norm": 0.38339976027415407, + "learning_rate": 4.3321060295866635e-05, + "loss": 1.0536, + "step": 862 + }, + { + "epoch": 1.12, + "grad_norm": 0.37688790408195166, + "learning_rate": 4.32360936366159e-05, + "loss": 1.012, + "step": 863 + }, + { + "epoch": 1.12, + "grad_norm": 0.4317538207582504, + "learning_rate": 4.315111227708224e-05, + "loss": 1.0505, + "step": 864 + }, + { + "epoch": 1.12, + "grad_norm": 0.4145324872228796, + "learning_rate": 4.306611660330208e-05, + "loss": 1.0496, + "step": 865 + }, + { + "epoch": 1.12, + "grad_norm": 0.416535227064448, + "learning_rate": 4.298110700137687e-05, + "loss": 0.9628, + "step": 866 + }, + { + "epoch": 1.12, + "grad_norm": 0.46564356187492717, + "learning_rate": 4.2896083857471345e-05, + "loss": 1.0016, + "step": 867 + }, + { + "epoch": 1.12, + "grad_norm": 0.4228980941889828, + "learning_rate": 4.281104755781172e-05, + "loss": 1.0904, + "step": 868 + }, + { + "epoch": 1.13, + "grad_norm": 0.4267821214430208, + "learning_rate": 4.272599848868402e-05, + "loss": 1.0544, + "step": 869 + }, + { + "epoch": 1.13, + "grad_norm": 0.45763332095792075, + "learning_rate": 4.264093703643223e-05, + "loss": 1.0686, + "step": 870 + }, + { + "epoch": 1.13, + "grad_norm": 0.4347555516548761, + "learning_rate": 4.255586358745662e-05, + "loss": 1.0264, + "step": 871 + }, + { + "epoch": 1.13, + "grad_norm": 0.3817726381103066, + "learning_rate": 4.247077852821194e-05, + "loss": 1.0045, + "step": 872 + }, + { + "epoch": 1.13, + "grad_norm": 0.3882808845457995, + "learning_rate": 4.2385682245205685e-05, + "loss": 1.0193, + "step": 873 + }, + { + "epoch": 1.14, + "grad_norm": 0.39410930252966775, + "learning_rate": 4.230057512499634e-05, + "loss": 0.9832, + "step": 874 + }, + { + "epoch": 1.14, + "grad_norm": 0.4373094593907156, + "learning_rate": 4.221545755419159e-05, + "loss": 1.0343, + "step": 875 + }, + { + "epoch": 1.14, + "grad_norm": 0.4462843721698891, + "learning_rate": 4.2130329919446646e-05, + "loss": 1.0324, + "step": 876 + }, + { + "epoch": 1.14, + "grad_norm": 0.4747274247448112, + "learning_rate": 4.20451926074624e-05, + "loss": 0.9903, + "step": 877 + }, + { + "epoch": 1.14, + "grad_norm": 0.4157472897596409, + "learning_rate": 4.196004600498369e-05, + "loss": 0.9266, + "step": 878 + }, + { + "epoch": 1.14, + "grad_norm": 0.41625958088960685, + "learning_rate": 4.1874890498797605e-05, + "loss": 0.9658, + "step": 879 + }, + { + "epoch": 1.15, + "grad_norm": 0.44784944130574333, + "learning_rate": 4.178972647573163e-05, + "loss": 0.9671, + "step": 880 + }, + { + "epoch": 1.15, + "grad_norm": 0.4116839177956385, + "learning_rate": 4.1704554322651975e-05, + "loss": 0.9591, + "step": 881 + }, + { + "epoch": 1.15, + "grad_norm": 0.4025569857639452, + "learning_rate": 4.161937442646176e-05, + "loss": 1.0072, + "step": 882 + }, + { + "epoch": 1.15, + "grad_norm": 0.41518478124763597, + "learning_rate": 4.1534187174099285e-05, + "loss": 1.0275, + "step": 883 + }, + { + "epoch": 1.15, + "grad_norm": 0.3987815564664466, + "learning_rate": 4.1448992952536275e-05, + "loss": 1.0039, + "step": 884 + }, + { + "epoch": 1.16, + "grad_norm": 0.4270378155679982, + "learning_rate": 4.136379214877609e-05, + "loss": 1.0369, + "step": 885 + }, + { + "epoch": 1.16, + "grad_norm": 0.42144733922972777, + "learning_rate": 4.127858514985203e-05, + "loss": 1.0269, + "step": 886 + }, + { + "epoch": 1.16, + "grad_norm": 0.4198664438272548, + "learning_rate": 4.1193372342825494e-05, + "loss": 1.0427, + "step": 887 + }, + { + "epoch": 1.16, + "grad_norm": 0.3985048256281719, + "learning_rate": 4.1108154114784275e-05, + "loss": 1.0702, + "step": 888 + }, + { + "epoch": 1.16, + "grad_norm": 0.605520808292362, + "learning_rate": 4.102293085284083e-05, + "loss": 0.9749, + "step": 889 + }, + { + "epoch": 1.16, + "grad_norm": 0.4150515863924052, + "learning_rate": 4.0937702944130426e-05, + "loss": 1.0231, + "step": 890 + }, + { + "epoch": 1.17, + "grad_norm": 0.3935997576565283, + "learning_rate": 4.085247077580948e-05, + "loss": 1.0014, + "step": 891 + }, + { + "epoch": 1.17, + "grad_norm": 0.399446131403209, + "learning_rate": 4.076723473505374e-05, + "loss": 0.9602, + "step": 892 + }, + { + "epoch": 1.17, + "grad_norm": 0.4406024397129952, + "learning_rate": 4.068199520905655e-05, + "loss": 1.0425, + "step": 893 + }, + { + "epoch": 1.17, + "grad_norm": 0.4036917571496492, + "learning_rate": 4.059675258502709e-05, + "loss": 0.973, + "step": 894 + }, + { + "epoch": 1.17, + "grad_norm": 0.4057196459433299, + "learning_rate": 4.05115072501886e-05, + "loss": 0.9997, + "step": 895 + }, + { + "epoch": 1.18, + "grad_norm": 0.4374124954708759, + "learning_rate": 4.0426259591776645e-05, + "loss": 0.9826, + "step": 896 + }, + { + "epoch": 1.18, + "grad_norm": 0.4545699371285546, + "learning_rate": 4.0341009997037356e-05, + "loss": 1.0554, + "step": 897 + }, + { + "epoch": 1.18, + "grad_norm": 0.4251917031237376, + "learning_rate": 4.025575885322563e-05, + "loss": 1.0217, + "step": 898 + }, + { + "epoch": 1.18, + "grad_norm": 0.3857651901893941, + "learning_rate": 4.0170506547603427e-05, + "loss": 1.0317, + "step": 899 + }, + { + "epoch": 1.18, + "grad_norm": 0.46323573798490897, + "learning_rate": 4.008525346743797e-05, + "loss": 1.0398, + "step": 900 + }, + { + "epoch": 1.18, + "grad_norm": 0.4011541121460918, + "learning_rate": 4e-05, + "loss": 1.0706, + "step": 901 + }, + { + "epoch": 1.19, + "grad_norm": 0.46493281221028004, + "learning_rate": 3.991474653256204e-05, + "loss": 1.0525, + "step": 902 + }, + { + "epoch": 1.19, + "grad_norm": 0.41683080924539023, + "learning_rate": 3.982949345239658e-05, + "loss": 1.0905, + "step": 903 + }, + { + "epoch": 1.19, + "grad_norm": 0.4750350025014512, + "learning_rate": 3.974424114677437e-05, + "loss": 1.049, + "step": 904 + }, + { + "epoch": 1.19, + "grad_norm": 0.3867445073614702, + "learning_rate": 3.965899000296266e-05, + "loss": 0.9624, + "step": 905 + }, + { + "epoch": 1.19, + "grad_norm": 0.378387661131469, + "learning_rate": 3.957374040822335e-05, + "loss": 1.0223, + "step": 906 + }, + { + "epoch": 1.2, + "grad_norm": 0.3905996390559077, + "learning_rate": 3.948849274981141e-05, + "loss": 1.0315, + "step": 907 + }, + { + "epoch": 1.2, + "grad_norm": 0.4139717689498189, + "learning_rate": 3.940324741497291e-05, + "loss": 0.9297, + "step": 908 + }, + { + "epoch": 1.2, + "grad_norm": 0.39086355684921514, + "learning_rate": 3.9318004790943465e-05, + "loss": 0.9684, + "step": 909 + }, + { + "epoch": 1.2, + "grad_norm": 0.4334915643736419, + "learning_rate": 3.923276526494627e-05, + "loss": 0.996, + "step": 910 + }, + { + "epoch": 1.2, + "grad_norm": 0.40782018986229496, + "learning_rate": 3.9147529224190536e-05, + "loss": 1.0875, + "step": 911 + }, + { + "epoch": 1.2, + "grad_norm": 0.43578702386625723, + "learning_rate": 3.906229705586959e-05, + "loss": 1.1214, + "step": 912 + }, + { + "epoch": 1.21, + "grad_norm": 0.414945683409524, + "learning_rate": 3.89770691471592e-05, + "loss": 1.1037, + "step": 913 + }, + { + "epoch": 1.21, + "grad_norm": 0.40665801579679106, + "learning_rate": 3.889184588521573e-05, + "loss": 0.9743, + "step": 914 + }, + { + "epoch": 1.21, + "grad_norm": 0.4064250611574517, + "learning_rate": 3.880662765717453e-05, + "loss": 0.8814, + "step": 915 + }, + { + "epoch": 1.21, + "grad_norm": 0.48023046298843347, + "learning_rate": 3.8721414850147985e-05, + "loss": 0.9663, + "step": 916 + }, + { + "epoch": 1.21, + "grad_norm": 0.42358024833566227, + "learning_rate": 3.8636207851223924e-05, + "loss": 1.0491, + "step": 917 + }, + { + "epoch": 1.22, + "grad_norm": 0.41522494786195835, + "learning_rate": 3.855100704746374e-05, + "loss": 1.033, + "step": 918 + }, + { + "epoch": 1.22, + "grad_norm": 0.40890517696706496, + "learning_rate": 3.8465812825900715e-05, + "loss": 1.0369, + "step": 919 + }, + { + "epoch": 1.22, + "grad_norm": 0.4325851866408538, + "learning_rate": 3.838062557353825e-05, + "loss": 0.9362, + "step": 920 + }, + { + "epoch": 1.22, + "grad_norm": 0.4185860919050069, + "learning_rate": 3.8295445677348025e-05, + "loss": 1.026, + "step": 921 + }, + { + "epoch": 1.22, + "grad_norm": 0.3975762375934804, + "learning_rate": 3.8210273524268375e-05, + "loss": 1.0412, + "step": 922 + }, + { + "epoch": 1.22, + "grad_norm": 0.41725298241987474, + "learning_rate": 3.8125109501202395e-05, + "loss": 1.0004, + "step": 923 + }, + { + "epoch": 1.23, + "grad_norm": 0.455183913149126, + "learning_rate": 3.803995399501632e-05, + "loss": 1.0594, + "step": 924 + }, + { + "epoch": 1.23, + "grad_norm": 0.3993993856483797, + "learning_rate": 3.795480739253761e-05, + "loss": 0.9761, + "step": 925 + }, + { + "epoch": 1.23, + "grad_norm": 0.41638796815161494, + "learning_rate": 3.786967008055337e-05, + "loss": 1.0369, + "step": 926 + }, + { + "epoch": 1.23, + "grad_norm": 0.40015112695810534, + "learning_rate": 3.7784542445808414e-05, + "loss": 1.0271, + "step": 927 + }, + { + "epoch": 1.23, + "grad_norm": 0.3995749494729548, + "learning_rate": 3.769942487500368e-05, + "loss": 1.0613, + "step": 928 + }, + { + "epoch": 1.24, + "grad_norm": 0.4073556267037492, + "learning_rate": 3.761431775479432e-05, + "loss": 1.0528, + "step": 929 + }, + { + "epoch": 1.24, + "grad_norm": 0.44218148822636044, + "learning_rate": 3.752922147178807e-05, + "loss": 1.0742, + "step": 930 + }, + { + "epoch": 1.24, + "grad_norm": 0.4435063485893757, + "learning_rate": 3.744413641254339e-05, + "loss": 1.0825, + "step": 931 + }, + { + "epoch": 1.24, + "grad_norm": 0.46841574994107515, + "learning_rate": 3.735906296356778e-05, + "loss": 1.0471, + "step": 932 + }, + { + "epoch": 1.24, + "grad_norm": 0.40093716627657294, + "learning_rate": 3.727400151131599e-05, + "loss": 1.0474, + "step": 933 + }, + { + "epoch": 1.25, + "grad_norm": 0.3866415067997244, + "learning_rate": 3.71889524421883e-05, + "loss": 1.0209, + "step": 934 + }, + { + "epoch": 1.25, + "grad_norm": 0.4881546110706673, + "learning_rate": 3.710391614252867e-05, + "loss": 1.0768, + "step": 935 + }, + { + "epoch": 1.25, + "grad_norm": 0.4133084639324523, + "learning_rate": 3.701889299862314e-05, + "loss": 1.0423, + "step": 936 + }, + { + "epoch": 1.25, + "grad_norm": 0.40523563084001196, + "learning_rate": 3.6933883396697936e-05, + "loss": 1.005, + "step": 937 + }, + { + "epoch": 1.25, + "grad_norm": 0.38757352418642405, + "learning_rate": 3.684888772291777e-05, + "loss": 0.9659, + "step": 938 + }, + { + "epoch": 1.25, + "grad_norm": 0.421394551890689, + "learning_rate": 3.676390636338411e-05, + "loss": 1.0454, + "step": 939 + }, + { + "epoch": 1.26, + "grad_norm": 0.45693070958342186, + "learning_rate": 3.667893970413337e-05, + "loss": 1.1459, + "step": 940 + }, + { + "epoch": 1.26, + "grad_norm": 0.4172025376377795, + "learning_rate": 3.659398813113522e-05, + "loss": 0.9954, + "step": 941 + }, + { + "epoch": 1.26, + "grad_norm": 0.3871624019510191, + "learning_rate": 3.650905203029075e-05, + "loss": 1.0441, + "step": 942 + }, + { + "epoch": 1.26, + "grad_norm": 0.38541342610032325, + "learning_rate": 3.642413178743083e-05, + "loss": 0.9465, + "step": 943 + }, + { + "epoch": 1.26, + "grad_norm": 0.4208031670525743, + "learning_rate": 3.633922778831423e-05, + "loss": 1.0367, + "step": 944 + }, + { + "epoch": 1.27, + "grad_norm": 0.41867209013040035, + "learning_rate": 3.6254340418625975e-05, + "loss": 1.0868, + "step": 945 + }, + { + "epoch": 1.27, + "grad_norm": 0.431758149074127, + "learning_rate": 3.6169470063975536e-05, + "loss": 1.0689, + "step": 946 + }, + { + "epoch": 1.27, + "grad_norm": 0.4988803338819952, + "learning_rate": 3.608461710989509e-05, + "loss": 1.0879, + "step": 947 + }, + { + "epoch": 1.27, + "grad_norm": 0.4094858411191625, + "learning_rate": 3.5999781941837755e-05, + "loss": 1.0332, + "step": 948 + }, + { + "epoch": 1.27, + "grad_norm": 0.3831847195845155, + "learning_rate": 3.591496494517589e-05, + "loss": 0.9751, + "step": 949 + }, + { + "epoch": 1.27, + "grad_norm": 0.40535692821947267, + "learning_rate": 3.5830166505199284e-05, + "loss": 1.0594, + "step": 950 + }, + { + "epoch": 1.28, + "grad_norm": 0.4875663789389966, + "learning_rate": 3.574538700711343e-05, + "loss": 0.9749, + "step": 951 + }, + { + "epoch": 1.28, + "grad_norm": 0.5155923998285772, + "learning_rate": 3.566062683603778e-05, + "loss": 0.9999, + "step": 952 + }, + { + "epoch": 1.28, + "grad_norm": 0.5280285947816189, + "learning_rate": 3.557588637700399e-05, + "loss": 1.1061, + "step": 953 + }, + { + "epoch": 1.28, + "grad_norm": 0.46573407357796753, + "learning_rate": 3.5491166014954174e-05, + "loss": 1.102, + "step": 954 + }, + { + "epoch": 1.28, + "grad_norm": 0.4122542582865379, + "learning_rate": 3.540646613473915e-05, + "loss": 1.0469, + "step": 955 + }, + { + "epoch": 1.29, + "grad_norm": 0.41414476980823367, + "learning_rate": 3.53217871211167e-05, + "loss": 0.9973, + "step": 956 + }, + { + "epoch": 1.29, + "grad_norm": 0.4030707611608045, + "learning_rate": 3.523712935874983e-05, + "loss": 0.9796, + "step": 957 + }, + { + "epoch": 1.29, + "grad_norm": 0.4235313349747291, + "learning_rate": 3.5152493232204975e-05, + "loss": 1.0601, + "step": 958 + }, + { + "epoch": 1.29, + "grad_norm": 0.4165235178302652, + "learning_rate": 3.5067879125950316e-05, + "loss": 1.0358, + "step": 959 + }, + { + "epoch": 1.29, + "grad_norm": 0.44083984701952955, + "learning_rate": 3.4983287424354e-05, + "loss": 1.0957, + "step": 960 + }, + { + "epoch": 1.29, + "grad_norm": 0.3781161039063518, + "learning_rate": 3.489871851168238e-05, + "loss": 0.9838, + "step": 961 + }, + { + "epoch": 1.3, + "grad_norm": 0.4095747724038915, + "learning_rate": 3.4814172772098314e-05, + "loss": 1.014, + "step": 962 + }, + { + "epoch": 1.3, + "grad_norm": 0.42197119558898466, + "learning_rate": 3.472965058965938e-05, + "loss": 1.0096, + "step": 963 + }, + { + "epoch": 1.3, + "grad_norm": 0.4339963388152155, + "learning_rate": 3.464515234831615e-05, + "loss": 1.0158, + "step": 964 + }, + { + "epoch": 1.3, + "grad_norm": 0.4284638765548976, + "learning_rate": 3.4560678431910424e-05, + "loss": 1.1047, + "step": 965 + }, + { + "epoch": 1.3, + "grad_norm": 0.3935144535755794, + "learning_rate": 3.447622922417355e-05, + "loss": 0.9925, + "step": 966 + }, + { + "epoch": 1.31, + "grad_norm": 0.45884343961025, + "learning_rate": 3.439180510872457e-05, + "loss": 1.0583, + "step": 967 + }, + { + "epoch": 1.31, + "grad_norm": 0.42439320759788374, + "learning_rate": 3.4307406469068604e-05, + "loss": 0.9305, + "step": 968 + }, + { + "epoch": 1.31, + "grad_norm": 0.45770082390324845, + "learning_rate": 3.4223033688594985e-05, + "loss": 1.054, + "step": 969 + }, + { + "epoch": 1.31, + "grad_norm": 0.4284786643981094, + "learning_rate": 3.4138687150575634e-05, + "loss": 0.9409, + "step": 970 + }, + { + "epoch": 1.31, + "grad_norm": 0.41356124058383237, + "learning_rate": 3.4054367238163215e-05, + "loss": 1.0739, + "step": 971 + }, + { + "epoch": 1.31, + "grad_norm": 0.4255832249412624, + "learning_rate": 3.3970074334389496e-05, + "loss": 1.0764, + "step": 972 + }, + { + "epoch": 1.32, + "grad_norm": 0.4337695536142702, + "learning_rate": 3.388580882216349e-05, + "loss": 1.0195, + "step": 973 + }, + { + "epoch": 1.32, + "grad_norm": 0.41363495650922455, + "learning_rate": 3.380157108426985e-05, + "loss": 1.0615, + "step": 974 + }, + { + "epoch": 1.32, + "grad_norm": 0.3950691247686479, + "learning_rate": 3.371736150336701e-05, + "loss": 1.0283, + "step": 975 + }, + { + "epoch": 1.32, + "grad_norm": 0.4042823691555822, + "learning_rate": 3.3633180461985505e-05, + "loss": 1.0309, + "step": 976 + }, + { + "epoch": 1.32, + "grad_norm": 0.3921158850479399, + "learning_rate": 3.354902834252627e-05, + "loss": 1.068, + "step": 977 + }, + { + "epoch": 1.33, + "grad_norm": 0.38349545732725654, + "learning_rate": 3.346490552725879e-05, + "loss": 1.0886, + "step": 978 + }, + { + "epoch": 1.33, + "grad_norm": 0.38689221457248724, + "learning_rate": 3.33808123983195e-05, + "loss": 0.987, + "step": 979 + }, + { + "epoch": 1.33, + "grad_norm": 0.38660550867425647, + "learning_rate": 3.329674933770992e-05, + "loss": 1.069, + "step": 980 + }, + { + "epoch": 1.33, + "grad_norm": 0.3917593746353493, + "learning_rate": 3.321271672729504e-05, + "loss": 0.9858, + "step": 981 + }, + { + "epoch": 1.33, + "grad_norm": 0.4292314072827653, + "learning_rate": 3.3128714948801474e-05, + "loss": 1.0477, + "step": 982 + }, + { + "epoch": 1.33, + "grad_norm": 0.479414638418211, + "learning_rate": 3.3044744383815835e-05, + "loss": 1.0763, + "step": 983 + }, + { + "epoch": 1.34, + "grad_norm": 0.380831894995463, + "learning_rate": 3.2960805413782884e-05, + "loss": 1.0393, + "step": 984 + }, + { + "epoch": 1.34, + "grad_norm": 0.42402274703362114, + "learning_rate": 3.2876898420003914e-05, + "loss": 1.0837, + "step": 985 + }, + { + "epoch": 1.34, + "grad_norm": 0.4571447203722258, + "learning_rate": 3.279302378363491e-05, + "loss": 1.0594, + "step": 986 + }, + { + "epoch": 1.34, + "grad_norm": 0.3776673281658531, + "learning_rate": 3.270918188568493e-05, + "loss": 1.0121, + "step": 987 + }, + { + "epoch": 1.34, + "grad_norm": 0.4367173448132159, + "learning_rate": 3.262537310701425e-05, + "loss": 0.9612, + "step": 988 + }, + { + "epoch": 1.35, + "grad_norm": 0.43679765208840926, + "learning_rate": 3.254159782833276e-05, + "loss": 1.0565, + "step": 989 + }, + { + "epoch": 1.35, + "grad_norm": 0.4018151260013493, + "learning_rate": 3.2457856430198126e-05, + "loss": 0.9975, + "step": 990 + }, + { + "epoch": 1.35, + "grad_norm": 0.40461959940721076, + "learning_rate": 3.237414929301412e-05, + "loss": 1.0255, + "step": 991 + }, + { + "epoch": 1.35, + "grad_norm": 0.41342378541540653, + "learning_rate": 3.2290476797028926e-05, + "loss": 1.024, + "step": 992 + }, + { + "epoch": 1.35, + "grad_norm": 0.3926173909201105, + "learning_rate": 3.220683932233328e-05, + "loss": 1.0877, + "step": 993 + }, + { + "epoch": 1.35, + "grad_norm": 0.3835623199834992, + "learning_rate": 3.21232372488589e-05, + "loss": 1.0992, + "step": 994 + }, + { + "epoch": 1.36, + "grad_norm": 0.39901809497083496, + "learning_rate": 3.2039670956376656e-05, + "loss": 1.0723, + "step": 995 + }, + { + "epoch": 1.36, + "grad_norm": 0.3979604537466272, + "learning_rate": 3.195614082449492e-05, + "loss": 1.0201, + "step": 996 + }, + { + "epoch": 1.36, + "grad_norm": 0.4057122427176845, + "learning_rate": 3.1872647232657723e-05, + "loss": 1.0885, + "step": 997 + }, + { + "epoch": 1.36, + "grad_norm": 0.39747060350754754, + "learning_rate": 3.17891905601432e-05, + "loss": 1.0544, + "step": 998 + }, + { + "epoch": 1.36, + "grad_norm": 0.4397658078291558, + "learning_rate": 3.1705771186061715e-05, + "loss": 1.0998, + "step": 999 + }, + { + "epoch": 1.37, + "grad_norm": 0.37373547663810053, + "learning_rate": 3.162238948935423e-05, + "loss": 1.0465, + "step": 1000 + }, + { + "epoch": 1.0, + "grad_norm": 0.4042576001255747, + "learning_rate": 3.153904584879052e-05, + "loss": 0.9206, + "step": 1001 + }, + { + "epoch": 1.0, + "grad_norm": 0.4042994886900337, + "learning_rate": 3.1455740642967545e-05, + "loss": 0.975, + "step": 1002 + }, + { + "epoch": 1.01, + "grad_norm": 0.4359721725421234, + "learning_rate": 3.1372474250307594e-05, + "loss": 0.9163, + "step": 1003 + }, + { + "epoch": 1.01, + "grad_norm": 0.4886423524029179, + "learning_rate": 3.128924704905673e-05, + "loss": 0.9956, + "step": 1004 + }, + { + "epoch": 1.01, + "grad_norm": 0.48669990170138744, + "learning_rate": 3.1206059417282894e-05, + "loss": 0.9874, + "step": 1005 + }, + { + "epoch": 1.01, + "grad_norm": 0.41954255928633066, + "learning_rate": 3.1122911732874356e-05, + "loss": 0.8986, + "step": 1006 + }, + { + "epoch": 1.01, + "grad_norm": 0.43363878644039366, + "learning_rate": 3.103980437353787e-05, + "loss": 0.9268, + "step": 1007 + }, + { + "epoch": 1.01, + "grad_norm": 0.5199775120765874, + "learning_rate": 3.0956737716797047e-05, + "loss": 0.9341, + "step": 1008 + }, + { + "epoch": 1.02, + "grad_norm": 0.40735757951139595, + "learning_rate": 3.087371213999056e-05, + "loss": 0.9142, + "step": 1009 + }, + { + "epoch": 1.02, + "grad_norm": 0.44449027493884186, + "learning_rate": 3.079072802027051e-05, + "loss": 0.966, + "step": 1010 + }, + { + "epoch": 1.02, + "grad_norm": 0.46590494286419365, + "learning_rate": 3.070778573460068e-05, + "loss": 0.8768, + "step": 1011 + }, + { + "epoch": 1.02, + "grad_norm": 0.45161453051587425, + "learning_rate": 3.062488565975476e-05, + "loss": 0.9299, + "step": 1012 + }, + { + "epoch": 1.02, + "grad_norm": 0.5022364894382346, + "learning_rate": 3.054202817231477e-05, + "loss": 0.9352, + "step": 1013 + }, + { + "epoch": 1.03, + "grad_norm": 0.46443439138730447, + "learning_rate": 3.0459213648669195e-05, + "loss": 0.8913, + "step": 1014 + }, + { + "epoch": 1.03, + "grad_norm": 0.41932307219261455, + "learning_rate": 3.0376442465011436e-05, + "loss": 0.8968, + "step": 1015 + }, + { + "epoch": 1.03, + "grad_norm": 0.42445864358441704, + "learning_rate": 3.0293714997337927e-05, + "loss": 0.8449, + "step": 1016 + }, + { + "epoch": 1.03, + "grad_norm": 0.4489777773688699, + "learning_rate": 3.0211031621446607e-05, + "loss": 0.927, + "step": 1017 + }, + { + "epoch": 1.03, + "grad_norm": 0.45180577504235525, + "learning_rate": 3.0128392712935044e-05, + "loss": 0.8834, + "step": 1018 + }, + { + "epoch": 1.03, + "grad_norm": 0.44680469596106914, + "learning_rate": 3.0045798647198882e-05, + "loss": 1.0176, + "step": 1019 + }, + { + "epoch": 1.04, + "grad_norm": 0.45747851649657734, + "learning_rate": 2.9963249799429986e-05, + "loss": 0.9036, + "step": 1020 + }, + { + "epoch": 1.04, + "grad_norm": 0.5045904501932169, + "learning_rate": 2.988074654461489e-05, + "loss": 1.0475, + "step": 1021 + }, + { + "epoch": 1.04, + "grad_norm": 0.47086144833942983, + "learning_rate": 2.9798289257532946e-05, + "loss": 0.9596, + "step": 1022 + }, + { + "epoch": 1.04, + "grad_norm": 0.4406706196288816, + "learning_rate": 2.9715878312754767e-05, + "loss": 1.0054, + "step": 1023 + }, + { + "epoch": 1.04, + "grad_norm": 0.44584179061175105, + "learning_rate": 2.9633514084640365e-05, + "loss": 0.8981, + "step": 1024 + }, + { + "epoch": 1.05, + "grad_norm": 0.462343843042957, + "learning_rate": 2.955119694733763e-05, + "loss": 0.974, + "step": 1025 + }, + { + "epoch": 1.05, + "grad_norm": 0.46767265335377156, + "learning_rate": 2.946892727478045e-05, + "loss": 1.0063, + "step": 1026 + }, + { + "epoch": 1.05, + "grad_norm": 0.43250194002958803, + "learning_rate": 2.9386705440687168e-05, + "loss": 0.9332, + "step": 1027 + }, + { + "epoch": 1.05, + "grad_norm": 0.44391321845917453, + "learning_rate": 2.9304531818558795e-05, + "loss": 0.8937, + "step": 1028 + }, + { + "epoch": 1.05, + "grad_norm": 0.45616826414927975, + "learning_rate": 2.9222406781677294e-05, + "loss": 0.869, + "step": 1029 + }, + { + "epoch": 1.05, + "grad_norm": 0.5670635396983207, + "learning_rate": 2.9140330703103992e-05, + "loss": 0.9697, + "step": 1030 + }, + { + "epoch": 1.06, + "grad_norm": 0.4860829361401993, + "learning_rate": 2.905830395567776e-05, + "loss": 0.9677, + "step": 1031 + }, + { + "epoch": 1.06, + "grad_norm": 0.4484206829172443, + "learning_rate": 2.8976326912013422e-05, + "loss": 0.9582, + "step": 1032 + }, + { + "epoch": 1.06, + "grad_norm": 0.46728002332884067, + "learning_rate": 2.8894399944499974e-05, + "loss": 0.9023, + "step": 1033 + }, + { + "epoch": 1.06, + "grad_norm": 0.48539702863685763, + "learning_rate": 2.8812523425299e-05, + "loss": 0.9725, + "step": 1034 + }, + { + "epoch": 1.06, + "grad_norm": 0.42521485032555006, + "learning_rate": 2.873069772634281e-05, + "loss": 0.9525, + "step": 1035 + }, + { + "epoch": 1.07, + "grad_norm": 0.4068824768950637, + "learning_rate": 2.8648923219332997e-05, + "loss": 0.8318, + "step": 1036 + }, + { + "epoch": 1.07, + "grad_norm": 0.45227216852040214, + "learning_rate": 2.856720027573848e-05, + "loss": 1.0211, + "step": 1037 + }, + { + "epoch": 1.07, + "grad_norm": 0.42310911927974604, + "learning_rate": 2.8485529266794043e-05, + "loss": 0.9422, + "step": 1038 + }, + { + "epoch": 1.07, + "grad_norm": 0.4494478185683011, + "learning_rate": 2.8403910563498482e-05, + "loss": 0.9577, + "step": 1039 + }, + { + "epoch": 1.07, + "grad_norm": 0.517885146963669, + "learning_rate": 2.832234453661304e-05, + "loss": 0.9551, + "step": 1040 + }, + { + "epoch": 1.07, + "grad_norm": 0.46117112897797924, + "learning_rate": 2.8240831556659635e-05, + "loss": 0.9336, + "step": 1041 + }, + { + "epoch": 1.08, + "grad_norm": 0.4610208217170147, + "learning_rate": 2.815937199391924e-05, + "loss": 0.926, + "step": 1042 + }, + { + "epoch": 1.08, + "grad_norm": 0.445775414660019, + "learning_rate": 2.807796621843016e-05, + "loss": 0.9737, + "step": 1043 + }, + { + "epoch": 1.08, + "grad_norm": 0.46809555676786746, + "learning_rate": 2.799661459998638e-05, + "loss": 0.9916, + "step": 1044 + }, + { + "epoch": 1.08, + "grad_norm": 0.4366867439876077, + "learning_rate": 2.7915317508135848e-05, + "loss": 0.9549, + "step": 1045 + }, + { + "epoch": 1.08, + "grad_norm": 0.388979809570948, + "learning_rate": 2.7834075312178838e-05, + "loss": 0.8967, + "step": 1046 + }, + { + "epoch": 1.09, + "grad_norm": 0.45918748975994583, + "learning_rate": 2.775288838116626e-05, + "loss": 1.032, + "step": 1047 + }, + { + "epoch": 1.09, + "grad_norm": 0.4607131980517622, + "learning_rate": 2.767175708389794e-05, + "loss": 0.9638, + "step": 1048 + }, + { + "epoch": 1.09, + "grad_norm": 0.4583573438714022, + "learning_rate": 2.759068178892105e-05, + "loss": 0.8574, + "step": 1049 + }, + { + "epoch": 1.09, + "grad_norm": 0.4506028295056863, + "learning_rate": 2.750966286452828e-05, + "loss": 0.904, + "step": 1050 + }, + { + "epoch": 1.0, + "grad_norm": 0.5185176506584814, + "learning_rate": 2.7428700678756334e-05, + "loss": 0.8967, + "step": 1051 + }, + { + "epoch": 1.0, + "grad_norm": 0.4898570882755602, + "learning_rate": 2.7347795599384097e-05, + "loss": 0.9361, + "step": 1052 + }, + { + "epoch": 1.01, + "grad_norm": 0.4715298019133342, + "learning_rate": 2.7266947993931113e-05, + "loss": 0.8779, + "step": 1053 + }, + { + "epoch": 1.01, + "grad_norm": 0.4635241319149805, + "learning_rate": 2.7186158229655805e-05, + "loss": 0.8891, + "step": 1054 + }, + { + "epoch": 1.01, + "grad_norm": 0.4623859512294341, + "learning_rate": 2.7105426673553855e-05, + "loss": 0.8135, + "step": 1055 + }, + { + "epoch": 1.01, + "grad_norm": 0.4484341139188101, + "learning_rate": 2.7024753692356526e-05, + "loss": 0.9299, + "step": 1056 + }, + { + "epoch": 1.01, + "grad_norm": 0.46758873631226766, + "learning_rate": 2.694413965252901e-05, + "loss": 0.7719, + "step": 1057 + }, + { + "epoch": 1.01, + "grad_norm": 0.44651575615691763, + "learning_rate": 2.686358492026873e-05, + "loss": 0.9053, + "step": 1058 + }, + { + "epoch": 1.02, + "grad_norm": 0.5155154510102731, + "learning_rate": 2.6783089861503717e-05, + "loss": 1.0115, + "step": 1059 + }, + { + "epoch": 1.02, + "grad_norm": 0.49165729879997466, + "learning_rate": 2.670265484189093e-05, + "loss": 1.0023, + "step": 1060 + }, + { + "epoch": 1.02, + "grad_norm": 0.46460007026601335, + "learning_rate": 2.6622280226814582e-05, + "loss": 0.8825, + "step": 1061 + }, + { + "epoch": 1.02, + "grad_norm": 0.46979936842787406, + "learning_rate": 2.6541966381384487e-05, + "loss": 0.9605, + "step": 1062 + }, + { + "epoch": 1.02, + "grad_norm": 0.46001266952399505, + "learning_rate": 2.6461713670434445e-05, + "loss": 0.9462, + "step": 1063 + }, + { + "epoch": 1.03, + "grad_norm": 0.4965534915567658, + "learning_rate": 2.6381522458520498e-05, + "loss": 0.8811, + "step": 1064 + }, + { + "epoch": 1.03, + "grad_norm": 0.49637266648924844, + "learning_rate": 2.6301393109919353e-05, + "loss": 0.9957, + "step": 1065 + }, + { + "epoch": 1.03, + "grad_norm": 0.43785076760992014, + "learning_rate": 2.6221325988626686e-05, + "loss": 0.9486, + "step": 1066 + }, + { + "epoch": 1.03, + "grad_norm": 0.49005821628705587, + "learning_rate": 2.61413214583555e-05, + "loss": 1.0479, + "step": 1067 + }, + { + "epoch": 1.03, + "grad_norm": 0.43997445074451363, + "learning_rate": 2.6061379882534466e-05, + "loss": 0.91, + "step": 1068 + }, + { + "epoch": 1.03, + "grad_norm": 0.47582752827416086, + "learning_rate": 2.5981501624306296e-05, + "loss": 0.9833, + "step": 1069 + }, + { + "epoch": 1.04, + "grad_norm": 0.475702334553824, + "learning_rate": 2.590168704652605e-05, + "loss": 0.9387, + "step": 1070 + }, + { + "epoch": 1.04, + "grad_norm": 0.46797543318073137, + "learning_rate": 2.582193651175954e-05, + "loss": 0.8675, + "step": 1071 + }, + { + "epoch": 1.04, + "grad_norm": 0.502869009197904, + "learning_rate": 2.5742250382281638e-05, + "loss": 0.95, + "step": 1072 + }, + { + "epoch": 1.04, + "grad_norm": 0.48932775538777384, + "learning_rate": 2.5662629020074647e-05, + "loss": 0.8834, + "step": 1073 + }, + { + "epoch": 1.04, + "grad_norm": 0.5117876032138348, + "learning_rate": 2.5583072786826678e-05, + "loss": 0.9326, + "step": 1074 + }, + { + "epoch": 1.05, + "grad_norm": 0.45206721861847016, + "learning_rate": 2.5503582043929963e-05, + "loss": 0.9122, + "step": 1075 + }, + { + "epoch": 1.05, + "grad_norm": 0.478569067482158, + "learning_rate": 2.542415715247926e-05, + "loss": 0.8907, + "step": 1076 + }, + { + "epoch": 1.05, + "grad_norm": 0.4615412716962746, + "learning_rate": 2.5344798473270152e-05, + "loss": 0.8945, + "step": 1077 + }, + { + "epoch": 1.05, + "grad_norm": 0.4844154360763313, + "learning_rate": 2.526550636679751e-05, + "loss": 0.9517, + "step": 1078 + }, + { + "epoch": 1.05, + "grad_norm": 0.5073794346028544, + "learning_rate": 2.5186281193253726e-05, + "loss": 0.9625, + "step": 1079 + }, + { + "epoch": 1.05, + "grad_norm": 0.49073451792300016, + "learning_rate": 2.510712331252719e-05, + "loss": 0.8753, + "step": 1080 + }, + { + "epoch": 1.06, + "grad_norm": 0.45076485507525016, + "learning_rate": 2.5028033084200566e-05, + "loss": 0.8979, + "step": 1081 + }, + { + "epoch": 1.06, + "grad_norm": 0.4963935920521398, + "learning_rate": 2.494901086754923e-05, + "loss": 1.0287, + "step": 1082 + }, + { + "epoch": 1.06, + "grad_norm": 0.4553597545112614, + "learning_rate": 2.4870057021539628e-05, + "loss": 0.8624, + "step": 1083 + }, + { + "epoch": 1.06, + "grad_norm": 0.4574238940524242, + "learning_rate": 2.4791171904827548e-05, + "loss": 0.7923, + "step": 1084 + }, + { + "epoch": 1.06, + "grad_norm": 0.4771229947789797, + "learning_rate": 2.4712355875756666e-05, + "loss": 0.8912, + "step": 1085 + }, + { + "epoch": 1.07, + "grad_norm": 0.6440729243869642, + "learning_rate": 2.4633609292356737e-05, + "loss": 1.022, + "step": 1086 + }, + { + "epoch": 1.07, + "grad_norm": 0.47469602978835046, + "learning_rate": 2.4554932512342117e-05, + "loss": 1.0285, + "step": 1087 + }, + { + "epoch": 1.07, + "grad_norm": 0.5204144570210799, + "learning_rate": 2.4476325893110008e-05, + "loss": 1.0108, + "step": 1088 + }, + { + "epoch": 1.07, + "grad_norm": 0.47985475663425353, + "learning_rate": 2.4397789791738974e-05, + "loss": 0.8768, + "step": 1089 + }, + { + "epoch": 1.07, + "grad_norm": 0.4962961122211798, + "learning_rate": 2.431932456498717e-05, + "loss": 0.9328, + "step": 1090 + }, + { + "epoch": 1.07, + "grad_norm": 0.5084494441414602, + "learning_rate": 2.4240930569290867e-05, + "loss": 0.9511, + "step": 1091 + }, + { + "epoch": 1.08, + "grad_norm": 0.5089598583736452, + "learning_rate": 2.416260816076269e-05, + "loss": 0.9491, + "step": 1092 + }, + { + "epoch": 1.08, + "grad_norm": 0.48221222121556606, + "learning_rate": 2.408435769519014e-05, + "loss": 1.0269, + "step": 1093 + }, + { + "epoch": 1.08, + "grad_norm": 0.4703359259662912, + "learning_rate": 2.4006179528033844e-05, + "loss": 0.9321, + "step": 1094 + }, + { + "epoch": 1.08, + "grad_norm": 0.4790398670509022, + "learning_rate": 2.3928074014426077e-05, + "loss": 0.9077, + "step": 1095 + }, + { + "epoch": 1.08, + "grad_norm": 2.9832894495461018, + "learning_rate": 2.3850041509169007e-05, + "loss": 1.0024, + "step": 1096 + }, + { + "epoch": 1.09, + "grad_norm": 0.5157931440937726, + "learning_rate": 2.3772082366733224e-05, + "loss": 0.9028, + "step": 1097 + }, + { + "epoch": 1.09, + "grad_norm": 0.597535205195858, + "learning_rate": 2.3694196941255988e-05, + "loss": 0.9252, + "step": 1098 + }, + { + "epoch": 1.09, + "grad_norm": 0.5068463743400509, + "learning_rate": 2.361638558653974e-05, + "loss": 0.9658, + "step": 1099 + }, + { + "epoch": 1.09, + "grad_norm": 0.5227973124909935, + "learning_rate": 2.3538648656050463e-05, + "loss": 0.8794, + "step": 1100 + } + ], + "logging_steps": 1.0, + "max_steps": 1638, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 50, + "total_flos": 1139920827580416.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-1100/training_args.bin b/checkpoint-1100/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..019ba23a7fdbf801e820866ee31c58074364f008 --- /dev/null +++ b/checkpoint-1100/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bac5e6482c01f264d88bde470e1276960bac33683cfdcf64b2409e1a5fb73c89 +size 6776 diff --git a/checkpoint-1100/zero_to_fp32.py b/checkpoint-1100/zero_to_fp32.py new file mode 100755 index 0000000000000000000000000000000000000000..24cc342e78d1a006c782b3a4cd68d9ce786d8fd8 --- /dev/null +++ b/checkpoint-1100/zero_to_fp32.py @@ -0,0 +1,604 @@ +#!/usr/bin/env python + +# Copyright (c) Microsoft Corporation. +# SPDX-License-Identifier: Apache-2.0 + +# DeepSpeed Team + +# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets +# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in +# the future. Once extracted, the weights don't require DeepSpeed and can be used in any +# application. +# +# example: python zero_to_fp32.py . pytorch_model.bin + +import argparse +import torch +import glob +import math +import os +import re +from collections import OrderedDict +from dataclasses import dataclass + +# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with +# DeepSpeed data structures it has to be available in the current python environment. +from deepspeed.utils import logger +from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS, + FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES, + FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS) + + +@dataclass +class zero_model_state: + buffers: dict() + param_shapes: dict() + shared_params: list + ds_version: int + frozen_param_shapes: dict() + frozen_param_fragments: dict() + + +debug = 0 + +# load to cpu +device = torch.device('cpu') + + +def atoi(text): + return int(text) if text.isdigit() else text + + +def natural_keys(text): + ''' + alist.sort(key=natural_keys) sorts in human order + http://nedbatchelder.com/blog/200712/human_sorting.html + (See Toothy's implementation in the comments) + ''' + return [atoi(c) for c in re.split(r'(\d+)', text)] + + +def get_model_state_file(checkpoint_dir, zero_stage): + if not os.path.isdir(checkpoint_dir): + raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist") + + # there should be only one file + if zero_stage <= 2: + file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt") + elif zero_stage == 3: + file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt") + + if not os.path.exists(file): + raise FileNotFoundError(f"can't find model states file at '{file}'") + + return file + + +def get_checkpoint_files(checkpoint_dir, glob_pattern): + # XXX: need to test that this simple glob rule works for multi-node setup too + ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys) + + if len(ckpt_files) == 0: + raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'") + + return ckpt_files + + +def get_optim_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt") + + +def get_model_state_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_model_states.pt") + + +def parse_model_states(files): + zero_model_states = [] + for file in files: + state_dict = torch.load(file, map_location=device) + + if BUFFER_NAMES not in state_dict: + raise ValueError(f"{file} is not a model state checkpoint") + buffer_names = state_dict[BUFFER_NAMES] + if debug: + print("Found buffers:", buffer_names) + + # recover just the buffers while restoring them to fp32 if they were saved in fp16 + buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names} + param_shapes = state_dict[PARAM_SHAPES] + + # collect parameters that are included in param_shapes + param_names = [] + for s in param_shapes: + for name in s.keys(): + param_names.append(name) + + # update with frozen parameters + frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None) + if frozen_param_shapes is not None: + if debug: + print(f"Found frozen_param_shapes: {frozen_param_shapes}") + param_names += list(frozen_param_shapes.keys()) + + # handle shared params + shared_params = [[k, v] for k, v in state_dict["shared_params"].items()] + + ds_version = state_dict.get(DS_VERSION, None) + + frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None) + + z_model_state = zero_model_state(buffers=buffers, + param_shapes=param_shapes, + shared_params=shared_params, + ds_version=ds_version, + frozen_param_shapes=frozen_param_shapes, + frozen_param_fragments=frozen_param_fragments) + zero_model_states.append(z_model_state) + + return zero_model_states + + +def parse_optim_states(files, ds_checkpoint_dir): + + total_files = len(files) + state_dicts = [] + for f in files: + state_dict = torch.load(f, map_location=device) + # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights + # and also handle the case where it was already removed by another helper script + state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None) + state_dicts.append(state_dict) + + if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]: + raise ValueError(f"{files[0]} is not a zero checkpoint") + zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE] + world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT] + + # For ZeRO-2 each param group can have different partition_count as data parallelism for expert + # parameters can be different from data parallelism for non-expert parameters. So we can just + # use the max of the partition_count to get the dp world_size. + + if type(world_size) is list: + world_size = max(world_size) + + if world_size != total_files: + raise ValueError( + f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. " + "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes." + ) + + # the groups are named differently in each stage + if zero_stage <= 2: + fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS + elif zero_stage == 3: + fp32_groups_key = FP32_FLAT_GROUPS + else: + raise ValueError(f"unknown zero stage {zero_stage}") + + if zero_stage <= 2: + fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))] + elif zero_stage == 3: + # if there is more than one param group, there will be multiple flattened tensors - one + # flattened tensor per group - for simplicity merge them into a single tensor + # + # XXX: could make the script more memory efficient for when there are multiple groups - it + # will require matching the sub-lists of param_shapes for each param group flattened tensor + + fp32_flat_groups = [ + torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts)) + ] + + return zero_stage, world_size, fp32_flat_groups + + +def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters): + """ + Returns fp32 state_dict reconstructed from ds checkpoint + + Args: + - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are) + + """ + print(f"Processing zero checkpoint '{ds_checkpoint_dir}'") + + optim_files = get_optim_files(ds_checkpoint_dir) + zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir) + print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}") + + model_files = get_model_state_files(ds_checkpoint_dir) + + zero_model_states = parse_model_states(model_files) + print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}') + + if zero_stage <= 2: + return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + elif zero_stage == 3: + return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + + +def _zero2_merge_frozen_params(state_dict, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + frozen_param_fragments = zero_model_states[0].frozen_param_fragments + + if debug: + num_elem = sum(s.numel() for s in frozen_param_shapes.values()) + print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in frozen_param_fragments.values()]) + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + state_dict[name] = frozen_param_fragments[name] + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _has_callable(obj, fn): + attr = getattr(obj, fn, None) + return callable(attr) + + +def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + + # Reconstruction protocol: + # + # XXX: document this + + if debug: + for i in range(world_size): + for j in range(len(fp32_flat_groups[0])): + print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}") + + # XXX: memory usage doubles here (zero2) + num_param_groups = len(fp32_flat_groups[0]) + merged_single_partition_of_fp32_groups = [] + for i in range(num_param_groups): + merged_partitions = [sd[i] for sd in fp32_flat_groups] + full_single_fp32_vector = torch.cat(merged_partitions, 0) + merged_single_partition_of_fp32_groups.append(full_single_fp32_vector) + avail_numel = sum( + [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups]) + + if debug: + wanted_params = sum([len(shapes) for shapes in param_shapes]) + wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes]) + # not asserting if there is a mismatch due to possible padding + print(f"Have {avail_numel} numels to process.") + print(f"Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + total_numel = 0 + total_params = 0 + for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups): + offset = 0 + avail_numel = full_single_fp32_vector.numel() + for name, shape in shapes.items(): + + unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape) + total_numel += unpartitioned_numel + total_params += 1 + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape) + offset += unpartitioned_numel + + # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and + # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex + # paddings performed in the code it's almost impossible to predict the exact numbers w/o the + # live optimizer object, so we are checking that the numbers are within the right range + align_to = 2 * world_size + + def zero2_align(x): + return align_to * math.ceil(x / align_to) + + if debug: + print(f"original offset={offset}, avail_numel={avail_numel}") + + offset = zero2_align(offset) + avail_numel = zero2_align(avail_numel) + + if debug: + print(f"aligned offset={offset}, avail_numel={avail_numel}") + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero2_merge_frozen_params(state_dict, zero_model_states) + + _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def zero3_partitioned_param_info(unpartitioned_numel, world_size): + remainder = unpartitioned_numel % world_size + padding_numel = (world_size - remainder) if remainder else 0 + partitioned_numel = math.ceil(unpartitioned_numel / world_size) + return partitioned_numel, padding_numel + + +def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + if debug: + for i in range(world_size): + num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values()) + print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in zero_model_states[0].frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states) + state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape) + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + avail_numel = fp32_flat_groups[0].numel() * world_size + # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each + # param, re-consolidating each param, while dealing with padding if any + + # merge list of dicts, preserving order + param_shapes = {k: v for d in param_shapes for k, v in d.items()} + + if debug: + for i in range(world_size): + print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}") + + wanted_params = len(param_shapes) + wanted_numel = sum(shape.numel() for shape in param_shapes.values()) + # not asserting if there is a mismatch due to possible padding + avail_numel = fp32_flat_groups[0].numel() * world_size + print(f"Trainable params: Have {avail_numel} numels to process.") + print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + offset = 0 + total_numel = 0 + total_params = 0 + for name, shape in param_shapes.items(): + + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + total_params += 1 + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + # XXX: memory usage doubles here + state_dict[name] = torch.cat( + tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)), + 0).narrow(0, 0, unpartitioned_numel).view(shape) + offset += partitioned_numel + + offset *= world_size + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero3_merge_frozen_params(state_dict, world_size, zero_model_states) + + _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with + ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example + via a model hub. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + + Returns: + - pytorch ``state_dict`` + + Note: this approach may not work if your application doesn't have sufficient free CPU memory and + you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with + the checkpoint. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + # do the training and checkpoint saving + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu + model = model.cpu() # move to cpu + model.load_state_dict(state_dict) + # submit to model hub or save the model to share with others + + In this example the ``model`` will no longer be usable in the deepspeed context of the same + application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead. + + """ + if tag is None: + latest_path = os.path.join(checkpoint_dir, 'latest') + if os.path.isfile(latest_path): + with open(latest_path, 'r') as fd: + tag = fd.read().strip() + else: + raise ValueError(f"Unable to find 'latest' file at {latest_path}") + + ds_checkpoint_dir = os.path.join(checkpoint_dir, tag) + + if not os.path.isdir(ds_checkpoint_dir): + raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist") + + return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters) + + +def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be + loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + """ + + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters) + print(f"Saving fp32 state dict to {output_file}") + torch.save(state_dict, output_file) + + +def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None): + """ + 1. Put the provided model to cpu + 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` + 3. Load it into the provided model + + Args: + - ``model``: the model object to update + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + + Returns: + - ``model`: modified model + + Make sure you have plenty of CPU memory available before you call this function. If you don't + have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it + conveniently placed for you in the checkpoint folder. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint + model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir) + # submit to model hub or save the model to share with others + + Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context + of the same application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + """ + logger.info(f"Extracting fp32 weights") + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag) + + logger.info(f"Overwriting model with fp32 weights") + model = model.cpu() + model.load_state_dict(state_dict, strict=False) + + return model + + +if __name__ == "__main__": + + parser = argparse.ArgumentParser() + parser.add_argument("checkpoint_dir", + type=str, + help="path to the desired checkpoint folder, e.g., path/checkpoint-12") + parser.add_argument( + "output_file", + type=str, + help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)") + parser.add_argument("-t", + "--tag", + type=str, + default=None, + help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1") + parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters") + parser.add_argument("-d", "--debug", action='store_true', help="enable debug") + args = parser.parse_args() + + debug = args.debug + + convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, + args.output_file, + tag=args.tag, + exclude_frozen_parameters=args.exclude_frozen_parameters) diff --git a/checkpoint-450/README.md b/checkpoint-450/README.md new file mode 100644 index 0000000000000000000000000000000000000000..16b1eacdd9353dec380a08ee77ce6ed5ab50f12e --- /dev/null +++ b/checkpoint-450/README.md @@ -0,0 +1,202 @@ +--- +library_name: peft +base_model: gotzmann/uni +--- + +# Model Card for Model ID + + + + + +## Model Details + +### Model Description + + + + + +- **Developed by:** [More Information Needed] +- **Funded by [optional]:** [More Information Needed] +- **Shared by [optional]:** [More Information Needed] +- **Model type:** [More Information Needed] +- **Language(s) (NLP):** [More Information Needed] +- **License:** [More Information Needed] +- **Finetuned from model [optional]:** [More Information Needed] + +### Model Sources [optional] + + + +- **Repository:** [More Information Needed] +- **Paper [optional]:** [More Information Needed] +- **Demo [optional]:** [More Information Needed] + +## Uses + + + +### Direct Use + + + +[More Information Needed] + +### Downstream Use [optional] + + + +[More Information Needed] + +### Out-of-Scope Use + + + +[More Information Needed] + +## Bias, Risks, and Limitations + + + +[More Information Needed] + +### Recommendations + + + +Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations. + +## How to Get Started with the Model + +Use the code below to get started with the model. + +[More Information Needed] + +## Training Details + +### Training Data + + + +[More Information Needed] + +### Training Procedure + + + +#### Preprocessing [optional] + +[More Information Needed] + + +#### Training Hyperparameters + +- **Training regime:** [More Information Needed] + +#### Speeds, Sizes, Times [optional] + + + +[More Information Needed] + +## Evaluation + + + +### Testing Data, Factors & Metrics + +#### Testing Data + + + +[More Information Needed] + +#### Factors + + + +[More Information Needed] + +#### Metrics + + + +[More Information Needed] + +### Results + +[More Information Needed] + +#### Summary + + + +## Model Examination [optional] + + + +[More Information Needed] + +## Environmental Impact + + + +Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700). + +- **Hardware Type:** [More Information Needed] +- **Hours used:** [More Information Needed] +- **Cloud Provider:** [More Information Needed] +- **Compute Region:** [More Information Needed] +- **Carbon Emitted:** [More Information Needed] + +## Technical Specifications [optional] + +### Model Architecture and Objective + +[More Information Needed] + +### Compute Infrastructure + +[More Information Needed] + +#### Hardware + +[More Information Needed] + +#### Software + +[More Information Needed] + +## Citation [optional] + + + +**BibTeX:** + +[More Information Needed] + +**APA:** + +[More Information Needed] + +## Glossary [optional] + + + +[More Information Needed] + +## More Information [optional] + +[More Information Needed] + +## Model Card Authors [optional] + +[More Information Needed] + +## Model Card Contact + +[More Information Needed] +### Framework versions + +- PEFT 0.10.0 \ No newline at end of file diff --git a/checkpoint-450/adapter_config.json b/checkpoint-450/adapter_config.json new file mode 100644 index 0000000000000000000000000000000000000000..3cd6dba5d79f7ca21fd4ad465cbbcac1e0960476 --- /dev/null +++ b/checkpoint-450/adapter_config.json @@ -0,0 +1,31 @@ +{ + "alpha_pattern": {}, + "auto_mapping": null, + "base_model_name_or_path": "gotzmann/uni", + "bias": "none", + "fan_in_fan_out": false, + "inference_mode": true, + "init_lora_weights": true, + "layer_replication": null, + "layers_pattern": null, + "layers_to_transform": null, + "loftq_config": {}, + "lora_alpha": 128, + "lora_dropout": 0.0, + "megatron_config": null, + "megatron_core": "megatron.core", + "modules_to_save": null, + "peft_type": "LORA", + "r": 128, + "rank_pattern": {}, + "revision": null, + "target_modules": [ + "v_proj", + "k_proj", + "q_proj", + "o_proj" + ], + "task_type": "CAUSAL_LM", + "use_dora": false, + "use_rslora": true +} \ No newline at end of file diff --git a/checkpoint-450/adapter_model.safetensors b/checkpoint-450/adapter_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..48cc46217d14642e6c5cf96af194d4afe7365fe0 --- /dev/null +++ b/checkpoint-450/adapter_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0189156bd202d07758f9b3af49e74d7a50c78a8e44013b92375c1c3deae1fd8b +size 1048664848 diff --git a/checkpoint-450/global_step450/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt b/checkpoint-450/global_step450/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..bc77e2dc53f2730674512f300c2657a3b2772ba0 --- /dev/null +++ b/checkpoint-450/global_step450/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:bcb0a7e1c71d757416c1979d9712bf9ae8da6a7dadcf890a182653ef6edd0b9a +size 787270042 diff --git a/checkpoint-450/global_step450/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt b/checkpoint-450/global_step450/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..934adb07c9cd7a6ae499f7f7b904ee70190b0d6d --- /dev/null +++ b/checkpoint-450/global_step450/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3816457b22b729fb8ccc5aea564c58c51f3feb22310620c1690c8ba78dc1f905 +size 787270042 diff --git a/checkpoint-450/global_step450/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt b/checkpoint-450/global_step450/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..9f877de0b9d4f04271bf4209f676857ab5f5caa1 --- /dev/null +++ b/checkpoint-450/global_step450/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:677fd137e293ab4ba17ecb1a7d00cc2c79a6c8ad5b97bd667b29c6a7d9238930 +size 787270042 diff --git a/checkpoint-450/global_step450/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt b/checkpoint-450/global_step450/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..b7ad2a23667cfab036d2f8d983a86b4f46e9fddb --- /dev/null +++ b/checkpoint-450/global_step450/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8045c9f7d95d85a7b56e6473d48be84ad92f75bb0d3de107b55e3abfca20530a +size 787270042 diff --git a/checkpoint-450/global_step450/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt b/checkpoint-450/global_step450/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..b37a37c078934d0b1a02e8514541456c0d8b8b9f --- /dev/null +++ b/checkpoint-450/global_step450/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2d1bd333d9570d6db4d01adf6edd0ce2a0925f28aa998c2d714a62d00ce593b8 +size 787270042 diff --git a/checkpoint-450/global_step450/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt b/checkpoint-450/global_step450/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..11a5ac9c1cff36b08c261dfc5794314d832a51b2 --- /dev/null +++ b/checkpoint-450/global_step450/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:57809e10728bce5408308872b7988bfd3a973523220ec1c9c250cd0aefb9183d +size 787270042 diff --git a/checkpoint-450/global_step450/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt b/checkpoint-450/global_step450/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..62b798efbebc4e455df0b773d4cac9b1e823c45a --- /dev/null +++ b/checkpoint-450/global_step450/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:170f6bf6c8936d2807f11d74ba597ce08924358b3c7079834a2c6f958f21742a +size 787270042 diff --git a/checkpoint-450/global_step450/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt b/checkpoint-450/global_step450/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..403021b1f29b98378c0927029aca697a2d95720e --- /dev/null +++ b/checkpoint-450/global_step450/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0b639bc02be13a9b0459df37e333bc1b29fa8077acde40bb792dd5d10cd41250 +size 787270042 diff --git a/checkpoint-450/global_step450/zero_pp_rank_0_mp_rank_00_model_states.pt b/checkpoint-450/global_step450/zero_pp_rank_0_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..b81c53b6eb98dd7c493af564306aa7a40a090304 --- /dev/null +++ b/checkpoint-450/global_step450/zero_pp_rank_0_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:090fe59e869db598283a64e5dc2789748a58152e1018b12e84e6ec9f22f6c1ef +size 653742 diff --git a/checkpoint-450/global_step450/zero_pp_rank_1_mp_rank_00_model_states.pt b/checkpoint-450/global_step450/zero_pp_rank_1_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..b704b54319eb800c66e48555423a4b30d53f1144 --- /dev/null +++ b/checkpoint-450/global_step450/zero_pp_rank_1_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a922dbda3030e4c2add850cfb9e11d1eaeeb822edb40340f263f9658f5a5a620 +size 653742 diff --git a/checkpoint-450/global_step450/zero_pp_rank_2_mp_rank_00_model_states.pt b/checkpoint-450/global_step450/zero_pp_rank_2_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..e21ae584e1f8c8042d4196587ad5c8e7d546f800 --- /dev/null +++ b/checkpoint-450/global_step450/zero_pp_rank_2_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7779c1112419f128a16587b227e6c4b2fb78c16eed4d94fa256dddc4d7eb0116 +size 653742 diff --git a/checkpoint-450/global_step450/zero_pp_rank_3_mp_rank_00_model_states.pt b/checkpoint-450/global_step450/zero_pp_rank_3_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..5746bff4623cf6a788a2eaf54ea1e532efcc6749 --- /dev/null +++ b/checkpoint-450/global_step450/zero_pp_rank_3_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:96c4349c65cff45d1bc9e710a0a5fcd2fb1fdcf8b86a3872f2e5693e087e4eab +size 653742 diff --git a/checkpoint-450/global_step450/zero_pp_rank_4_mp_rank_00_model_states.pt b/checkpoint-450/global_step450/zero_pp_rank_4_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..c36f00f85a1f6367fa0237e914023cfb6f339094 --- /dev/null +++ b/checkpoint-450/global_step450/zero_pp_rank_4_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9fa3fb28afd32f6ce6d88deb1ff883dffea4785b833f53c0bbaec9ce14e7b643 +size 653742 diff --git a/checkpoint-450/global_step450/zero_pp_rank_5_mp_rank_00_model_states.pt b/checkpoint-450/global_step450/zero_pp_rank_5_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..48285cca8a70311956589d91b9ab25b98c2e429b --- /dev/null +++ b/checkpoint-450/global_step450/zero_pp_rank_5_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:60ac36ebcb8f6bcefb72bf66bc760c5cc4ead6edddccbb6be2aa84af7b5191e0 +size 653742 diff --git a/checkpoint-450/global_step450/zero_pp_rank_6_mp_rank_00_model_states.pt b/checkpoint-450/global_step450/zero_pp_rank_6_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..69685d0a55c64731170e89c8b4239b38a56a8857 --- /dev/null +++ b/checkpoint-450/global_step450/zero_pp_rank_6_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f54b4c130357abfe0a3bc252acbb4965805212c35ffb1b66c3f81d2b582307c9 +size 653742 diff --git a/checkpoint-450/global_step450/zero_pp_rank_7_mp_rank_00_model_states.pt b/checkpoint-450/global_step450/zero_pp_rank_7_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..f11cd1a664e4e1cb15f347f01098821fcfbb89f2 --- /dev/null +++ b/checkpoint-450/global_step450/zero_pp_rank_7_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ac62b78b9cc99f4838aa43e132cb1829fa014d54127a23ef7742cce9d6d33e09 +size 653742 diff --git a/checkpoint-450/latest b/checkpoint-450/latest new file mode 100644 index 0000000000000000000000000000000000000000..1480cb77d39204b528f1515cd41be7cdbe05e78b --- /dev/null +++ b/checkpoint-450/latest @@ -0,0 +1 @@ +global_step450 \ No newline at end of file diff --git a/checkpoint-450/rng_state_0.pth b/checkpoint-450/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..b346349ce12dd5a17d4b91ed2a5722bb52550950 --- /dev/null +++ b/checkpoint-450/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ad8a35afd8967cbb748405387e44426e43ad127028e826eddc9b67d2ca873c85 +size 15984 diff --git a/checkpoint-450/rng_state_1.pth b/checkpoint-450/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..68f3c6994456cb8d0592a5375d99503c8924b1c4 --- /dev/null +++ b/checkpoint-450/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f338ce80d7c441076bfc8c53b84067a0181f5a14e80c13d5acb8150b659f4d73 +size 15984 diff --git a/checkpoint-450/rng_state_2.pth b/checkpoint-450/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..be044f6ceeed587d30e80c2f72d5aa19fdc9947b --- /dev/null +++ b/checkpoint-450/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c9fbc9fa428939be10b46779f0eb5cd833e0da426b1cbdee77b3a55b6952235b +size 15984 diff --git a/checkpoint-450/rng_state_3.pth b/checkpoint-450/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..fc825249656a9b858782542bd3f4386250f1dfe0 --- /dev/null +++ b/checkpoint-450/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ac55dba0b79d5fa4699d239da2f966d52040d576d31234ac8d4632e6956481bc +size 15984 diff --git a/checkpoint-450/rng_state_4.pth b/checkpoint-450/rng_state_4.pth new file mode 100644 index 0000000000000000000000000000000000000000..d30f52a44be563c152ae09db6ae934da6da0d3ed --- /dev/null +++ b/checkpoint-450/rng_state_4.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:af2d0c015100768ffa23faf3b6c2d54ea89eb045603e30e55cd211e06ff34972 +size 15984 diff --git a/checkpoint-450/rng_state_5.pth b/checkpoint-450/rng_state_5.pth new file mode 100644 index 0000000000000000000000000000000000000000..c8715d27ab23ae545d58039cf949cc44ecc1da5e --- /dev/null +++ b/checkpoint-450/rng_state_5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c60a1b40608e34bc801c8231f97b81c53b5290dfaed1b9cd0ccbeca29574a991 +size 15984 diff --git a/checkpoint-450/rng_state_6.pth b/checkpoint-450/rng_state_6.pth new file mode 100644 index 0000000000000000000000000000000000000000..1ed791b6ef76eadf0b0c55a5733411771e2ae027 --- /dev/null +++ b/checkpoint-450/rng_state_6.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3ad6a142a403eb9aafc4a3a9a856bca648fe31fd22d796867baca31fb13656aa +size 15984 diff --git a/checkpoint-450/rng_state_7.pth b/checkpoint-450/rng_state_7.pth new file mode 100644 index 0000000000000000000000000000000000000000..800c3bbbc5edf7db01a8316069d439c5fb8d8c30 --- /dev/null +++ b/checkpoint-450/rng_state_7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:38bc23a138cc800b22881742c0f3f9a71731a9a7111c6058a0077e6274d21773 +size 15984 diff --git a/checkpoint-450/scheduler.pt b/checkpoint-450/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..e46ab54fb8147fbaae665f86673db3154fa95ca4 --- /dev/null +++ b/checkpoint-450/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8233715ac83c710af8824b8baf5d0d3ff081dedd7c651ffe5e70c9cee90b4ace +size 1064 diff --git a/checkpoint-450/special_tokens_map.json b/checkpoint-450/special_tokens_map.json new file mode 100644 index 0000000000000000000000000000000000000000..72ecfeeb7e14d244c936169d2ed139eeae235ef1 --- /dev/null +++ b/checkpoint-450/special_tokens_map.json @@ -0,0 +1,24 @@ +{ + "bos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "eos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "pad_token": "", + "unk_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + } +} diff --git a/checkpoint-450/tokenizer.model b/checkpoint-450/tokenizer.model new file mode 100644 index 0000000000000000000000000000000000000000..6c00c742ce03c627d6cd5b795984876fa49fa899 --- /dev/null +++ b/checkpoint-450/tokenizer.model @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347 +size 499723 diff --git a/checkpoint-450/tokenizer_config.json b/checkpoint-450/tokenizer_config.json new file mode 100644 index 0000000000000000000000000000000000000000..bb5a9f09d8c0f3c32c66fc6118fe5c76c5c6fd90 --- /dev/null +++ b/checkpoint-450/tokenizer_config.json @@ -0,0 +1,45 @@ +{ + "add_bos_token": true, + "add_eos_token": false, + "add_prefix_space": false, + "added_tokens_decoder": { + "0": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "1": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "2": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + } + }, + "bos_token": "", + "chat_template": "{% if messages[0]['role'] == 'system' %}{% set system_message = messages[0]['content'] %}{% endif %}{% if system_message is defined %}{{ '' + '### System:\\n\\n' + system_message }}{% endif %}{% for message in messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ '\\n\\n### Human:\\n\\n' + content }}{% elif message['role'] == 'assistant' %}{{ '\\n\\n### Assistant:\\n\\n' + content + '' }}{% endif %}{% endfor %}", + "clean_up_tokenization_spaces": false, + "eos_token": "", + "legacy": false, + "model_max_length": 1000000000000000019884624838656, + "pad_token": "", + "padding_side": "right", + "sp_model_kwargs": {}, + "spaces_between_special_tokens": false, + "split_special_tokens": false, + "tokenizer_class": "LlamaTokenizer", + "unk_token": "", + "use_default_system_prompt": false +} diff --git a/checkpoint-450/trainer_state.json b/checkpoint-450/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..2573fd6fdd6034ae34ad05bb5f303907770590b6 --- /dev/null +++ b/checkpoint-450/trainer_state.json @@ -0,0 +1,3171 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 0.823045267489712, + "eval_steps": 500, + "global_step": 450, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "grad_norm": 0.849355824164473, + "learning_rate": 4.878048780487805e-07, + "loss": 1.3655, + "step": 1 + }, + { + "epoch": 0.0, + "grad_norm": 10.01567518957158, + "learning_rate": 9.75609756097561e-07, + "loss": 1.5767, + "step": 2 + }, + { + "epoch": 0.01, + "grad_norm": 0.6466000875559635, + "learning_rate": 1.4634146341463414e-06, + "loss": 1.3913, + "step": 3 + }, + { + "epoch": 0.01, + "grad_norm": 0.6644565932010504, + "learning_rate": 1.951219512195122e-06, + "loss": 1.3218, + "step": 4 + }, + { + "epoch": 0.01, + "grad_norm": 0.571354207588475, + "learning_rate": 2.4390243902439027e-06, + "loss": 1.3597, + "step": 5 + }, + { + "epoch": 0.01, + "grad_norm": 0.31036262839244955, + "learning_rate": 2.926829268292683e-06, + "loss": 1.2832, + "step": 6 + }, + { + "epoch": 0.01, + "grad_norm": 0.2622135027188184, + "learning_rate": 3.414634146341464e-06, + "loss": 1.2161, + "step": 7 + }, + { + "epoch": 0.01, + "grad_norm": 0.296824630261661, + "learning_rate": 3.902439024390244e-06, + "loss": 1.2985, + "step": 8 + }, + { + "epoch": 0.02, + "grad_norm": 0.2557267467361569, + "learning_rate": 4.390243902439025e-06, + "loss": 1.3175, + "step": 9 + }, + { + "epoch": 0.02, + "grad_norm": 0.23418939513890769, + "learning_rate": 4.8780487804878055e-06, + "loss": 1.2617, + "step": 10 + }, + { + "epoch": 0.02, + "grad_norm": 0.2364760983285843, + "learning_rate": 5.365853658536586e-06, + "loss": 1.3103, + "step": 11 + }, + { + "epoch": 0.02, + "grad_norm": 0.23893034721889, + "learning_rate": 5.853658536585366e-06, + "loss": 1.2405, + "step": 12 + }, + { + "epoch": 0.02, + "grad_norm": 0.25563593295485887, + "learning_rate": 6.341463414634147e-06, + "loss": 1.2831, + "step": 13 + }, + { + "epoch": 0.03, + "grad_norm": 0.23239975352661665, + "learning_rate": 6.829268292682928e-06, + "loss": 1.3125, + "step": 14 + }, + { + "epoch": 0.03, + "grad_norm": 0.3092813858209507, + "learning_rate": 7.317073170731707e-06, + "loss": 1.2422, + "step": 15 + }, + { + "epoch": 0.03, + "grad_norm": 0.282563380367434, + "learning_rate": 7.804878048780489e-06, + "loss": 1.2453, + "step": 16 + }, + { + "epoch": 0.03, + "grad_norm": 0.22065680088315018, + "learning_rate": 8.292682926829268e-06, + "loss": 1.2491, + "step": 17 + }, + { + "epoch": 0.03, + "grad_norm": 0.22777800877980184, + "learning_rate": 8.78048780487805e-06, + "loss": 1.2655, + "step": 18 + }, + { + "epoch": 0.03, + "grad_norm": 0.22145212540177928, + "learning_rate": 9.268292682926831e-06, + "loss": 1.2413, + "step": 19 + }, + { + "epoch": 0.04, + "grad_norm": 0.22482351883112714, + "learning_rate": 9.756097560975611e-06, + "loss": 1.2653, + "step": 20 + }, + { + "epoch": 0.04, + "grad_norm": 0.20823080508385733, + "learning_rate": 1.024390243902439e-05, + "loss": 1.2374, + "step": 21 + }, + { + "epoch": 0.04, + "grad_norm": 0.26025492562935737, + "learning_rate": 1.0731707317073172e-05, + "loss": 1.2065, + "step": 22 + }, + { + "epoch": 0.04, + "grad_norm": 0.2150252124176173, + "learning_rate": 1.1219512195121953e-05, + "loss": 1.2782, + "step": 23 + }, + { + "epoch": 0.04, + "grad_norm": 0.2505915177425618, + "learning_rate": 1.1707317073170731e-05, + "loss": 1.2742, + "step": 24 + }, + { + "epoch": 0.05, + "grad_norm": 0.20129223044786942, + "learning_rate": 1.2195121951219513e-05, + "loss": 1.3366, + "step": 25 + }, + { + "epoch": 0.05, + "grad_norm": 0.1973508510397107, + "learning_rate": 1.2682926829268294e-05, + "loss": 1.2476, + "step": 26 + }, + { + "epoch": 0.05, + "grad_norm": 0.27103325392437194, + "learning_rate": 1.3170731707317076e-05, + "loss": 1.2325, + "step": 27 + }, + { + "epoch": 0.05, + "grad_norm": 0.17954976411006285, + "learning_rate": 1.3658536585365855e-05, + "loss": 1.2523, + "step": 28 + }, + { + "epoch": 0.05, + "grad_norm": 0.22216997851088888, + "learning_rate": 1.4146341463414635e-05, + "loss": 1.3297, + "step": 29 + }, + { + "epoch": 0.05, + "grad_norm": 0.2071458864548587, + "learning_rate": 1.4634146341463415e-05, + "loss": 1.2127, + "step": 30 + }, + { + "epoch": 0.06, + "grad_norm": 0.18039422081622164, + "learning_rate": 1.5121951219512196e-05, + "loss": 1.2509, + "step": 31 + }, + { + "epoch": 0.06, + "grad_norm": 0.18631254372974412, + "learning_rate": 1.5609756097560978e-05, + "loss": 1.2247, + "step": 32 + }, + { + "epoch": 0.06, + "grad_norm": 0.18843872523649827, + "learning_rate": 1.6097560975609757e-05, + "loss": 1.195, + "step": 33 + }, + { + "epoch": 0.06, + "grad_norm": 0.2163847267778325, + "learning_rate": 1.6585365853658537e-05, + "loss": 1.2179, + "step": 34 + }, + { + "epoch": 0.06, + "grad_norm": 0.19687688475496104, + "learning_rate": 1.7073170731707317e-05, + "loss": 1.2763, + "step": 35 + }, + { + "epoch": 0.07, + "grad_norm": 0.20409643064887947, + "learning_rate": 1.75609756097561e-05, + "loss": 1.253, + "step": 36 + }, + { + "epoch": 0.07, + "grad_norm": 0.1879182661759335, + "learning_rate": 1.804878048780488e-05, + "loss": 1.2586, + "step": 37 + }, + { + "epoch": 0.07, + "grad_norm": 0.19400648948514373, + "learning_rate": 1.8536585365853663e-05, + "loss": 1.2154, + "step": 38 + }, + { + "epoch": 0.07, + "grad_norm": 0.1878879343148452, + "learning_rate": 1.902439024390244e-05, + "loss": 1.2304, + "step": 39 + }, + { + "epoch": 0.07, + "grad_norm": 0.17687475469924052, + "learning_rate": 1.9512195121951222e-05, + "loss": 1.2351, + "step": 40 + }, + { + "epoch": 0.07, + "grad_norm": 0.18223935625384885, + "learning_rate": 2e-05, + "loss": 1.2222, + "step": 41 + }, + { + "epoch": 0.08, + "grad_norm": 0.1943061629408338, + "learning_rate": 2.048780487804878e-05, + "loss": 1.2044, + "step": 42 + }, + { + "epoch": 0.08, + "grad_norm": 0.17027514338700078, + "learning_rate": 2.0975609756097564e-05, + "loss": 1.1548, + "step": 43 + }, + { + "epoch": 0.08, + "grad_norm": 0.18553769630586192, + "learning_rate": 2.1463414634146344e-05, + "loss": 1.2721, + "step": 44 + }, + { + "epoch": 0.08, + "grad_norm": 0.19732826914228765, + "learning_rate": 2.1951219512195124e-05, + "loss": 1.3097, + "step": 45 + }, + { + "epoch": 0.08, + "grad_norm": 0.18714230986631472, + "learning_rate": 2.2439024390243907e-05, + "loss": 1.2662, + "step": 46 + }, + { + "epoch": 0.09, + "grad_norm": 0.19988987568002223, + "learning_rate": 2.2926829268292683e-05, + "loss": 1.2904, + "step": 47 + }, + { + "epoch": 0.09, + "grad_norm": 0.17744650133390918, + "learning_rate": 2.3414634146341463e-05, + "loss": 1.1825, + "step": 48 + }, + { + "epoch": 0.09, + "grad_norm": 0.16576734763834533, + "learning_rate": 2.3902439024390246e-05, + "loss": 1.1858, + "step": 49 + }, + { + "epoch": 0.09, + "grad_norm": 0.179591794065527, + "learning_rate": 2.4390243902439026e-05, + "loss": 1.2711, + "step": 50 + }, + { + "epoch": 0.09, + "grad_norm": 0.17923464471176911, + "learning_rate": 2.4878048780487805e-05, + "loss": 1.2289, + "step": 51 + }, + { + "epoch": 0.1, + "grad_norm": 0.18991742907836837, + "learning_rate": 2.536585365853659e-05, + "loss": 1.3097, + "step": 52 + }, + { + "epoch": 0.1, + "grad_norm": 0.19849796137254636, + "learning_rate": 2.5853658536585368e-05, + "loss": 1.2489, + "step": 53 + }, + { + "epoch": 0.1, + "grad_norm": 0.17452371110976383, + "learning_rate": 2.634146341463415e-05, + "loss": 1.2461, + "step": 54 + }, + { + "epoch": 0.1, + "grad_norm": 0.17671022353085036, + "learning_rate": 2.682926829268293e-05, + "loss": 1.153, + "step": 55 + }, + { + "epoch": 0.1, + "grad_norm": 0.36820559192096686, + "learning_rate": 2.731707317073171e-05, + "loss": 1.2431, + "step": 56 + }, + { + "epoch": 0.1, + "grad_norm": 0.20331468526494198, + "learning_rate": 2.7804878048780487e-05, + "loss": 1.2575, + "step": 57 + }, + { + "epoch": 0.11, + "grad_norm": 0.2402486598118377, + "learning_rate": 2.829268292682927e-05, + "loss": 1.2538, + "step": 58 + }, + { + "epoch": 0.11, + "grad_norm": 0.2549409484173144, + "learning_rate": 2.878048780487805e-05, + "loss": 1.2065, + "step": 59 + }, + { + "epoch": 0.11, + "grad_norm": 0.2053105349872685, + "learning_rate": 2.926829268292683e-05, + "loss": 1.2094, + "step": 60 + }, + { + "epoch": 0.11, + "grad_norm": 0.17971910872957886, + "learning_rate": 2.9756097560975613e-05, + "loss": 1.228, + "step": 61 + }, + { + "epoch": 0.11, + "grad_norm": 0.1885853654992973, + "learning_rate": 3.0243902439024392e-05, + "loss": 1.2286, + "step": 62 + }, + { + "epoch": 0.12, + "grad_norm": 0.1848524571968613, + "learning_rate": 3.073170731707317e-05, + "loss": 1.2718, + "step": 63 + }, + { + "epoch": 0.12, + "grad_norm": 0.18734105883548513, + "learning_rate": 3.1219512195121955e-05, + "loss": 1.2357, + "step": 64 + }, + { + "epoch": 0.12, + "grad_norm": 0.17774668052121825, + "learning_rate": 3.170731707317074e-05, + "loss": 1.1509, + "step": 65 + }, + { + "epoch": 0.12, + "grad_norm": 0.17890968008080646, + "learning_rate": 3.2195121951219514e-05, + "loss": 1.1924, + "step": 66 + }, + { + "epoch": 0.12, + "grad_norm": 0.18249273371332375, + "learning_rate": 3.268292682926829e-05, + "loss": 1.2545, + "step": 67 + }, + { + "epoch": 0.12, + "grad_norm": 0.21064122671902577, + "learning_rate": 3.3170731707317074e-05, + "loss": 1.2832, + "step": 68 + }, + { + "epoch": 0.13, + "grad_norm": 0.1820064171955093, + "learning_rate": 3.365853658536586e-05, + "loss": 1.2071, + "step": 69 + }, + { + "epoch": 0.13, + "grad_norm": 0.16996662800553433, + "learning_rate": 3.414634146341463e-05, + "loss": 1.2073, + "step": 70 + }, + { + "epoch": 0.13, + "grad_norm": 0.1618669302922445, + "learning_rate": 3.4634146341463416e-05, + "loss": 1.1289, + "step": 71 + }, + { + "epoch": 0.13, + "grad_norm": 0.18948744950985544, + "learning_rate": 3.51219512195122e-05, + "loss": 1.2915, + "step": 72 + }, + { + "epoch": 0.13, + "grad_norm": 0.18326143691603383, + "learning_rate": 3.5609756097560976e-05, + "loss": 1.2238, + "step": 73 + }, + { + "epoch": 0.14, + "grad_norm": 0.17410704510700503, + "learning_rate": 3.609756097560976e-05, + "loss": 1.1784, + "step": 74 + }, + { + "epoch": 0.14, + "grad_norm": 0.1983667344995625, + "learning_rate": 3.658536585365854e-05, + "loss": 1.2452, + "step": 75 + }, + { + "epoch": 0.14, + "grad_norm": 0.3416310763369357, + "learning_rate": 3.7073170731707325e-05, + "loss": 1.1972, + "step": 76 + }, + { + "epoch": 0.14, + "grad_norm": 0.2776466983511955, + "learning_rate": 3.75609756097561e-05, + "loss": 1.3121, + "step": 77 + }, + { + "epoch": 0.14, + "grad_norm": 0.20026129636576834, + "learning_rate": 3.804878048780488e-05, + "loss": 1.2436, + "step": 78 + }, + { + "epoch": 0.14, + "grad_norm": 0.21064549243917835, + "learning_rate": 3.853658536585366e-05, + "loss": 1.2064, + "step": 79 + }, + { + "epoch": 0.15, + "grad_norm": 0.22119482175714267, + "learning_rate": 3.9024390243902444e-05, + "loss": 1.2715, + "step": 80 + }, + { + "epoch": 0.15, + "grad_norm": 0.23047133748844142, + "learning_rate": 3.951219512195122e-05, + "loss": 1.2888, + "step": 81 + }, + { + "epoch": 0.15, + "grad_norm": 0.18741863156973176, + "learning_rate": 4e-05, + "loss": 1.248, + "step": 82 + }, + { + "epoch": 0.15, + "grad_norm": 0.1747859810629604, + "learning_rate": 4.0487804878048786e-05, + "loss": 1.1683, + "step": 83 + }, + { + "epoch": 0.15, + "grad_norm": 0.1896944798413341, + "learning_rate": 4.097560975609756e-05, + "loss": 1.2155, + "step": 84 + }, + { + "epoch": 0.16, + "grad_norm": 0.18724128114363303, + "learning_rate": 4.1463414634146346e-05, + "loss": 1.2273, + "step": 85 + }, + { + "epoch": 0.16, + "grad_norm": 0.17368125504855478, + "learning_rate": 4.195121951219513e-05, + "loss": 1.224, + "step": 86 + }, + { + "epoch": 0.16, + "grad_norm": 0.18371141013625703, + "learning_rate": 4.2439024390243905e-05, + "loss": 1.2294, + "step": 87 + }, + { + "epoch": 0.16, + "grad_norm": 0.1791029365673714, + "learning_rate": 4.292682926829269e-05, + "loss": 1.2895, + "step": 88 + }, + { + "epoch": 0.16, + "grad_norm": 0.20259974283859655, + "learning_rate": 4.341463414634147e-05, + "loss": 1.1841, + "step": 89 + }, + { + "epoch": 0.16, + "grad_norm": 0.17457456183272174, + "learning_rate": 4.390243902439025e-05, + "loss": 1.2357, + "step": 90 + }, + { + "epoch": 0.17, + "grad_norm": 0.1815824380789748, + "learning_rate": 4.439024390243903e-05, + "loss": 1.2304, + "step": 91 + }, + { + "epoch": 0.17, + "grad_norm": 0.17566480599583392, + "learning_rate": 4.4878048780487814e-05, + "loss": 1.242, + "step": 92 + }, + { + "epoch": 0.17, + "grad_norm": 0.18422975005984474, + "learning_rate": 4.536585365853658e-05, + "loss": 1.2177, + "step": 93 + }, + { + "epoch": 0.17, + "grad_norm": 0.16796781877940678, + "learning_rate": 4.5853658536585366e-05, + "loss": 1.1482, + "step": 94 + }, + { + "epoch": 0.17, + "grad_norm": 0.18636131653783305, + "learning_rate": 4.634146341463415e-05, + "loss": 1.1758, + "step": 95 + }, + { + "epoch": 0.18, + "grad_norm": 0.1823665700289814, + "learning_rate": 4.6829268292682926e-05, + "loss": 1.289, + "step": 96 + }, + { + "epoch": 0.18, + "grad_norm": 0.1719900691262439, + "learning_rate": 4.731707317073171e-05, + "loss": 1.1626, + "step": 97 + }, + { + "epoch": 0.18, + "grad_norm": 0.17937994168039778, + "learning_rate": 4.780487804878049e-05, + "loss": 1.175, + "step": 98 + }, + { + "epoch": 0.18, + "grad_norm": 0.16631851422106986, + "learning_rate": 4.829268292682927e-05, + "loss": 1.2177, + "step": 99 + }, + { + "epoch": 0.18, + "grad_norm": 0.19143696232800309, + "learning_rate": 4.878048780487805e-05, + "loss": 1.3071, + "step": 100 + }, + { + "epoch": 0.18, + "grad_norm": 0.17859506638780318, + "learning_rate": 4.9268292682926835e-05, + "loss": 1.2351, + "step": 101 + }, + { + "epoch": 0.19, + "grad_norm": 0.18381520321248196, + "learning_rate": 4.975609756097561e-05, + "loss": 1.2342, + "step": 102 + }, + { + "epoch": 0.19, + "grad_norm": 0.17968218683773912, + "learning_rate": 5.0243902439024394e-05, + "loss": 1.2074, + "step": 103 + }, + { + "epoch": 0.19, + "grad_norm": 0.18139489969339018, + "learning_rate": 5.073170731707318e-05, + "loss": 1.1558, + "step": 104 + }, + { + "epoch": 0.19, + "grad_norm": 0.17366624842514394, + "learning_rate": 5.121951219512195e-05, + "loss": 1.1897, + "step": 105 + }, + { + "epoch": 0.19, + "grad_norm": 0.16034845455223745, + "learning_rate": 5.1707317073170736e-05, + "loss": 1.179, + "step": 106 + }, + { + "epoch": 0.2, + "grad_norm": 0.17583069577827776, + "learning_rate": 5.219512195121952e-05, + "loss": 1.1856, + "step": 107 + }, + { + "epoch": 0.2, + "grad_norm": 0.1853758076989552, + "learning_rate": 5.26829268292683e-05, + "loss": 1.2072, + "step": 108 + }, + { + "epoch": 0.2, + "grad_norm": 0.19597443965936462, + "learning_rate": 5.317073170731708e-05, + "loss": 1.2271, + "step": 109 + }, + { + "epoch": 0.2, + "grad_norm": 0.1899206334098331, + "learning_rate": 5.365853658536586e-05, + "loss": 1.1961, + "step": 110 + }, + { + "epoch": 0.2, + "grad_norm": 0.17463763837757018, + "learning_rate": 5.4146341463414645e-05, + "loss": 1.2049, + "step": 111 + }, + { + "epoch": 0.2, + "grad_norm": 0.20431371701229986, + "learning_rate": 5.463414634146342e-05, + "loss": 1.2891, + "step": 112 + }, + { + "epoch": 0.21, + "grad_norm": 0.1814475107638498, + "learning_rate": 5.51219512195122e-05, + "loss": 1.2346, + "step": 113 + }, + { + "epoch": 0.21, + "grad_norm": 0.1883849423207823, + "learning_rate": 5.5609756097560974e-05, + "loss": 1.244, + "step": 114 + }, + { + "epoch": 0.21, + "grad_norm": 0.1857258128640568, + "learning_rate": 5.609756097560976e-05, + "loss": 1.2669, + "step": 115 + }, + { + "epoch": 0.21, + "grad_norm": 0.1740768514118401, + "learning_rate": 5.658536585365854e-05, + "loss": 1.2414, + "step": 116 + }, + { + "epoch": 0.21, + "grad_norm": 0.1919320335584178, + "learning_rate": 5.7073170731707317e-05, + "loss": 1.2886, + "step": 117 + }, + { + "epoch": 0.22, + "grad_norm": 0.18288775167828136, + "learning_rate": 5.75609756097561e-05, + "loss": 1.1875, + "step": 118 + }, + { + "epoch": 0.22, + "grad_norm": 0.18208588867750863, + "learning_rate": 5.804878048780488e-05, + "loss": 1.2388, + "step": 119 + }, + { + "epoch": 0.22, + "grad_norm": 0.1743260015658331, + "learning_rate": 5.853658536585366e-05, + "loss": 1.1762, + "step": 120 + }, + { + "epoch": 0.22, + "grad_norm": 0.17856046291517946, + "learning_rate": 5.902439024390244e-05, + "loss": 1.2888, + "step": 121 + }, + { + "epoch": 0.22, + "grad_norm": 0.17493794870966536, + "learning_rate": 5.9512195121951225e-05, + "loss": 1.2222, + "step": 122 + }, + { + "epoch": 0.22, + "grad_norm": 0.1909202655203384, + "learning_rate": 6.000000000000001e-05, + "loss": 1.2414, + "step": 123 + }, + { + "epoch": 0.23, + "grad_norm": 0.18345819482834988, + "learning_rate": 6.0487804878048785e-05, + "loss": 1.2756, + "step": 124 + }, + { + "epoch": 0.23, + "grad_norm": 0.2057069352956621, + "learning_rate": 6.097560975609757e-05, + "loss": 1.261, + "step": 125 + }, + { + "epoch": 0.23, + "grad_norm": 0.299775882469108, + "learning_rate": 6.146341463414634e-05, + "loss": 1.2566, + "step": 126 + }, + { + "epoch": 0.23, + "grad_norm": 0.1869687633018095, + "learning_rate": 6.195121951219513e-05, + "loss": 1.3039, + "step": 127 + }, + { + "epoch": 0.23, + "grad_norm": 0.17747149926197442, + "learning_rate": 6.243902439024391e-05, + "loss": 1.2524, + "step": 128 + }, + { + "epoch": 0.24, + "grad_norm": 0.17885157788044242, + "learning_rate": 6.29268292682927e-05, + "loss": 1.2455, + "step": 129 + }, + { + "epoch": 0.24, + "grad_norm": 0.17617298187845123, + "learning_rate": 6.341463414634148e-05, + "loss": 1.2009, + "step": 130 + }, + { + "epoch": 0.24, + "grad_norm": 0.20164176323497066, + "learning_rate": 6.390243902439025e-05, + "loss": 1.2634, + "step": 131 + }, + { + "epoch": 0.24, + "grad_norm": 0.20459903417307612, + "learning_rate": 6.439024390243903e-05, + "loss": 1.1963, + "step": 132 + }, + { + "epoch": 0.24, + "grad_norm": 0.1863755486334296, + "learning_rate": 6.487804878048781e-05, + "loss": 1.2387, + "step": 133 + }, + { + "epoch": 0.25, + "grad_norm": 0.19265866140295207, + "learning_rate": 6.536585365853658e-05, + "loss": 1.2688, + "step": 134 + }, + { + "epoch": 0.25, + "grad_norm": 0.1823425868969493, + "learning_rate": 6.585365853658536e-05, + "loss": 1.2041, + "step": 135 + }, + { + "epoch": 0.25, + "grad_norm": 0.2016853266472781, + "learning_rate": 6.634146341463415e-05, + "loss": 1.1223, + "step": 136 + }, + { + "epoch": 0.25, + "grad_norm": 0.17282675192463448, + "learning_rate": 6.682926829268293e-05, + "loss": 1.1879, + "step": 137 + }, + { + "epoch": 0.25, + "grad_norm": 0.17398811693399288, + "learning_rate": 6.731707317073171e-05, + "loss": 1.2682, + "step": 138 + }, + { + "epoch": 0.25, + "grad_norm": 0.18516916965434696, + "learning_rate": 6.78048780487805e-05, + "loss": 1.1666, + "step": 139 + }, + { + "epoch": 0.26, + "grad_norm": 0.1852213129647933, + "learning_rate": 6.829268292682927e-05, + "loss": 1.2501, + "step": 140 + }, + { + "epoch": 0.26, + "grad_norm": 0.17915948766591883, + "learning_rate": 6.878048780487805e-05, + "loss": 1.2264, + "step": 141 + }, + { + "epoch": 0.26, + "grad_norm": 0.21599939417233183, + "learning_rate": 6.926829268292683e-05, + "loss": 1.2376, + "step": 142 + }, + { + "epoch": 0.26, + "grad_norm": 0.17839304459521851, + "learning_rate": 6.975609756097562e-05, + "loss": 1.2353, + "step": 143 + }, + { + "epoch": 0.26, + "grad_norm": 0.20826913231380875, + "learning_rate": 7.02439024390244e-05, + "loss": 1.1901, + "step": 144 + }, + { + "epoch": 0.27, + "grad_norm": 0.20788894913361589, + "learning_rate": 7.073170731707318e-05, + "loss": 1.2577, + "step": 145 + }, + { + "epoch": 0.27, + "grad_norm": 0.18420055842301297, + "learning_rate": 7.121951219512195e-05, + "loss": 1.1393, + "step": 146 + }, + { + "epoch": 0.27, + "grad_norm": 0.19903048468685589, + "learning_rate": 7.170731707317073e-05, + "loss": 1.2321, + "step": 147 + }, + { + "epoch": 0.27, + "grad_norm": 0.19074116314985748, + "learning_rate": 7.219512195121952e-05, + "loss": 1.1912, + "step": 148 + }, + { + "epoch": 0.27, + "grad_norm": 0.2353816469403903, + "learning_rate": 7.26829268292683e-05, + "loss": 1.28, + "step": 149 + }, + { + "epoch": 0.27, + "grad_norm": 0.21634875684769345, + "learning_rate": 7.317073170731708e-05, + "loss": 1.3312, + "step": 150 + }, + { + "epoch": 0.28, + "grad_norm": 0.18290969006743918, + "learning_rate": 7.365853658536587e-05, + "loss": 1.2214, + "step": 151 + }, + { + "epoch": 0.28, + "grad_norm": 0.18484243897545208, + "learning_rate": 7.414634146341465e-05, + "loss": 1.1895, + "step": 152 + }, + { + "epoch": 0.28, + "grad_norm": 0.21882343112978872, + "learning_rate": 7.463414634146342e-05, + "loss": 1.2219, + "step": 153 + }, + { + "epoch": 0.28, + "grad_norm": 0.19868284379241205, + "learning_rate": 7.51219512195122e-05, + "loss": 1.2176, + "step": 154 + }, + { + "epoch": 0.28, + "grad_norm": 0.20912516312950613, + "learning_rate": 7.560975609756097e-05, + "loss": 1.242, + "step": 155 + }, + { + "epoch": 0.29, + "grad_norm": 0.23811880045549916, + "learning_rate": 7.609756097560976e-05, + "loss": 1.2838, + "step": 156 + }, + { + "epoch": 0.29, + "grad_norm": 0.19511077122033713, + "learning_rate": 7.658536585365854e-05, + "loss": 1.1594, + "step": 157 + }, + { + "epoch": 0.29, + "grad_norm": 0.20094129399534238, + "learning_rate": 7.707317073170732e-05, + "loss": 1.2966, + "step": 158 + }, + { + "epoch": 0.29, + "grad_norm": 0.19366245038292418, + "learning_rate": 7.75609756097561e-05, + "loss": 1.2246, + "step": 159 + }, + { + "epoch": 0.29, + "grad_norm": 0.19409570223867306, + "learning_rate": 7.804878048780489e-05, + "loss": 1.2312, + "step": 160 + }, + { + "epoch": 0.29, + "grad_norm": 0.2087258457033805, + "learning_rate": 7.853658536585366e-05, + "loss": 1.2169, + "step": 161 + }, + { + "epoch": 0.3, + "grad_norm": 0.18765223996270428, + "learning_rate": 7.902439024390244e-05, + "loss": 1.2383, + "step": 162 + }, + { + "epoch": 0.3, + "grad_norm": 0.20734180224147242, + "learning_rate": 7.951219512195122e-05, + "loss": 1.2587, + "step": 163 + }, + { + "epoch": 0.3, + "grad_norm": 0.24690929540287834, + "learning_rate": 8e-05, + "loss": 1.1951, + "step": 164 + }, + { + "epoch": 0.3, + "grad_norm": 0.2003538797619543, + "learning_rate": 7.999990914797545e-05, + "loss": 1.1982, + "step": 165 + }, + { + "epoch": 0.3, + "grad_norm": 0.22469075613510484, + "learning_rate": 7.99996365923145e-05, + "loss": 1.2355, + "step": 166 + }, + { + "epoch": 0.31, + "grad_norm": 0.21870100788336058, + "learning_rate": 7.999918233425526e-05, + "loss": 1.1103, + "step": 167 + }, + { + "epoch": 0.31, + "grad_norm": 0.20939989594131886, + "learning_rate": 7.999854637586122e-05, + "loss": 1.1966, + "step": 168 + }, + { + "epoch": 0.31, + "grad_norm": 0.43108211416237796, + "learning_rate": 7.999772872002132e-05, + "loss": 1.2882, + "step": 169 + }, + { + "epoch": 0.31, + "grad_norm": 0.27045413432174487, + "learning_rate": 7.999672937044984e-05, + "loss": 1.2399, + "step": 170 + }, + { + "epoch": 0.31, + "grad_norm": 0.19700483036740515, + "learning_rate": 7.999554833168642e-05, + "loss": 1.202, + "step": 171 + }, + { + "epoch": 0.31, + "grad_norm": 0.3335979493370708, + "learning_rate": 7.999418560909604e-05, + "loss": 1.1995, + "step": 172 + }, + { + "epoch": 0.32, + "grad_norm": 0.3165803974474567, + "learning_rate": 7.999264120886902e-05, + "loss": 1.1569, + "step": 173 + }, + { + "epoch": 0.32, + "grad_norm": 0.1951699080346223, + "learning_rate": 7.999091513802093e-05, + "loss": 1.1778, + "step": 174 + }, + { + "epoch": 0.32, + "grad_norm": 0.2087559121749787, + "learning_rate": 7.998900740439265e-05, + "loss": 1.1736, + "step": 175 + }, + { + "epoch": 0.32, + "grad_norm": 0.20345180977460478, + "learning_rate": 7.998691801665024e-05, + "loss": 1.2281, + "step": 176 + }, + { + "epoch": 0.32, + "grad_norm": 0.24617644827252333, + "learning_rate": 7.998464698428495e-05, + "loss": 1.2072, + "step": 177 + }, + { + "epoch": 0.33, + "grad_norm": 0.2469050959356265, + "learning_rate": 7.998219431761318e-05, + "loss": 1.2242, + "step": 178 + }, + { + "epoch": 0.33, + "grad_norm": 0.19529317748460623, + "learning_rate": 7.997956002777642e-05, + "loss": 1.2567, + "step": 179 + }, + { + "epoch": 0.33, + "grad_norm": 0.19048389491381376, + "learning_rate": 7.99767441267412e-05, + "loss": 1.2982, + "step": 180 + }, + { + "epoch": 0.33, + "grad_norm": 0.2085799116493225, + "learning_rate": 7.997374662729904e-05, + "loss": 1.1254, + "step": 181 + }, + { + "epoch": 0.33, + "grad_norm": 0.20636853256378995, + "learning_rate": 7.997056754306636e-05, + "loss": 1.2435, + "step": 182 + }, + { + "epoch": 0.33, + "grad_norm": 0.20590016382290252, + "learning_rate": 7.99672068884845e-05, + "loss": 1.2658, + "step": 183 + }, + { + "epoch": 0.34, + "grad_norm": 0.1931166169764433, + "learning_rate": 7.996366467881955e-05, + "loss": 1.1637, + "step": 184 + }, + { + "epoch": 0.34, + "grad_norm": 0.18873318157988098, + "learning_rate": 7.995994093016237e-05, + "loss": 1.1335, + "step": 185 + }, + { + "epoch": 0.34, + "grad_norm": 0.19210254625199108, + "learning_rate": 7.995603565942846e-05, + "loss": 1.1928, + "step": 186 + }, + { + "epoch": 0.34, + "grad_norm": 0.2130986479765664, + "learning_rate": 7.995194888435792e-05, + "loss": 1.2158, + "step": 187 + }, + { + "epoch": 0.34, + "grad_norm": 0.22003854501814088, + "learning_rate": 7.994768062351532e-05, + "loss": 1.2288, + "step": 188 + }, + { + "epoch": 0.35, + "grad_norm": 0.20330803191993058, + "learning_rate": 7.994323089628968e-05, + "loss": 1.2426, + "step": 189 + }, + { + "epoch": 0.35, + "grad_norm": 0.20567314642208634, + "learning_rate": 7.993859972289434e-05, + "loss": 1.2649, + "step": 190 + }, + { + "epoch": 0.35, + "grad_norm": 0.21556663727342962, + "learning_rate": 7.993378712436686e-05, + "loss": 1.2545, + "step": 191 + }, + { + "epoch": 0.35, + "grad_norm": 0.20309165469109888, + "learning_rate": 7.992879312256897e-05, + "loss": 1.3338, + "step": 192 + }, + { + "epoch": 0.35, + "grad_norm": 0.19574356669421325, + "learning_rate": 7.992361774018641e-05, + "loss": 1.278, + "step": 193 + }, + { + "epoch": 0.35, + "grad_norm": 0.2763613746722313, + "learning_rate": 7.991826100072891e-05, + "loss": 1.2571, + "step": 194 + }, + { + "epoch": 0.36, + "grad_norm": 0.19346552479915102, + "learning_rate": 7.991272292852996e-05, + "loss": 1.2027, + "step": 195 + }, + { + "epoch": 0.36, + "grad_norm": 0.2281167812123908, + "learning_rate": 7.990700354874683e-05, + "loss": 1.2586, + "step": 196 + }, + { + "epoch": 0.36, + "grad_norm": 0.19699013712137542, + "learning_rate": 7.990110288736042e-05, + "loss": 1.1371, + "step": 197 + }, + { + "epoch": 0.36, + "grad_norm": 0.21768209981475933, + "learning_rate": 7.989502097117503e-05, + "loss": 1.2522, + "step": 198 + }, + { + "epoch": 0.36, + "grad_norm": 0.21335427847754582, + "learning_rate": 7.988875782781838e-05, + "loss": 1.2437, + "step": 199 + }, + { + "epoch": 0.37, + "grad_norm": 0.21856710629066897, + "learning_rate": 7.988231348574147e-05, + "loss": 1.2135, + "step": 200 + }, + { + "epoch": 0.37, + "grad_norm": 0.20482062658774797, + "learning_rate": 7.987568797421836e-05, + "loss": 1.1755, + "step": 201 + }, + { + "epoch": 0.37, + "grad_norm": 0.2017756813960897, + "learning_rate": 7.986888132334608e-05, + "loss": 1.1699, + "step": 202 + }, + { + "epoch": 0.37, + "grad_norm": 0.20496443848153809, + "learning_rate": 7.986189356404458e-05, + "loss": 1.2125, + "step": 203 + }, + { + "epoch": 0.37, + "grad_norm": 0.2134603800558358, + "learning_rate": 7.985472472805643e-05, + "loss": 1.2391, + "step": 204 + }, + { + "epoch": 0.37, + "grad_norm": 0.2364175573420861, + "learning_rate": 7.98473748479468e-05, + "loss": 1.2384, + "step": 205 + }, + { + "epoch": 0.38, + "grad_norm": 0.1872419861598724, + "learning_rate": 7.983984395710326e-05, + "loss": 1.1457, + "step": 206 + }, + { + "epoch": 0.38, + "grad_norm": 0.28222194007095774, + "learning_rate": 7.983213208973566e-05, + "loss": 1.2952, + "step": 207 + }, + { + "epoch": 0.38, + "grad_norm": 0.1916094851162064, + "learning_rate": 7.982423928087593e-05, + "loss": 1.1763, + "step": 208 + }, + { + "epoch": 0.38, + "grad_norm": 0.18446245256166657, + "learning_rate": 7.981616556637795e-05, + "loss": 1.1863, + "step": 209 + }, + { + "epoch": 0.38, + "grad_norm": 0.195191961022491, + "learning_rate": 7.980791098291737e-05, + "loss": 1.2036, + "step": 210 + }, + { + "epoch": 0.39, + "grad_norm": 0.2652439657825496, + "learning_rate": 7.979947556799151e-05, + "loss": 1.2834, + "step": 211 + }, + { + "epoch": 0.39, + "grad_norm": 0.24308438957843412, + "learning_rate": 7.979085935991906e-05, + "loss": 1.234, + "step": 212 + }, + { + "epoch": 0.39, + "grad_norm": 0.21294701043622016, + "learning_rate": 7.978206239784004e-05, + "loss": 1.3006, + "step": 213 + }, + { + "epoch": 0.39, + "grad_norm": 0.25809277041859524, + "learning_rate": 7.977308472171553e-05, + "loss": 1.2272, + "step": 214 + }, + { + "epoch": 0.39, + "grad_norm": 0.193463860107294, + "learning_rate": 7.976392637232754e-05, + "loss": 1.2295, + "step": 215 + }, + { + "epoch": 0.4, + "grad_norm": 0.2150023760609626, + "learning_rate": 7.975458739127877e-05, + "loss": 1.2135, + "step": 216 + }, + { + "epoch": 0.4, + "grad_norm": 0.22590495955605894, + "learning_rate": 7.974506782099253e-05, + "loss": 1.2532, + "step": 217 + }, + { + "epoch": 0.4, + "grad_norm": 0.21023744668403702, + "learning_rate": 7.973536770471242e-05, + "loss": 1.2472, + "step": 218 + }, + { + "epoch": 0.4, + "grad_norm": 0.2345749799511543, + "learning_rate": 7.972548708650218e-05, + "loss": 1.1791, + "step": 219 + }, + { + "epoch": 0.4, + "grad_norm": 0.2158876734005217, + "learning_rate": 7.971542601124553e-05, + "loss": 1.2483, + "step": 220 + }, + { + "epoch": 0.4, + "grad_norm": 0.29455339949432446, + "learning_rate": 7.970518452464593e-05, + "loss": 1.2894, + "step": 221 + }, + { + "epoch": 0.41, + "grad_norm": 0.23983708730626851, + "learning_rate": 7.969476267322636e-05, + "loss": 1.271, + "step": 222 + }, + { + "epoch": 0.41, + "grad_norm": 0.1922400905426158, + "learning_rate": 7.968416050432912e-05, + "loss": 1.2139, + "step": 223 + }, + { + "epoch": 0.41, + "grad_norm": 0.2238136844422931, + "learning_rate": 7.967337806611568e-05, + "loss": 1.2655, + "step": 224 + }, + { + "epoch": 0.41, + "grad_norm": 0.21230292828267672, + "learning_rate": 7.966241540756631e-05, + "loss": 1.2406, + "step": 225 + }, + { + "epoch": 0.41, + "grad_norm": 0.26656119419070456, + "learning_rate": 7.965127257848004e-05, + "loss": 1.2595, + "step": 226 + }, + { + "epoch": 0.42, + "grad_norm": 0.22381385502992684, + "learning_rate": 7.963994962947426e-05, + "loss": 1.1737, + "step": 227 + }, + { + "epoch": 0.42, + "grad_norm": 0.20056702203994298, + "learning_rate": 7.962844661198462e-05, + "loss": 1.1969, + "step": 228 + }, + { + "epoch": 0.42, + "grad_norm": 0.20148701321526885, + "learning_rate": 7.961676357826478e-05, + "loss": 1.2151, + "step": 229 + }, + { + "epoch": 0.42, + "grad_norm": 0.20034834807028637, + "learning_rate": 7.960490058138604e-05, + "loss": 1.1455, + "step": 230 + }, + { + "epoch": 0.42, + "grad_norm": 0.21050838521846033, + "learning_rate": 7.959285767523732e-05, + "loss": 1.2223, + "step": 231 + }, + { + "epoch": 0.42, + "grad_norm": 0.20904772138969777, + "learning_rate": 7.95806349145247e-05, + "loss": 1.2534, + "step": 232 + }, + { + "epoch": 0.43, + "grad_norm": 0.20307877304792957, + "learning_rate": 7.956823235477134e-05, + "loss": 1.1352, + "step": 233 + }, + { + "epoch": 0.43, + "grad_norm": 0.20501105270897094, + "learning_rate": 7.95556500523171e-05, + "loss": 1.2031, + "step": 234 + }, + { + "epoch": 0.43, + "grad_norm": 0.19800586972038586, + "learning_rate": 7.954288806431838e-05, + "loss": 1.2567, + "step": 235 + }, + { + "epoch": 0.43, + "grad_norm": 0.2175102450594135, + "learning_rate": 7.952994644874777e-05, + "loss": 1.2538, + "step": 236 + }, + { + "epoch": 0.43, + "grad_norm": 0.22698189300067595, + "learning_rate": 7.951682526439391e-05, + "loss": 1.3088, + "step": 237 + }, + { + "epoch": 0.44, + "grad_norm": 0.19208392014975315, + "learning_rate": 7.950352457086109e-05, + "loss": 1.2336, + "step": 238 + }, + { + "epoch": 0.44, + "grad_norm": 0.27004086334319655, + "learning_rate": 7.949004442856905e-05, + "loss": 1.2012, + "step": 239 + }, + { + "epoch": 0.44, + "grad_norm": 0.23420974954538043, + "learning_rate": 7.947638489875272e-05, + "loss": 1.2244, + "step": 240 + }, + { + "epoch": 0.44, + "grad_norm": 0.20514399124802024, + "learning_rate": 7.946254604346186e-05, + "loss": 1.2548, + "step": 241 + }, + { + "epoch": 0.44, + "grad_norm": 0.19334973602372896, + "learning_rate": 7.944852792556092e-05, + "loss": 1.2104, + "step": 242 + }, + { + "epoch": 0.44, + "grad_norm": 0.1992640714537956, + "learning_rate": 7.943433060872858e-05, + "loss": 1.2628, + "step": 243 + }, + { + "epoch": 0.45, + "grad_norm": 0.203284617090413, + "learning_rate": 7.941995415745761e-05, + "loss": 1.2002, + "step": 244 + }, + { + "epoch": 0.45, + "grad_norm": 0.22795306969682058, + "learning_rate": 7.94053986370545e-05, + "loss": 1.2215, + "step": 245 + }, + { + "epoch": 0.45, + "grad_norm": 0.20789041346838505, + "learning_rate": 7.939066411363915e-05, + "loss": 1.0998, + "step": 246 + }, + { + "epoch": 0.45, + "grad_norm": 0.22354868884742066, + "learning_rate": 7.937575065414464e-05, + "loss": 1.2564, + "step": 247 + }, + { + "epoch": 0.45, + "grad_norm": 0.21176392726647736, + "learning_rate": 7.936065832631687e-05, + "loss": 1.2816, + "step": 248 + }, + { + "epoch": 0.46, + "grad_norm": 0.19967179557235587, + "learning_rate": 7.934538719871427e-05, + "loss": 1.1961, + "step": 249 + }, + { + "epoch": 0.46, + "grad_norm": 0.210819577350627, + "learning_rate": 7.932993734070747e-05, + "loss": 1.2167, + "step": 250 + }, + { + "epoch": 0.46, + "grad_norm": 0.21537794551756187, + "learning_rate": 7.931430882247903e-05, + "loss": 1.2341, + "step": 251 + }, + { + "epoch": 0.46, + "grad_norm": 0.22850872387256574, + "learning_rate": 7.929850171502304e-05, + "loss": 1.1686, + "step": 252 + }, + { + "epoch": 0.46, + "grad_norm": 0.22380366415076383, + "learning_rate": 7.928251609014493e-05, + "loss": 1.1462, + "step": 253 + }, + { + "epoch": 0.46, + "grad_norm": 0.22426923149036065, + "learning_rate": 7.926635202046102e-05, + "loss": 1.1792, + "step": 254 + }, + { + "epoch": 0.47, + "grad_norm": 0.42082703321103965, + "learning_rate": 7.925000957939822e-05, + "loss": 1.2718, + "step": 255 + }, + { + "epoch": 0.47, + "grad_norm": 0.2235432774854074, + "learning_rate": 7.92334888411937e-05, + "loss": 1.2598, + "step": 256 + }, + { + "epoch": 0.47, + "grad_norm": 0.281644028934108, + "learning_rate": 7.92167898808946e-05, + "loss": 1.2205, + "step": 257 + }, + { + "epoch": 0.47, + "grad_norm": 0.2037705143888748, + "learning_rate": 7.919991277435763e-05, + "loss": 1.1737, + "step": 258 + }, + { + "epoch": 0.47, + "grad_norm": 0.20917419230028977, + "learning_rate": 7.918285759824879e-05, + "loss": 1.2035, + "step": 259 + }, + { + "epoch": 0.48, + "grad_norm": 0.20510847570635518, + "learning_rate": 7.916562443004292e-05, + "loss": 1.2135, + "step": 260 + }, + { + "epoch": 0.48, + "grad_norm": 0.25172483071092466, + "learning_rate": 7.914821334802342e-05, + "loss": 1.2218, + "step": 261 + }, + { + "epoch": 0.48, + "grad_norm": 0.21102706700634313, + "learning_rate": 7.91306244312819e-05, + "loss": 1.1738, + "step": 262 + }, + { + "epoch": 0.48, + "grad_norm": 0.22626060872645815, + "learning_rate": 7.911285775971781e-05, + "loss": 1.238, + "step": 263 + }, + { + "epoch": 0.48, + "grad_norm": 0.22448567539778486, + "learning_rate": 7.909491341403805e-05, + "loss": 1.2404, + "step": 264 + }, + { + "epoch": 0.48, + "grad_norm": 0.2019099786139193, + "learning_rate": 7.907679147575661e-05, + "loss": 1.213, + "step": 265 + }, + { + "epoch": 0.49, + "grad_norm": 0.24307234839096267, + "learning_rate": 7.905849202719422e-05, + "loss": 1.2322, + "step": 266 + }, + { + "epoch": 0.49, + "grad_norm": 0.19801890521743487, + "learning_rate": 7.904001515147802e-05, + "loss": 1.2448, + "step": 267 + }, + { + "epoch": 0.49, + "grad_norm": 0.2102742273575385, + "learning_rate": 7.902136093254106e-05, + "loss": 1.1657, + "step": 268 + }, + { + "epoch": 0.49, + "grad_norm": 0.2173464476815016, + "learning_rate": 7.900252945512201e-05, + "loss": 1.2549, + "step": 269 + }, + { + "epoch": 0.49, + "grad_norm": 0.20957275458699595, + "learning_rate": 7.898352080476479e-05, + "loss": 1.2536, + "step": 270 + }, + { + "epoch": 0.5, + "grad_norm": 0.20691966388952363, + "learning_rate": 7.896433506781811e-05, + "loss": 1.2661, + "step": 271 + }, + { + "epoch": 0.5, + "grad_norm": 0.2276662275112648, + "learning_rate": 7.894497233143509e-05, + "loss": 1.2409, + "step": 272 + }, + { + "epoch": 0.5, + "grad_norm": 0.23854109569301263, + "learning_rate": 7.892543268357297e-05, + "loss": 1.2681, + "step": 273 + }, + { + "epoch": 0.5, + "grad_norm": 0.2233864156677627, + "learning_rate": 7.890571621299252e-05, + "loss": 1.1687, + "step": 274 + }, + { + "epoch": 0.5, + "grad_norm": 0.20114129147925475, + "learning_rate": 7.888582300925787e-05, + "loss": 1.2184, + "step": 275 + }, + { + "epoch": 0.5, + "grad_norm": 0.2154654670569462, + "learning_rate": 7.886575316273586e-05, + "loss": 1.1982, + "step": 276 + }, + { + "epoch": 0.51, + "grad_norm": 0.2292982209343639, + "learning_rate": 7.884550676459583e-05, + "loss": 1.2129, + "step": 277 + }, + { + "epoch": 0.51, + "grad_norm": 0.21302713135229548, + "learning_rate": 7.882508390680908e-05, + "loss": 1.1605, + "step": 278 + }, + { + "epoch": 0.51, + "grad_norm": 0.2123661020671048, + "learning_rate": 7.88044846821485e-05, + "loss": 1.2308, + "step": 279 + }, + { + "epoch": 0.51, + "grad_norm": 0.2080577410800404, + "learning_rate": 7.878370918418818e-05, + "loss": 1.2195, + "step": 280 + }, + { + "epoch": 0.51, + "grad_norm": 0.19663901881127385, + "learning_rate": 7.876275750730289e-05, + "loss": 1.1591, + "step": 281 + }, + { + "epoch": 0.52, + "grad_norm": 0.20534502031312163, + "learning_rate": 7.874162974666776e-05, + "loss": 1.2664, + "step": 282 + }, + { + "epoch": 0.52, + "grad_norm": 0.23240445399513837, + "learning_rate": 7.872032599825779e-05, + "loss": 1.2151, + "step": 283 + }, + { + "epoch": 0.52, + "grad_norm": 0.2672527316717507, + "learning_rate": 7.86988463588474e-05, + "loss": 1.2406, + "step": 284 + }, + { + "epoch": 0.52, + "grad_norm": 0.19893903058743695, + "learning_rate": 7.867719092601003e-05, + "loss": 1.1291, + "step": 285 + }, + { + "epoch": 0.52, + "grad_norm": 0.33275268109930917, + "learning_rate": 7.865535979811768e-05, + "loss": 1.1406, + "step": 286 + }, + { + "epoch": 0.52, + "grad_norm": 0.2373619455690358, + "learning_rate": 7.863335307434045e-05, + "loss": 1.2799, + "step": 287 + }, + { + "epoch": 0.53, + "grad_norm": 0.263235735390858, + "learning_rate": 7.861117085464612e-05, + "loss": 1.2415, + "step": 288 + }, + { + "epoch": 0.53, + "grad_norm": 0.25884281780784324, + "learning_rate": 7.858881323979965e-05, + "loss": 1.3919, + "step": 289 + }, + { + "epoch": 0.53, + "grad_norm": 0.25426288332255736, + "learning_rate": 7.85662803313628e-05, + "loss": 1.174, + "step": 290 + }, + { + "epoch": 0.53, + "grad_norm": 0.26655405527881243, + "learning_rate": 7.854357223169356e-05, + "loss": 1.2806, + "step": 291 + }, + { + "epoch": 0.53, + "grad_norm": 0.20909844432349833, + "learning_rate": 7.852068904394579e-05, + "loss": 1.2627, + "step": 292 + }, + { + "epoch": 0.54, + "grad_norm": 0.21307115068935759, + "learning_rate": 7.849763087206866e-05, + "loss": 1.1879, + "step": 293 + }, + { + "epoch": 0.54, + "grad_norm": 0.25009949471398946, + "learning_rate": 7.847439782080628e-05, + "loss": 1.2881, + "step": 294 + }, + { + "epoch": 0.54, + "grad_norm": 0.20960783418679174, + "learning_rate": 7.845098999569712e-05, + "loss": 1.2723, + "step": 295 + }, + { + "epoch": 0.54, + "grad_norm": 0.24968832437925104, + "learning_rate": 7.842740750307362e-05, + "loss": 1.2029, + "step": 296 + }, + { + "epoch": 0.54, + "grad_norm": 0.22981196585125677, + "learning_rate": 7.84036504500616e-05, + "loss": 1.1695, + "step": 297 + }, + { + "epoch": 0.55, + "grad_norm": 0.2320606844751365, + "learning_rate": 7.837971894457991e-05, + "loss": 1.2317, + "step": 298 + }, + { + "epoch": 0.55, + "grad_norm": 0.23051459673906124, + "learning_rate": 7.835561309533981e-05, + "loss": 1.2046, + "step": 299 + }, + { + "epoch": 0.55, + "grad_norm": 0.2510027231060586, + "learning_rate": 7.833133301184457e-05, + "loss": 1.199, + "step": 300 + }, + { + "epoch": 0.55, + "grad_norm": 0.23601180466018787, + "learning_rate": 7.830687880438895e-05, + "loss": 1.1755, + "step": 301 + }, + { + "epoch": 0.55, + "grad_norm": 0.24740820934385369, + "learning_rate": 7.828225058405864e-05, + "loss": 1.2054, + "step": 302 + }, + { + "epoch": 0.55, + "grad_norm": 0.23065372979111173, + "learning_rate": 7.825744846272984e-05, + "loss": 1.2066, + "step": 303 + }, + { + "epoch": 0.56, + "grad_norm": 0.22385077334838213, + "learning_rate": 7.823247255306866e-05, + "loss": 1.2147, + "step": 304 + }, + { + "epoch": 0.56, + "grad_norm": 0.42981213948386104, + "learning_rate": 7.820732296853074e-05, + "loss": 1.2314, + "step": 305 + }, + { + "epoch": 0.56, + "grad_norm": 0.21122844902751076, + "learning_rate": 7.818199982336058e-05, + "loss": 1.1462, + "step": 306 + }, + { + "epoch": 0.56, + "grad_norm": 0.23374869692118933, + "learning_rate": 7.815650323259117e-05, + "loss": 1.2051, + "step": 307 + }, + { + "epoch": 0.56, + "grad_norm": 0.21662363795962128, + "learning_rate": 7.813083331204332e-05, + "loss": 1.1575, + "step": 308 + }, + { + "epoch": 0.57, + "grad_norm": 0.2088315773384112, + "learning_rate": 7.810499017832526e-05, + "loss": 1.1316, + "step": 309 + }, + { + "epoch": 0.57, + "grad_norm": 0.2095238410730976, + "learning_rate": 7.807897394883203e-05, + "loss": 1.2087, + "step": 310 + }, + { + "epoch": 0.57, + "grad_norm": 0.22672932127256515, + "learning_rate": 7.805278474174499e-05, + "loss": 1.2512, + "step": 311 + }, + { + "epoch": 0.57, + "grad_norm": 0.21873052340922736, + "learning_rate": 7.802642267603126e-05, + "loss": 1.1909, + "step": 312 + }, + { + "epoch": 0.57, + "grad_norm": 0.219814521916342, + "learning_rate": 7.79998878714432e-05, + "loss": 1.1669, + "step": 313 + }, + { + "epoch": 0.57, + "grad_norm": 0.3049426027257317, + "learning_rate": 7.797318044851786e-05, + "loss": 1.1797, + "step": 314 + }, + { + "epoch": 0.58, + "grad_norm": 0.22309435690065985, + "learning_rate": 7.794630052857638e-05, + "loss": 1.1417, + "step": 315 + }, + { + "epoch": 0.58, + "grad_norm": 0.3891885169154885, + "learning_rate": 7.791924823372354e-05, + "loss": 1.2369, + "step": 316 + }, + { + "epoch": 0.58, + "grad_norm": 0.24780269452456372, + "learning_rate": 7.789202368684711e-05, + "loss": 1.2521, + "step": 317 + }, + { + "epoch": 0.58, + "grad_norm": 0.21660460720269362, + "learning_rate": 7.786462701161738e-05, + "loss": 1.2151, + "step": 318 + }, + { + "epoch": 0.58, + "grad_norm": 0.23635409466561857, + "learning_rate": 7.783705833248649e-05, + "loss": 1.2363, + "step": 319 + }, + { + "epoch": 0.59, + "grad_norm": 0.2616135839903218, + "learning_rate": 7.780931777468797e-05, + "loss": 1.2428, + "step": 320 + }, + { + "epoch": 0.59, + "grad_norm": 0.21461059159245083, + "learning_rate": 7.77814054642361e-05, + "loss": 1.1434, + "step": 321 + }, + { + "epoch": 0.59, + "grad_norm": 0.25348824286656163, + "learning_rate": 7.775332152792539e-05, + "loss": 1.2368, + "step": 322 + }, + { + "epoch": 0.59, + "grad_norm": 0.22275034726331247, + "learning_rate": 7.772506609332995e-05, + "loss": 1.1827, + "step": 323 + }, + { + "epoch": 0.59, + "grad_norm": 0.25030821228147526, + "learning_rate": 7.769663928880298e-05, + "loss": 1.2428, + "step": 324 + }, + { + "epoch": 0.59, + "grad_norm": 0.22251804398745534, + "learning_rate": 7.766804124347608e-05, + "loss": 1.1889, + "step": 325 + }, + { + "epoch": 0.6, + "grad_norm": 0.23381455520411995, + "learning_rate": 7.763927208725879e-05, + "loss": 1.2115, + "step": 326 + }, + { + "epoch": 0.6, + "grad_norm": 0.27341902651946226, + "learning_rate": 7.761033195083791e-05, + "loss": 1.2535, + "step": 327 + }, + { + "epoch": 0.6, + "grad_norm": 0.24862471659814522, + "learning_rate": 7.758122096567694e-05, + "loss": 1.2128, + "step": 328 + }, + { + "epoch": 0.6, + "grad_norm": 0.2251357082045494, + "learning_rate": 7.755193926401547e-05, + "loss": 1.2334, + "step": 329 + }, + { + "epoch": 0.6, + "grad_norm": 0.3173274941622932, + "learning_rate": 7.752248697886857e-05, + "loss": 1.226, + "step": 330 + }, + { + "epoch": 0.61, + "grad_norm": 0.23056440717672175, + "learning_rate": 7.74928642440263e-05, + "loss": 1.2339, + "step": 331 + }, + { + "epoch": 0.61, + "grad_norm": 0.2801507500859342, + "learning_rate": 7.746307119405286e-05, + "loss": 1.287, + "step": 332 + }, + { + "epoch": 0.61, + "grad_norm": 0.2267818430426272, + "learning_rate": 7.743310796428622e-05, + "loss": 1.1916, + "step": 333 + }, + { + "epoch": 0.61, + "grad_norm": 0.2777329160365585, + "learning_rate": 7.74029746908374e-05, + "loss": 1.252, + "step": 334 + }, + { + "epoch": 0.61, + "grad_norm": 0.25289169762353, + "learning_rate": 7.737267151058983e-05, + "loss": 1.2153, + "step": 335 + }, + { + "epoch": 0.61, + "grad_norm": 0.2424670686901653, + "learning_rate": 7.734219856119875e-05, + "loss": 1.2227, + "step": 336 + }, + { + "epoch": 0.62, + "grad_norm": 0.22747092217441645, + "learning_rate": 7.731155598109067e-05, + "loss": 1.19, + "step": 337 + }, + { + "epoch": 0.62, + "grad_norm": 0.2307810940100189, + "learning_rate": 7.728074390946257e-05, + "loss": 1.1818, + "step": 338 + }, + { + "epoch": 0.62, + "grad_norm": 0.2583402574655623, + "learning_rate": 7.724976248628142e-05, + "loss": 1.1608, + "step": 339 + }, + { + "epoch": 0.62, + "grad_norm": 0.22140209760890694, + "learning_rate": 7.721861185228347e-05, + "loss": 1.1245, + "step": 340 + }, + { + "epoch": 0.62, + "grad_norm": 0.25859310758244686, + "learning_rate": 7.718729214897362e-05, + "loss": 1.2247, + "step": 341 + }, + { + "epoch": 0.63, + "grad_norm": 0.26371179531372124, + "learning_rate": 7.715580351862482e-05, + "loss": 1.2128, + "step": 342 + }, + { + "epoch": 0.63, + "grad_norm": 0.26575541302851047, + "learning_rate": 7.712414610427733e-05, + "loss": 1.2443, + "step": 343 + }, + { + "epoch": 0.63, + "grad_norm": 0.269978305197599, + "learning_rate": 7.709232004973816e-05, + "loss": 1.2231, + "step": 344 + }, + { + "epoch": 0.63, + "grad_norm": 0.26583998705977047, + "learning_rate": 7.70603254995804e-05, + "loss": 1.2476, + "step": 345 + }, + { + "epoch": 0.63, + "grad_norm": 0.24256062164066097, + "learning_rate": 7.702816259914253e-05, + "loss": 1.2901, + "step": 346 + }, + { + "epoch": 0.63, + "grad_norm": 0.3463123472658915, + "learning_rate": 7.699583149452779e-05, + "loss": 1.3277, + "step": 347 + }, + { + "epoch": 0.64, + "grad_norm": 0.2269096590531878, + "learning_rate": 7.696333233260345e-05, + "loss": 1.2047, + "step": 348 + }, + { + "epoch": 0.64, + "grad_norm": 0.25136883001050025, + "learning_rate": 7.693066526100031e-05, + "loss": 1.1619, + "step": 349 + }, + { + "epoch": 0.64, + "grad_norm": 0.2565112571116145, + "learning_rate": 7.68978304281118e-05, + "loss": 1.2389, + "step": 350 + }, + { + "epoch": 0.64, + "grad_norm": 0.22175779550828703, + "learning_rate": 7.686482798309349e-05, + "loss": 1.2238, + "step": 351 + }, + { + "epoch": 0.64, + "grad_norm": 0.22588304332216555, + "learning_rate": 7.683165807586234e-05, + "loss": 1.174, + "step": 352 + }, + { + "epoch": 0.65, + "grad_norm": 0.24889474296529737, + "learning_rate": 7.6798320857096e-05, + "loss": 1.2366, + "step": 353 + }, + { + "epoch": 0.65, + "grad_norm": 0.27339703806525034, + "learning_rate": 7.676481647823214e-05, + "loss": 1.2356, + "step": 354 + }, + { + "epoch": 0.65, + "grad_norm": 0.23424666722888365, + "learning_rate": 7.673114509146782e-05, + "loss": 1.2089, + "step": 355 + }, + { + "epoch": 0.65, + "grad_norm": 0.27978285392461766, + "learning_rate": 7.66973068497587e-05, + "loss": 1.2609, + "step": 356 + }, + { + "epoch": 0.65, + "grad_norm": 0.2509423350138824, + "learning_rate": 7.666330190681844e-05, + "loss": 1.1777, + "step": 357 + }, + { + "epoch": 0.65, + "grad_norm": 0.23007730927468031, + "learning_rate": 7.662913041711793e-05, + "loss": 1.154, + "step": 358 + }, + { + "epoch": 0.66, + "grad_norm": 0.2438648674953112, + "learning_rate": 7.659479253588462e-05, + "loss": 1.2257, + "step": 359 + }, + { + "epoch": 0.66, + "grad_norm": 0.28816093242092233, + "learning_rate": 7.65602884191018e-05, + "loss": 1.2558, + "step": 360 + }, + { + "epoch": 0.66, + "grad_norm": 0.24972815300596035, + "learning_rate": 7.652561822350793e-05, + "loss": 1.2837, + "step": 361 + }, + { + "epoch": 0.66, + "grad_norm": 0.2543189139697063, + "learning_rate": 7.649078210659587e-05, + "loss": 1.2193, + "step": 362 + }, + { + "epoch": 0.66, + "grad_norm": 0.2237937956718952, + "learning_rate": 7.645578022661224e-05, + "loss": 1.2237, + "step": 363 + }, + { + "epoch": 0.67, + "grad_norm": 0.29742029408787396, + "learning_rate": 7.642061274255657e-05, + "loss": 1.2116, + "step": 364 + }, + { + "epoch": 0.67, + "grad_norm": 0.2462883147335493, + "learning_rate": 7.638527981418075e-05, + "loss": 1.1827, + "step": 365 + }, + { + "epoch": 0.67, + "grad_norm": 0.2647802498907096, + "learning_rate": 7.634978160198817e-05, + "loss": 1.2739, + "step": 366 + }, + { + "epoch": 0.67, + "grad_norm": 0.22360398779217264, + "learning_rate": 7.631411826723306e-05, + "loss": 1.2185, + "step": 367 + }, + { + "epoch": 0.67, + "grad_norm": 0.2635048004593543, + "learning_rate": 7.627828997191973e-05, + "loss": 1.2317, + "step": 368 + }, + { + "epoch": 0.67, + "grad_norm": 0.2764803449917684, + "learning_rate": 7.624229687880184e-05, + "loss": 1.1923, + "step": 369 + }, + { + "epoch": 0.68, + "grad_norm": 0.25724943233414527, + "learning_rate": 7.620613915138166e-05, + "loss": 1.2218, + "step": 370 + }, + { + "epoch": 0.68, + "grad_norm": 0.2858318045794755, + "learning_rate": 7.61698169539093e-05, + "loss": 1.1496, + "step": 371 + }, + { + "epoch": 0.68, + "grad_norm": 0.23547216647460364, + "learning_rate": 7.613333045138206e-05, + "loss": 1.1905, + "step": 372 + }, + { + "epoch": 0.68, + "grad_norm": 0.22984814903684375, + "learning_rate": 7.609667980954355e-05, + "loss": 1.2009, + "step": 373 + }, + { + "epoch": 0.68, + "grad_norm": 0.2551903754079084, + "learning_rate": 7.605986519488301e-05, + "loss": 1.2042, + "step": 374 + }, + { + "epoch": 0.69, + "grad_norm": 0.2508257410125616, + "learning_rate": 7.602288677463457e-05, + "loss": 1.2468, + "step": 375 + }, + { + "epoch": 0.69, + "grad_norm": 0.25324577774935964, + "learning_rate": 7.598574471677644e-05, + "loss": 1.2603, + "step": 376 + }, + { + "epoch": 0.69, + "grad_norm": 0.35888776531769967, + "learning_rate": 7.59484391900302e-05, + "loss": 1.1929, + "step": 377 + }, + { + "epoch": 0.69, + "grad_norm": 0.22048517191014724, + "learning_rate": 7.591097036385994e-05, + "loss": 1.1783, + "step": 378 + }, + { + "epoch": 0.69, + "grad_norm": 0.2781160412746083, + "learning_rate": 7.587333840847162e-05, + "loss": 1.3397, + "step": 379 + }, + { + "epoch": 0.7, + "grad_norm": 0.24033046830332258, + "learning_rate": 7.583554349481222e-05, + "loss": 1.2436, + "step": 380 + }, + { + "epoch": 0.7, + "grad_norm": 0.26413762380260003, + "learning_rate": 7.579758579456893e-05, + "loss": 1.1917, + "step": 381 + }, + { + "epoch": 0.7, + "grad_norm": 0.2390937887338632, + "learning_rate": 7.575946548016847e-05, + "loss": 1.2186, + "step": 382 + }, + { + "epoch": 0.7, + "grad_norm": 0.25131263043429275, + "learning_rate": 7.572118272477622e-05, + "loss": 1.2538, + "step": 383 + }, + { + "epoch": 0.7, + "grad_norm": 0.223974104870702, + "learning_rate": 7.568273770229546e-05, + "loss": 1.2165, + "step": 384 + }, + { + "epoch": 0.7, + "grad_norm": 0.25840356830252875, + "learning_rate": 7.564413058736663e-05, + "loss": 1.1848, + "step": 385 + }, + { + "epoch": 0.71, + "grad_norm": 0.2723156683076603, + "learning_rate": 7.560536155536641e-05, + "loss": 1.1982, + "step": 386 + }, + { + "epoch": 0.71, + "grad_norm": 0.265687427976889, + "learning_rate": 7.556643078240708e-05, + "loss": 1.231, + "step": 387 + }, + { + "epoch": 0.71, + "grad_norm": 0.25152762080976077, + "learning_rate": 7.552733844533562e-05, + "loss": 1.1974, + "step": 388 + }, + { + "epoch": 0.71, + "grad_norm": 0.2366049485053541, + "learning_rate": 7.548808472173292e-05, + "loss": 1.3119, + "step": 389 + }, + { + "epoch": 0.71, + "grad_norm": 0.22092196577077122, + "learning_rate": 7.5448669789913e-05, + "loss": 1.195, + "step": 390 + }, + { + "epoch": 0.72, + "grad_norm": 0.22667521540462374, + "learning_rate": 7.540909382892217e-05, + "loss": 1.1431, + "step": 391 + }, + { + "epoch": 0.72, + "grad_norm": 0.25432207282646513, + "learning_rate": 7.536935701853823e-05, + "loss": 1.2173, + "step": 392 + }, + { + "epoch": 0.72, + "grad_norm": 0.29950506457923864, + "learning_rate": 7.53294595392697e-05, + "loss": 1.1962, + "step": 393 + }, + { + "epoch": 0.72, + "grad_norm": 0.24735689607229913, + "learning_rate": 7.528940157235487e-05, + "loss": 1.2053, + "step": 394 + }, + { + "epoch": 0.72, + "grad_norm": 0.24394198607459663, + "learning_rate": 7.524918329976114e-05, + "loss": 1.1979, + "step": 395 + }, + { + "epoch": 0.72, + "grad_norm": 0.2630369372689188, + "learning_rate": 7.520880490418409e-05, + "loss": 1.2111, + "step": 396 + }, + { + "epoch": 0.73, + "grad_norm": 0.26275028416291457, + "learning_rate": 7.516826656904664e-05, + "loss": 1.2133, + "step": 397 + }, + { + "epoch": 0.73, + "grad_norm": 0.23938074620956928, + "learning_rate": 7.512756847849831e-05, + "loss": 1.1355, + "step": 398 + }, + { + "epoch": 0.73, + "grad_norm": 0.3724960610098138, + "learning_rate": 7.508671081741428e-05, + "loss": 1.2572, + "step": 399 + }, + { + "epoch": 0.73, + "grad_norm": 0.24161685847894723, + "learning_rate": 7.504569377139462e-05, + "loss": 1.1706, + "step": 400 + }, + { + "epoch": 0.73, + "grad_norm": 0.26121591322670523, + "learning_rate": 7.50045175267634e-05, + "loss": 1.2135, + "step": 401 + }, + { + "epoch": 0.74, + "grad_norm": 0.2465579498164775, + "learning_rate": 7.496318227056788e-05, + "loss": 1.1641, + "step": 402 + }, + { + "epoch": 0.74, + "grad_norm": 0.2556288696122787, + "learning_rate": 7.492168819057767e-05, + "loss": 1.2939, + "step": 403 + }, + { + "epoch": 0.74, + "grad_norm": 0.261481216336303, + "learning_rate": 7.488003547528382e-05, + "loss": 1.2026, + "step": 404 + }, + { + "epoch": 0.74, + "grad_norm": 0.2389415135676362, + "learning_rate": 7.483822431389799e-05, + "loss": 1.2131, + "step": 405 + }, + { + "epoch": 0.74, + "grad_norm": 0.2559201956627192, + "learning_rate": 7.479625489635162e-05, + "loss": 1.1246, + "step": 406 + }, + { + "epoch": 0.74, + "grad_norm": 0.27127932491822604, + "learning_rate": 7.475412741329504e-05, + "loss": 1.2429, + "step": 407 + }, + { + "epoch": 0.75, + "grad_norm": 0.27006004008695594, + "learning_rate": 7.47118420560966e-05, + "loss": 1.2388, + "step": 408 + }, + { + "epoch": 0.75, + "grad_norm": 0.23716823297200537, + "learning_rate": 7.466939901684182e-05, + "loss": 1.1264, + "step": 409 + }, + { + "epoch": 0.75, + "grad_norm": 0.2885373898669248, + "learning_rate": 7.462679848833252e-05, + "loss": 1.2786, + "step": 410 + }, + { + "epoch": 0.75, + "grad_norm": 0.49215227598639927, + "learning_rate": 7.458404066408588e-05, + "loss": 1.2386, + "step": 411 + }, + { + "epoch": 0.75, + "grad_norm": 0.24235735604947403, + "learning_rate": 7.454112573833368e-05, + "loss": 1.1423, + "step": 412 + }, + { + "epoch": 0.76, + "grad_norm": 0.2584614748054343, + "learning_rate": 7.449805390602127e-05, + "loss": 1.2669, + "step": 413 + }, + { + "epoch": 0.76, + "grad_norm": 0.23806123085998873, + "learning_rate": 7.445482536280684e-05, + "loss": 1.1763, + "step": 414 + }, + { + "epoch": 0.76, + "grad_norm": 0.24459517607786851, + "learning_rate": 7.441144030506043e-05, + "loss": 1.198, + "step": 415 + }, + { + "epoch": 0.76, + "grad_norm": 0.25801616402700395, + "learning_rate": 7.436789892986304e-05, + "loss": 1.2136, + "step": 416 + }, + { + "epoch": 0.76, + "grad_norm": 0.2814819942392514, + "learning_rate": 7.432420143500578e-05, + "loss": 1.2398, + "step": 417 + }, + { + "epoch": 0.76, + "grad_norm": 0.22134709322606153, + "learning_rate": 7.428034801898893e-05, + "loss": 1.1592, + "step": 418 + }, + { + "epoch": 0.77, + "grad_norm": 0.2899677536995633, + "learning_rate": 7.42363388810211e-05, + "loss": 1.2296, + "step": 419 + }, + { + "epoch": 0.77, + "grad_norm": 0.24005943230262294, + "learning_rate": 7.419217422101822e-05, + "loss": 1.2223, + "step": 420 + }, + { + "epoch": 0.77, + "grad_norm": 0.26417562369496167, + "learning_rate": 7.414785423960275e-05, + "loss": 1.2261, + "step": 421 + }, + { + "epoch": 0.77, + "grad_norm": 0.2580815883535521, + "learning_rate": 7.410337913810271e-05, + "loss": 1.2021, + "step": 422 + }, + { + "epoch": 0.77, + "grad_norm": 0.25242217589496435, + "learning_rate": 7.405874911855071e-05, + "loss": 1.239, + "step": 423 + }, + { + "epoch": 0.78, + "grad_norm": 0.21991733999839932, + "learning_rate": 7.401396438368315e-05, + "loss": 1.1716, + "step": 424 + }, + { + "epoch": 0.78, + "grad_norm": 0.40116538322720213, + "learning_rate": 7.396902513693924e-05, + "loss": 1.2773, + "step": 425 + }, + { + "epoch": 0.78, + "grad_norm": 0.277333939455099, + "learning_rate": 7.392393158246002e-05, + "loss": 1.2574, + "step": 426 + }, + { + "epoch": 0.78, + "grad_norm": 0.27146087746385755, + "learning_rate": 7.387868392508756e-05, + "loss": 1.2243, + "step": 427 + }, + { + "epoch": 0.78, + "grad_norm": 0.255881055620786, + "learning_rate": 7.38332823703639e-05, + "loss": 1.223, + "step": 428 + }, + { + "epoch": 0.78, + "grad_norm": 0.24807364856677255, + "learning_rate": 7.378772712453021e-05, + "loss": 1.1985, + "step": 429 + }, + { + "epoch": 0.79, + "grad_norm": 0.25746257617764423, + "learning_rate": 7.37420183945258e-05, + "loss": 1.2502, + "step": 430 + }, + { + "epoch": 0.79, + "grad_norm": 0.28851991049982234, + "learning_rate": 7.369615638798722e-05, + "loss": 1.2535, + "step": 431 + }, + { + "epoch": 0.79, + "grad_norm": 0.24113389811604363, + "learning_rate": 7.365014131324725e-05, + "loss": 1.2227, + "step": 432 + }, + { + "epoch": 0.79, + "grad_norm": 0.2414465151257969, + "learning_rate": 7.360397337933405e-05, + "loss": 1.1884, + "step": 433 + }, + { + "epoch": 0.79, + "grad_norm": 0.2735463134699831, + "learning_rate": 7.355765279597011e-05, + "loss": 1.2756, + "step": 434 + }, + { + "epoch": 0.8, + "grad_norm": 0.2588437452987293, + "learning_rate": 7.351117977357139e-05, + "loss": 1.2108, + "step": 435 + }, + { + "epoch": 0.8, + "grad_norm": 0.26573294117796553, + "learning_rate": 7.346455452324629e-05, + "loss": 1.1821, + "step": 436 + }, + { + "epoch": 0.8, + "grad_norm": 0.2555476577827304, + "learning_rate": 7.341777725679473e-05, + "loss": 1.1937, + "step": 437 + }, + { + "epoch": 0.8, + "grad_norm": 0.2867704132108098, + "learning_rate": 7.337084818670716e-05, + "loss": 1.2272, + "step": 438 + }, + { + "epoch": 0.8, + "grad_norm": 0.27726678115981157, + "learning_rate": 7.332376752616367e-05, + "loss": 1.2331, + "step": 439 + }, + { + "epoch": 0.8, + "grad_norm": 0.26955338021079955, + "learning_rate": 7.32765354890329e-05, + "loss": 1.1731, + "step": 440 + }, + { + "epoch": 0.81, + "grad_norm": 0.25250321202536524, + "learning_rate": 7.322915228987116e-05, + "loss": 1.2653, + "step": 441 + }, + { + "epoch": 0.81, + "grad_norm": 0.24748844179765395, + "learning_rate": 7.318161814392143e-05, + "loss": 1.24, + "step": 442 + }, + { + "epoch": 0.81, + "grad_norm": 0.28177805247356325, + "learning_rate": 7.313393326711239e-05, + "loss": 1.185, + "step": 443 + }, + { + "epoch": 0.81, + "grad_norm": 0.24093242000396312, + "learning_rate": 7.30860978760574e-05, + "loss": 1.1994, + "step": 444 + }, + { + "epoch": 0.81, + "grad_norm": 0.26277803901457075, + "learning_rate": 7.30381121880536e-05, + "loss": 1.212, + "step": 445 + }, + { + "epoch": 0.82, + "grad_norm": 0.2506524258682433, + "learning_rate": 7.298997642108079e-05, + "loss": 1.2421, + "step": 446 + }, + { + "epoch": 0.82, + "grad_norm": 0.2840599700015824, + "learning_rate": 7.294169079380061e-05, + "loss": 1.1818, + "step": 447 + }, + { + "epoch": 0.82, + "grad_norm": 0.24892184038117549, + "learning_rate": 7.289325552555538e-05, + "loss": 1.1916, + "step": 448 + }, + { + "epoch": 0.82, + "grad_norm": 0.2700898428541357, + "learning_rate": 7.284467083636722e-05, + "loss": 1.2517, + "step": 449 + }, + { + "epoch": 0.82, + "grad_norm": 0.2617848546539419, + "learning_rate": 7.279593694693698e-05, + "loss": 1.2063, + "step": 450 + } + ], + "logging_steps": 1.0, + "max_steps": 1638, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 50, + "total_flos": 466496262242304.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-450/training_args.bin b/checkpoint-450/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..c5d2416a3b70bb5260978ec9996f00154a724ba7 --- /dev/null +++ b/checkpoint-450/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b22e8f9d51a16d03a2c506fa3d1eafa8f4b1ae992992c2086a4d435ffd97387e +size 6712 diff --git a/checkpoint-450/zero_to_fp32.py b/checkpoint-450/zero_to_fp32.py new file mode 100755 index 0000000000000000000000000000000000000000..24cc342e78d1a006c782b3a4cd68d9ce786d8fd8 --- /dev/null +++ b/checkpoint-450/zero_to_fp32.py @@ -0,0 +1,604 @@ +#!/usr/bin/env python + +# Copyright (c) Microsoft Corporation. +# SPDX-License-Identifier: Apache-2.0 + +# DeepSpeed Team + +# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets +# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in +# the future. Once extracted, the weights don't require DeepSpeed and can be used in any +# application. +# +# example: python zero_to_fp32.py . pytorch_model.bin + +import argparse +import torch +import glob +import math +import os +import re +from collections import OrderedDict +from dataclasses import dataclass + +# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with +# DeepSpeed data structures it has to be available in the current python environment. +from deepspeed.utils import logger +from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS, + FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES, + FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS) + + +@dataclass +class zero_model_state: + buffers: dict() + param_shapes: dict() + shared_params: list + ds_version: int + frozen_param_shapes: dict() + frozen_param_fragments: dict() + + +debug = 0 + +# load to cpu +device = torch.device('cpu') + + +def atoi(text): + return int(text) if text.isdigit() else text + + +def natural_keys(text): + ''' + alist.sort(key=natural_keys) sorts in human order + http://nedbatchelder.com/blog/200712/human_sorting.html + (See Toothy's implementation in the comments) + ''' + return [atoi(c) for c in re.split(r'(\d+)', text)] + + +def get_model_state_file(checkpoint_dir, zero_stage): + if not os.path.isdir(checkpoint_dir): + raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist") + + # there should be only one file + if zero_stage <= 2: + file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt") + elif zero_stage == 3: + file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt") + + if not os.path.exists(file): + raise FileNotFoundError(f"can't find model states file at '{file}'") + + return file + + +def get_checkpoint_files(checkpoint_dir, glob_pattern): + # XXX: need to test that this simple glob rule works for multi-node setup too + ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys) + + if len(ckpt_files) == 0: + raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'") + + return ckpt_files + + +def get_optim_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt") + + +def get_model_state_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_model_states.pt") + + +def parse_model_states(files): + zero_model_states = [] + for file in files: + state_dict = torch.load(file, map_location=device) + + if BUFFER_NAMES not in state_dict: + raise ValueError(f"{file} is not a model state checkpoint") + buffer_names = state_dict[BUFFER_NAMES] + if debug: + print("Found buffers:", buffer_names) + + # recover just the buffers while restoring them to fp32 if they were saved in fp16 + buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names} + param_shapes = state_dict[PARAM_SHAPES] + + # collect parameters that are included in param_shapes + param_names = [] + for s in param_shapes: + for name in s.keys(): + param_names.append(name) + + # update with frozen parameters + frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None) + if frozen_param_shapes is not None: + if debug: + print(f"Found frozen_param_shapes: {frozen_param_shapes}") + param_names += list(frozen_param_shapes.keys()) + + # handle shared params + shared_params = [[k, v] for k, v in state_dict["shared_params"].items()] + + ds_version = state_dict.get(DS_VERSION, None) + + frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None) + + z_model_state = zero_model_state(buffers=buffers, + param_shapes=param_shapes, + shared_params=shared_params, + ds_version=ds_version, + frozen_param_shapes=frozen_param_shapes, + frozen_param_fragments=frozen_param_fragments) + zero_model_states.append(z_model_state) + + return zero_model_states + + +def parse_optim_states(files, ds_checkpoint_dir): + + total_files = len(files) + state_dicts = [] + for f in files: + state_dict = torch.load(f, map_location=device) + # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights + # and also handle the case where it was already removed by another helper script + state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None) + state_dicts.append(state_dict) + + if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]: + raise ValueError(f"{files[0]} is not a zero checkpoint") + zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE] + world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT] + + # For ZeRO-2 each param group can have different partition_count as data parallelism for expert + # parameters can be different from data parallelism for non-expert parameters. So we can just + # use the max of the partition_count to get the dp world_size. + + if type(world_size) is list: + world_size = max(world_size) + + if world_size != total_files: + raise ValueError( + f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. " + "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes." + ) + + # the groups are named differently in each stage + if zero_stage <= 2: + fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS + elif zero_stage == 3: + fp32_groups_key = FP32_FLAT_GROUPS + else: + raise ValueError(f"unknown zero stage {zero_stage}") + + if zero_stage <= 2: + fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))] + elif zero_stage == 3: + # if there is more than one param group, there will be multiple flattened tensors - one + # flattened tensor per group - for simplicity merge them into a single tensor + # + # XXX: could make the script more memory efficient for when there are multiple groups - it + # will require matching the sub-lists of param_shapes for each param group flattened tensor + + fp32_flat_groups = [ + torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts)) + ] + + return zero_stage, world_size, fp32_flat_groups + + +def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters): + """ + Returns fp32 state_dict reconstructed from ds checkpoint + + Args: + - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are) + + """ + print(f"Processing zero checkpoint '{ds_checkpoint_dir}'") + + optim_files = get_optim_files(ds_checkpoint_dir) + zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir) + print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}") + + model_files = get_model_state_files(ds_checkpoint_dir) + + zero_model_states = parse_model_states(model_files) + print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}') + + if zero_stage <= 2: + return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + elif zero_stage == 3: + return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + + +def _zero2_merge_frozen_params(state_dict, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + frozen_param_fragments = zero_model_states[0].frozen_param_fragments + + if debug: + num_elem = sum(s.numel() for s in frozen_param_shapes.values()) + print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in frozen_param_fragments.values()]) + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + state_dict[name] = frozen_param_fragments[name] + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _has_callable(obj, fn): + attr = getattr(obj, fn, None) + return callable(attr) + + +def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + + # Reconstruction protocol: + # + # XXX: document this + + if debug: + for i in range(world_size): + for j in range(len(fp32_flat_groups[0])): + print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}") + + # XXX: memory usage doubles here (zero2) + num_param_groups = len(fp32_flat_groups[0]) + merged_single_partition_of_fp32_groups = [] + for i in range(num_param_groups): + merged_partitions = [sd[i] for sd in fp32_flat_groups] + full_single_fp32_vector = torch.cat(merged_partitions, 0) + merged_single_partition_of_fp32_groups.append(full_single_fp32_vector) + avail_numel = sum( + [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups]) + + if debug: + wanted_params = sum([len(shapes) for shapes in param_shapes]) + wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes]) + # not asserting if there is a mismatch due to possible padding + print(f"Have {avail_numel} numels to process.") + print(f"Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + total_numel = 0 + total_params = 0 + for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups): + offset = 0 + avail_numel = full_single_fp32_vector.numel() + for name, shape in shapes.items(): + + unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape) + total_numel += unpartitioned_numel + total_params += 1 + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape) + offset += unpartitioned_numel + + # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and + # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex + # paddings performed in the code it's almost impossible to predict the exact numbers w/o the + # live optimizer object, so we are checking that the numbers are within the right range + align_to = 2 * world_size + + def zero2_align(x): + return align_to * math.ceil(x / align_to) + + if debug: + print(f"original offset={offset}, avail_numel={avail_numel}") + + offset = zero2_align(offset) + avail_numel = zero2_align(avail_numel) + + if debug: + print(f"aligned offset={offset}, avail_numel={avail_numel}") + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero2_merge_frozen_params(state_dict, zero_model_states) + + _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def zero3_partitioned_param_info(unpartitioned_numel, world_size): + remainder = unpartitioned_numel % world_size + padding_numel = (world_size - remainder) if remainder else 0 + partitioned_numel = math.ceil(unpartitioned_numel / world_size) + return partitioned_numel, padding_numel + + +def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + if debug: + for i in range(world_size): + num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values()) + print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in zero_model_states[0].frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states) + state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape) + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + avail_numel = fp32_flat_groups[0].numel() * world_size + # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each + # param, re-consolidating each param, while dealing with padding if any + + # merge list of dicts, preserving order + param_shapes = {k: v for d in param_shapes for k, v in d.items()} + + if debug: + for i in range(world_size): + print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}") + + wanted_params = len(param_shapes) + wanted_numel = sum(shape.numel() for shape in param_shapes.values()) + # not asserting if there is a mismatch due to possible padding + avail_numel = fp32_flat_groups[0].numel() * world_size + print(f"Trainable params: Have {avail_numel} numels to process.") + print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + offset = 0 + total_numel = 0 + total_params = 0 + for name, shape in param_shapes.items(): + + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + total_params += 1 + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + # XXX: memory usage doubles here + state_dict[name] = torch.cat( + tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)), + 0).narrow(0, 0, unpartitioned_numel).view(shape) + offset += partitioned_numel + + offset *= world_size + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero3_merge_frozen_params(state_dict, world_size, zero_model_states) + + _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with + ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example + via a model hub. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + + Returns: + - pytorch ``state_dict`` + + Note: this approach may not work if your application doesn't have sufficient free CPU memory and + you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with + the checkpoint. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + # do the training and checkpoint saving + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu + model = model.cpu() # move to cpu + model.load_state_dict(state_dict) + # submit to model hub or save the model to share with others + + In this example the ``model`` will no longer be usable in the deepspeed context of the same + application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead. + + """ + if tag is None: + latest_path = os.path.join(checkpoint_dir, 'latest') + if os.path.isfile(latest_path): + with open(latest_path, 'r') as fd: + tag = fd.read().strip() + else: + raise ValueError(f"Unable to find 'latest' file at {latest_path}") + + ds_checkpoint_dir = os.path.join(checkpoint_dir, tag) + + if not os.path.isdir(ds_checkpoint_dir): + raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist") + + return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters) + + +def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be + loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + """ + + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters) + print(f"Saving fp32 state dict to {output_file}") + torch.save(state_dict, output_file) + + +def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None): + """ + 1. Put the provided model to cpu + 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` + 3. Load it into the provided model + + Args: + - ``model``: the model object to update + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + + Returns: + - ``model`: modified model + + Make sure you have plenty of CPU memory available before you call this function. If you don't + have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it + conveniently placed for you in the checkpoint folder. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint + model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir) + # submit to model hub or save the model to share with others + + Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context + of the same application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + """ + logger.info(f"Extracting fp32 weights") + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag) + + logger.info(f"Overwriting model with fp32 weights") + model = model.cpu() + model.load_state_dict(state_dict, strict=False) + + return model + + +if __name__ == "__main__": + + parser = argparse.ArgumentParser() + parser.add_argument("checkpoint_dir", + type=str, + help="path to the desired checkpoint folder, e.g., path/checkpoint-12") + parser.add_argument( + "output_file", + type=str, + help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)") + parser.add_argument("-t", + "--tag", + type=str, + default=None, + help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1") + parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters") + parser.add_argument("-d", "--debug", action='store_true', help="enable debug") + args = parser.parse_args() + + debug = args.debug + + convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, + args.output_file, + tag=args.tag, + exclude_frozen_parameters=args.exclude_frozen_parameters) diff --git a/checkpoint-500/README.md b/checkpoint-500/README.md new file mode 100644 index 0000000000000000000000000000000000000000..16b1eacdd9353dec380a08ee77ce6ed5ab50f12e --- /dev/null +++ b/checkpoint-500/README.md @@ -0,0 +1,202 @@ +--- +library_name: peft +base_model: gotzmann/uni +--- + +# Model Card for Model ID + + + + + +## Model Details + +### Model Description + + + + + +- **Developed by:** [More Information Needed] +- **Funded by [optional]:** [More Information Needed] +- **Shared by [optional]:** [More Information Needed] +- **Model type:** [More Information Needed] +- **Language(s) (NLP):** [More Information Needed] +- **License:** [More Information Needed] +- **Finetuned from model [optional]:** [More Information Needed] + +### Model Sources [optional] + + + +- **Repository:** [More Information Needed] +- **Paper [optional]:** [More Information Needed] +- **Demo [optional]:** [More Information Needed] + +## Uses + + + +### Direct Use + + + +[More Information Needed] + +### Downstream Use [optional] + + + +[More Information Needed] + +### Out-of-Scope Use + + + +[More Information Needed] + +## Bias, Risks, and Limitations + + + +[More Information Needed] + +### Recommendations + + + +Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations. + +## How to Get Started with the Model + +Use the code below to get started with the model. + +[More Information Needed] + +## Training Details + +### Training Data + + + +[More Information Needed] + +### Training Procedure + + + +#### Preprocessing [optional] + +[More Information Needed] + + +#### Training Hyperparameters + +- **Training regime:** [More Information Needed] + +#### Speeds, Sizes, Times [optional] + + + +[More Information Needed] + +## Evaluation + + + +### Testing Data, Factors & Metrics + +#### Testing Data + + + +[More Information Needed] + +#### Factors + + + +[More Information Needed] + +#### Metrics + + + +[More Information Needed] + +### Results + +[More Information Needed] + +#### Summary + + + +## Model Examination [optional] + + + +[More Information Needed] + +## Environmental Impact + + + +Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700). + +- **Hardware Type:** [More Information Needed] +- **Hours used:** [More Information Needed] +- **Cloud Provider:** [More Information Needed] +- **Compute Region:** [More Information Needed] +- **Carbon Emitted:** [More Information Needed] + +## Technical Specifications [optional] + +### Model Architecture and Objective + +[More Information Needed] + +### Compute Infrastructure + +[More Information Needed] + +#### Hardware + +[More Information Needed] + +#### Software + +[More Information Needed] + +## Citation [optional] + + + +**BibTeX:** + +[More Information Needed] + +**APA:** + +[More Information Needed] + +## Glossary [optional] + + + +[More Information Needed] + +## More Information [optional] + +[More Information Needed] + +## Model Card Authors [optional] + +[More Information Needed] + +## Model Card Contact + +[More Information Needed] +### Framework versions + +- PEFT 0.10.0 \ No newline at end of file diff --git a/checkpoint-500/adapter_config.json b/checkpoint-500/adapter_config.json new file mode 100644 index 0000000000000000000000000000000000000000..3cd6dba5d79f7ca21fd4ad465cbbcac1e0960476 --- /dev/null +++ b/checkpoint-500/adapter_config.json @@ -0,0 +1,31 @@ +{ + "alpha_pattern": {}, + "auto_mapping": null, + "base_model_name_or_path": "gotzmann/uni", + "bias": "none", + "fan_in_fan_out": false, + "inference_mode": true, + "init_lora_weights": true, + "layer_replication": null, + "layers_pattern": null, + "layers_to_transform": null, + "loftq_config": {}, + "lora_alpha": 128, + "lora_dropout": 0.0, + "megatron_config": null, + "megatron_core": "megatron.core", + "modules_to_save": null, + "peft_type": "LORA", + "r": 128, + "rank_pattern": {}, + "revision": null, + "target_modules": [ + "v_proj", + "k_proj", + "q_proj", + "o_proj" + ], + "task_type": "CAUSAL_LM", + "use_dora": false, + "use_rslora": true +} \ No newline at end of file diff --git a/checkpoint-500/adapter_model.safetensors b/checkpoint-500/adapter_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..25998e6401d2eddee182d853ec7076e001d867bb --- /dev/null +++ b/checkpoint-500/adapter_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:83758c1ddea613ef778179c38abd9da6838eb7dd974e77e3e56c84640f4107f0 +size 1048664848 diff --git a/checkpoint-500/global_step500/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt b/checkpoint-500/global_step500/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..30cde85576578f264f48c87e8ab62f94960e448c --- /dev/null +++ b/checkpoint-500/global_step500/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0d14f988a761a7ce00710b0b629c9fcd09cb201de111af5c7cedd4518b0c683a +size 787270042 diff --git a/checkpoint-500/global_step500/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt b/checkpoint-500/global_step500/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..951371adaa80900bff0bc86f90eb2a70f070ff54 --- /dev/null +++ b/checkpoint-500/global_step500/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:b7a9b590ff45a757f25e2c576d88c5a1ab260ae0dab9a61ec933d171371ea21b +size 787270042 diff --git a/checkpoint-500/global_step500/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt b/checkpoint-500/global_step500/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..6c0fe2b0b6bc490c50cb4e776266717393ff1a17 --- /dev/null +++ b/checkpoint-500/global_step500/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:b29865cf6ac925303112462d79c8f920ffa70517aaa32fc92c33d8ce11660e8a +size 787270042 diff --git a/checkpoint-500/global_step500/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt b/checkpoint-500/global_step500/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..b028efe4d32ea387151a6229c3e499005b4aa565 --- /dev/null +++ b/checkpoint-500/global_step500/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6eac043d481e9c249f1d38e77b3ee662b36e55fa35a81f4e7b572b8958e513ff +size 787270042 diff --git a/checkpoint-500/global_step500/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt b/checkpoint-500/global_step500/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..23a39c058d10522a1fb90b9ebaeb0d0dee4afcda --- /dev/null +++ b/checkpoint-500/global_step500/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:722060c492e959244200c0c3fd8e64d5653e040fca5655c0b737386cf39eea26 +size 787270042 diff --git a/checkpoint-500/global_step500/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt b/checkpoint-500/global_step500/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..c04d178900a3db573531a14582cdad8abf9384f4 --- /dev/null +++ b/checkpoint-500/global_step500/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9eb253220e37a6cef37c54b6a5828f27e5144b24016dc7eb58d892f5a0af94aa +size 787270042 diff --git a/checkpoint-500/global_step500/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt b/checkpoint-500/global_step500/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..086ae6788796fb71d5d42a67a710abc19a63fafb --- /dev/null +++ b/checkpoint-500/global_step500/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:f11fc7f8012e1e361f24c6cd0adffb2642a0d691544be4bddcdbf2d5fb7a7c76 +size 787270042 diff --git a/checkpoint-500/global_step500/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt b/checkpoint-500/global_step500/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..e998dff96600e95d931a97ccc30f41214d965eb8 --- /dev/null +++ b/checkpoint-500/global_step500/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:cc078293f190630f7e77c6b3f2fe73669f2057b6ce982490fe425ad19fd0ccf7 +size 787270042 diff --git a/checkpoint-500/global_step500/zero_pp_rank_0_mp_rank_00_model_states.pt b/checkpoint-500/global_step500/zero_pp_rank_0_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..7b89211b0c52111571e2de9f61fdcca7c60d3a4b --- /dev/null +++ b/checkpoint-500/global_step500/zero_pp_rank_0_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bc93ec286c6a45468a2ec8a2733f7c25526cf847e51bafb4eb0596094a5e9f49 +size 653742 diff --git a/checkpoint-500/global_step500/zero_pp_rank_1_mp_rank_00_model_states.pt b/checkpoint-500/global_step500/zero_pp_rank_1_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..66f8e3a20049977242bae7db50de408317a9be5e --- /dev/null +++ b/checkpoint-500/global_step500/zero_pp_rank_1_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cd5e50551ae370d21a908003a7bb17c2fb644cf991279d1a990036afc0525251 +size 653742 diff --git a/checkpoint-500/global_step500/zero_pp_rank_2_mp_rank_00_model_states.pt b/checkpoint-500/global_step500/zero_pp_rank_2_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..76f4775785d2987e539fd855932130a4bfb9d610 --- /dev/null +++ b/checkpoint-500/global_step500/zero_pp_rank_2_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:87eb5b17005b0580a82f8e1669ba240a18471d321a5944410c3ec48ea7a847fb +size 653742 diff --git a/checkpoint-500/global_step500/zero_pp_rank_3_mp_rank_00_model_states.pt b/checkpoint-500/global_step500/zero_pp_rank_3_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..71627c723bead832d2456242b1cf1c563213e7d2 --- /dev/null +++ b/checkpoint-500/global_step500/zero_pp_rank_3_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a2842b4114ea3bb10c42198fdfc0a2958005d39a235d39988f95eaa7e90a93ca +size 653742 diff --git a/checkpoint-500/global_step500/zero_pp_rank_4_mp_rank_00_model_states.pt b/checkpoint-500/global_step500/zero_pp_rank_4_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..2fd51663fe1eadce5468ea6c2682f132dec70e1d --- /dev/null +++ b/checkpoint-500/global_step500/zero_pp_rank_4_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c68448b2d7ebaf4c38271b8c9cd7a48f1906f2ef23e3ddbeef5c39545508ac00 +size 653742 diff --git a/checkpoint-500/global_step500/zero_pp_rank_5_mp_rank_00_model_states.pt b/checkpoint-500/global_step500/zero_pp_rank_5_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..3570b4d435301b46f73b5abb029637ae66b1cd6f --- /dev/null +++ b/checkpoint-500/global_step500/zero_pp_rank_5_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ca270f4813e13d2c89663a5b1a70ec46408b87096075443f24a10c95b0c2863d +size 653742 diff --git a/checkpoint-500/global_step500/zero_pp_rank_6_mp_rank_00_model_states.pt b/checkpoint-500/global_step500/zero_pp_rank_6_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..97eac0f6180f2a455a1a67ce25aa79cec19e2f1b --- /dev/null +++ b/checkpoint-500/global_step500/zero_pp_rank_6_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f8eef63270ad1d645a07b1a0b5a0bf3cfaba8785a660af7a96d7f3e5156363e3 +size 653742 diff --git a/checkpoint-500/global_step500/zero_pp_rank_7_mp_rank_00_model_states.pt b/checkpoint-500/global_step500/zero_pp_rank_7_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..eecfad8f6db4dbe91d5524150ea9641b7926d820 --- /dev/null +++ b/checkpoint-500/global_step500/zero_pp_rank_7_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:50e64b923ed1ba99bbf51c5d8eb3a3448a03037421db1c1a6f65839ee6e8c258 +size 653742 diff --git a/checkpoint-500/latest b/checkpoint-500/latest new file mode 100644 index 0000000000000000000000000000000000000000..f0b47ce15fff9a01b2a416a473b2148085048a50 --- /dev/null +++ b/checkpoint-500/latest @@ -0,0 +1 @@ +global_step500 \ No newline at end of file diff --git a/checkpoint-500/rng_state_0.pth b/checkpoint-500/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..b346349ce12dd5a17d4b91ed2a5722bb52550950 --- /dev/null +++ b/checkpoint-500/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ad8a35afd8967cbb748405387e44426e43ad127028e826eddc9b67d2ca873c85 +size 15984 diff --git a/checkpoint-500/rng_state_1.pth b/checkpoint-500/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..68f3c6994456cb8d0592a5375d99503c8924b1c4 --- /dev/null +++ b/checkpoint-500/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f338ce80d7c441076bfc8c53b84067a0181f5a14e80c13d5acb8150b659f4d73 +size 15984 diff --git a/checkpoint-500/rng_state_2.pth b/checkpoint-500/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..be044f6ceeed587d30e80c2f72d5aa19fdc9947b --- /dev/null +++ b/checkpoint-500/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c9fbc9fa428939be10b46779f0eb5cd833e0da426b1cbdee77b3a55b6952235b +size 15984 diff --git a/checkpoint-500/rng_state_3.pth b/checkpoint-500/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..fc825249656a9b858782542bd3f4386250f1dfe0 --- /dev/null +++ b/checkpoint-500/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ac55dba0b79d5fa4699d239da2f966d52040d576d31234ac8d4632e6956481bc +size 15984 diff --git a/checkpoint-500/rng_state_4.pth b/checkpoint-500/rng_state_4.pth new file mode 100644 index 0000000000000000000000000000000000000000..d30f52a44be563c152ae09db6ae934da6da0d3ed --- /dev/null +++ b/checkpoint-500/rng_state_4.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:af2d0c015100768ffa23faf3b6c2d54ea89eb045603e30e55cd211e06ff34972 +size 15984 diff --git a/checkpoint-500/rng_state_5.pth b/checkpoint-500/rng_state_5.pth new file mode 100644 index 0000000000000000000000000000000000000000..c8715d27ab23ae545d58039cf949cc44ecc1da5e --- /dev/null +++ b/checkpoint-500/rng_state_5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c60a1b40608e34bc801c8231f97b81c53b5290dfaed1b9cd0ccbeca29574a991 +size 15984 diff --git a/checkpoint-500/rng_state_6.pth b/checkpoint-500/rng_state_6.pth new file mode 100644 index 0000000000000000000000000000000000000000..1ed791b6ef76eadf0b0c55a5733411771e2ae027 --- /dev/null +++ b/checkpoint-500/rng_state_6.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3ad6a142a403eb9aafc4a3a9a856bca648fe31fd22d796867baca31fb13656aa +size 15984 diff --git a/checkpoint-500/rng_state_7.pth b/checkpoint-500/rng_state_7.pth new file mode 100644 index 0000000000000000000000000000000000000000..800c3bbbc5edf7db01a8316069d439c5fb8d8c30 --- /dev/null +++ b/checkpoint-500/rng_state_7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:38bc23a138cc800b22881742c0f3f9a71731a9a7111c6058a0077e6274d21773 +size 15984 diff --git a/checkpoint-500/scheduler.pt b/checkpoint-500/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..92346318f4ff3265cc7c4b0b86c905c695421b97 --- /dev/null +++ b/checkpoint-500/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f864fbe1f67c6816a27efc600470a00e4227c0e65abc1cef188658f5ca2a8f8b +size 1064 diff --git a/checkpoint-500/special_tokens_map.json b/checkpoint-500/special_tokens_map.json new file mode 100644 index 0000000000000000000000000000000000000000..72ecfeeb7e14d244c936169d2ed139eeae235ef1 --- /dev/null +++ b/checkpoint-500/special_tokens_map.json @@ -0,0 +1,24 @@ +{ + "bos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "eos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "pad_token": "", + "unk_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + } +} diff --git a/checkpoint-500/tokenizer.model b/checkpoint-500/tokenizer.model new file mode 100644 index 0000000000000000000000000000000000000000..6c00c742ce03c627d6cd5b795984876fa49fa899 --- /dev/null +++ b/checkpoint-500/tokenizer.model @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347 +size 499723 diff --git a/checkpoint-500/tokenizer_config.json b/checkpoint-500/tokenizer_config.json new file mode 100644 index 0000000000000000000000000000000000000000..bb5a9f09d8c0f3c32c66fc6118fe5c76c5c6fd90 --- /dev/null +++ b/checkpoint-500/tokenizer_config.json @@ -0,0 +1,45 @@ +{ + "add_bos_token": true, + "add_eos_token": false, + "add_prefix_space": false, + "added_tokens_decoder": { + "0": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "1": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "2": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + } + }, + "bos_token": "", + "chat_template": "{% if messages[0]['role'] == 'system' %}{% set system_message = messages[0]['content'] %}{% endif %}{% if system_message is defined %}{{ '' + '### System:\\n\\n' + system_message }}{% endif %}{% for message in messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ '\\n\\n### Human:\\n\\n' + content }}{% elif message['role'] == 'assistant' %}{{ '\\n\\n### Assistant:\\n\\n' + content + '' }}{% endif %}{% endfor %}", + "clean_up_tokenization_spaces": false, + "eos_token": "", + "legacy": false, + "model_max_length": 1000000000000000019884624838656, + "pad_token": "", + "padding_side": "right", + "sp_model_kwargs": {}, + "spaces_between_special_tokens": false, + "split_special_tokens": false, + "tokenizer_class": "LlamaTokenizer", + "unk_token": "", + "use_default_system_prompt": false +} diff --git a/checkpoint-500/trainer_state.json b/checkpoint-500/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..979a3c8490cb57b1be16d7fd6fdd2b11c86916d1 --- /dev/null +++ b/checkpoint-500/trainer_state.json @@ -0,0 +1,3521 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 0.9144947416552355, + "eval_steps": 500, + "global_step": 500, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "grad_norm": 0.849355824164473, + "learning_rate": 4.878048780487805e-07, + "loss": 1.3655, + "step": 1 + }, + { + "epoch": 0.0, + "grad_norm": 10.01567518957158, + "learning_rate": 9.75609756097561e-07, + "loss": 1.5767, + "step": 2 + }, + { + "epoch": 0.01, + "grad_norm": 0.6466000875559635, + "learning_rate": 1.4634146341463414e-06, + "loss": 1.3913, + "step": 3 + }, + { + "epoch": 0.01, + "grad_norm": 0.6644565932010504, + "learning_rate": 1.951219512195122e-06, + "loss": 1.3218, + "step": 4 + }, + { + "epoch": 0.01, + "grad_norm": 0.571354207588475, + "learning_rate": 2.4390243902439027e-06, + "loss": 1.3597, + "step": 5 + }, + { + "epoch": 0.01, + "grad_norm": 0.31036262839244955, + "learning_rate": 2.926829268292683e-06, + "loss": 1.2832, + "step": 6 + }, + { + "epoch": 0.01, + "grad_norm": 0.2622135027188184, + "learning_rate": 3.414634146341464e-06, + "loss": 1.2161, + "step": 7 + }, + { + "epoch": 0.01, + "grad_norm": 0.296824630261661, + "learning_rate": 3.902439024390244e-06, + "loss": 1.2985, + "step": 8 + }, + { + "epoch": 0.02, + "grad_norm": 0.2557267467361569, + "learning_rate": 4.390243902439025e-06, + "loss": 1.3175, + "step": 9 + }, + { + "epoch": 0.02, + "grad_norm": 0.23418939513890769, + "learning_rate": 4.8780487804878055e-06, + "loss": 1.2617, + "step": 10 + }, + { + "epoch": 0.02, + "grad_norm": 0.2364760983285843, + "learning_rate": 5.365853658536586e-06, + "loss": 1.3103, + "step": 11 + }, + { + "epoch": 0.02, + "grad_norm": 0.23893034721889, + "learning_rate": 5.853658536585366e-06, + "loss": 1.2405, + "step": 12 + }, + { + "epoch": 0.02, + "grad_norm": 0.25563593295485887, + "learning_rate": 6.341463414634147e-06, + "loss": 1.2831, + "step": 13 + }, + { + "epoch": 0.03, + "grad_norm": 0.23239975352661665, + "learning_rate": 6.829268292682928e-06, + "loss": 1.3125, + "step": 14 + }, + { + "epoch": 0.03, + "grad_norm": 0.3092813858209507, + "learning_rate": 7.317073170731707e-06, + "loss": 1.2422, + "step": 15 + }, + { + "epoch": 0.03, + "grad_norm": 0.282563380367434, + "learning_rate": 7.804878048780489e-06, + "loss": 1.2453, + "step": 16 + }, + { + "epoch": 0.03, + "grad_norm": 0.22065680088315018, + "learning_rate": 8.292682926829268e-06, + "loss": 1.2491, + "step": 17 + }, + { + "epoch": 0.03, + "grad_norm": 0.22777800877980184, + "learning_rate": 8.78048780487805e-06, + "loss": 1.2655, + "step": 18 + }, + { + "epoch": 0.03, + "grad_norm": 0.22145212540177928, + "learning_rate": 9.268292682926831e-06, + "loss": 1.2413, + "step": 19 + }, + { + "epoch": 0.04, + "grad_norm": 0.22482351883112714, + "learning_rate": 9.756097560975611e-06, + "loss": 1.2653, + "step": 20 + }, + { + "epoch": 0.04, + "grad_norm": 0.20823080508385733, + "learning_rate": 1.024390243902439e-05, + "loss": 1.2374, + "step": 21 + }, + { + "epoch": 0.04, + "grad_norm": 0.26025492562935737, + "learning_rate": 1.0731707317073172e-05, + "loss": 1.2065, + "step": 22 + }, + { + "epoch": 0.04, + "grad_norm": 0.2150252124176173, + "learning_rate": 1.1219512195121953e-05, + "loss": 1.2782, + "step": 23 + }, + { + "epoch": 0.04, + "grad_norm": 0.2505915177425618, + "learning_rate": 1.1707317073170731e-05, + "loss": 1.2742, + "step": 24 + }, + { + "epoch": 0.05, + "grad_norm": 0.20129223044786942, + "learning_rate": 1.2195121951219513e-05, + "loss": 1.3366, + "step": 25 + }, + { + "epoch": 0.05, + "grad_norm": 0.1973508510397107, + "learning_rate": 1.2682926829268294e-05, + "loss": 1.2476, + "step": 26 + }, + { + "epoch": 0.05, + "grad_norm": 0.27103325392437194, + "learning_rate": 1.3170731707317076e-05, + "loss": 1.2325, + "step": 27 + }, + { + "epoch": 0.05, + "grad_norm": 0.17954976411006285, + "learning_rate": 1.3658536585365855e-05, + "loss": 1.2523, + "step": 28 + }, + { + "epoch": 0.05, + "grad_norm": 0.22216997851088888, + "learning_rate": 1.4146341463414635e-05, + "loss": 1.3297, + "step": 29 + }, + { + "epoch": 0.05, + "grad_norm": 0.2071458864548587, + "learning_rate": 1.4634146341463415e-05, + "loss": 1.2127, + "step": 30 + }, + { + "epoch": 0.06, + "grad_norm": 0.18039422081622164, + "learning_rate": 1.5121951219512196e-05, + "loss": 1.2509, + "step": 31 + }, + { + "epoch": 0.06, + "grad_norm": 0.18631254372974412, + "learning_rate": 1.5609756097560978e-05, + "loss": 1.2247, + "step": 32 + }, + { + "epoch": 0.06, + "grad_norm": 0.18843872523649827, + "learning_rate": 1.6097560975609757e-05, + "loss": 1.195, + "step": 33 + }, + { + "epoch": 0.06, + "grad_norm": 0.2163847267778325, + "learning_rate": 1.6585365853658537e-05, + "loss": 1.2179, + "step": 34 + }, + { + "epoch": 0.06, + "grad_norm": 0.19687688475496104, + "learning_rate": 1.7073170731707317e-05, + "loss": 1.2763, + "step": 35 + }, + { + "epoch": 0.07, + "grad_norm": 0.20409643064887947, + "learning_rate": 1.75609756097561e-05, + "loss": 1.253, + "step": 36 + }, + { + "epoch": 0.07, + "grad_norm": 0.1879182661759335, + "learning_rate": 1.804878048780488e-05, + "loss": 1.2586, + "step": 37 + }, + { + "epoch": 0.07, + "grad_norm": 0.19400648948514373, + "learning_rate": 1.8536585365853663e-05, + "loss": 1.2154, + "step": 38 + }, + { + "epoch": 0.07, + "grad_norm": 0.1878879343148452, + "learning_rate": 1.902439024390244e-05, + "loss": 1.2304, + "step": 39 + }, + { + "epoch": 0.07, + "grad_norm": 0.17687475469924052, + "learning_rate": 1.9512195121951222e-05, + "loss": 1.2351, + "step": 40 + }, + { + "epoch": 0.07, + "grad_norm": 0.18223935625384885, + "learning_rate": 2e-05, + "loss": 1.2222, + "step": 41 + }, + { + "epoch": 0.08, + "grad_norm": 0.1943061629408338, + "learning_rate": 2.048780487804878e-05, + "loss": 1.2044, + "step": 42 + }, + { + "epoch": 0.08, + "grad_norm": 0.17027514338700078, + "learning_rate": 2.0975609756097564e-05, + "loss": 1.1548, + "step": 43 + }, + { + "epoch": 0.08, + "grad_norm": 0.18553769630586192, + "learning_rate": 2.1463414634146344e-05, + "loss": 1.2721, + "step": 44 + }, + { + "epoch": 0.08, + "grad_norm": 0.19732826914228765, + "learning_rate": 2.1951219512195124e-05, + "loss": 1.3097, + "step": 45 + }, + { + "epoch": 0.08, + "grad_norm": 0.18714230986631472, + "learning_rate": 2.2439024390243907e-05, + "loss": 1.2662, + "step": 46 + }, + { + "epoch": 0.09, + "grad_norm": 0.19988987568002223, + "learning_rate": 2.2926829268292683e-05, + "loss": 1.2904, + "step": 47 + }, + { + "epoch": 0.09, + "grad_norm": 0.17744650133390918, + "learning_rate": 2.3414634146341463e-05, + "loss": 1.1825, + "step": 48 + }, + { + "epoch": 0.09, + "grad_norm": 0.16576734763834533, + "learning_rate": 2.3902439024390246e-05, + "loss": 1.1858, + "step": 49 + }, + { + "epoch": 0.09, + "grad_norm": 0.179591794065527, + "learning_rate": 2.4390243902439026e-05, + "loss": 1.2711, + "step": 50 + }, + { + "epoch": 0.09, + "grad_norm": 0.17923464471176911, + "learning_rate": 2.4878048780487805e-05, + "loss": 1.2289, + "step": 51 + }, + { + "epoch": 0.1, + "grad_norm": 0.18991742907836837, + "learning_rate": 2.536585365853659e-05, + "loss": 1.3097, + "step": 52 + }, + { + "epoch": 0.1, + "grad_norm": 0.19849796137254636, + "learning_rate": 2.5853658536585368e-05, + "loss": 1.2489, + "step": 53 + }, + { + "epoch": 0.1, + "grad_norm": 0.17452371110976383, + "learning_rate": 2.634146341463415e-05, + "loss": 1.2461, + "step": 54 + }, + { + "epoch": 0.1, + "grad_norm": 0.17671022353085036, + "learning_rate": 2.682926829268293e-05, + "loss": 1.153, + "step": 55 + }, + { + "epoch": 0.1, + "grad_norm": 0.36820559192096686, + "learning_rate": 2.731707317073171e-05, + "loss": 1.2431, + "step": 56 + }, + { + "epoch": 0.1, + "grad_norm": 0.20331468526494198, + "learning_rate": 2.7804878048780487e-05, + "loss": 1.2575, + "step": 57 + }, + { + "epoch": 0.11, + "grad_norm": 0.2402486598118377, + "learning_rate": 2.829268292682927e-05, + "loss": 1.2538, + "step": 58 + }, + { + "epoch": 0.11, + "grad_norm": 0.2549409484173144, + "learning_rate": 2.878048780487805e-05, + "loss": 1.2065, + "step": 59 + }, + { + "epoch": 0.11, + "grad_norm": 0.2053105349872685, + "learning_rate": 2.926829268292683e-05, + "loss": 1.2094, + "step": 60 + }, + { + "epoch": 0.11, + "grad_norm": 0.17971910872957886, + "learning_rate": 2.9756097560975613e-05, + "loss": 1.228, + "step": 61 + }, + { + "epoch": 0.11, + "grad_norm": 0.1885853654992973, + "learning_rate": 3.0243902439024392e-05, + "loss": 1.2286, + "step": 62 + }, + { + "epoch": 0.12, + "grad_norm": 0.1848524571968613, + "learning_rate": 3.073170731707317e-05, + "loss": 1.2718, + "step": 63 + }, + { + "epoch": 0.12, + "grad_norm": 0.18734105883548513, + "learning_rate": 3.1219512195121955e-05, + "loss": 1.2357, + "step": 64 + }, + { + "epoch": 0.12, + "grad_norm": 0.17774668052121825, + "learning_rate": 3.170731707317074e-05, + "loss": 1.1509, + "step": 65 + }, + { + "epoch": 0.12, + "grad_norm": 0.17890968008080646, + "learning_rate": 3.2195121951219514e-05, + "loss": 1.1924, + "step": 66 + }, + { + "epoch": 0.12, + "grad_norm": 0.18249273371332375, + "learning_rate": 3.268292682926829e-05, + "loss": 1.2545, + "step": 67 + }, + { + "epoch": 0.12, + "grad_norm": 0.21064122671902577, + "learning_rate": 3.3170731707317074e-05, + "loss": 1.2832, + "step": 68 + }, + { + "epoch": 0.13, + "grad_norm": 0.1820064171955093, + "learning_rate": 3.365853658536586e-05, + "loss": 1.2071, + "step": 69 + }, + { + "epoch": 0.13, + "grad_norm": 0.16996662800553433, + "learning_rate": 3.414634146341463e-05, + "loss": 1.2073, + "step": 70 + }, + { + "epoch": 0.13, + "grad_norm": 0.1618669302922445, + "learning_rate": 3.4634146341463416e-05, + "loss": 1.1289, + "step": 71 + }, + { + "epoch": 0.13, + "grad_norm": 0.18948744950985544, + "learning_rate": 3.51219512195122e-05, + "loss": 1.2915, + "step": 72 + }, + { + "epoch": 0.13, + "grad_norm": 0.18326143691603383, + "learning_rate": 3.5609756097560976e-05, + "loss": 1.2238, + "step": 73 + }, + { + "epoch": 0.14, + "grad_norm": 0.17410704510700503, + "learning_rate": 3.609756097560976e-05, + "loss": 1.1784, + "step": 74 + }, + { + "epoch": 0.14, + "grad_norm": 0.1983667344995625, + "learning_rate": 3.658536585365854e-05, + "loss": 1.2452, + "step": 75 + }, + { + "epoch": 0.14, + "grad_norm": 0.3416310763369357, + "learning_rate": 3.7073170731707325e-05, + "loss": 1.1972, + "step": 76 + }, + { + "epoch": 0.14, + "grad_norm": 0.2776466983511955, + "learning_rate": 3.75609756097561e-05, + "loss": 1.3121, + "step": 77 + }, + { + "epoch": 0.14, + "grad_norm": 0.20026129636576834, + "learning_rate": 3.804878048780488e-05, + "loss": 1.2436, + "step": 78 + }, + { + "epoch": 0.14, + "grad_norm": 0.21064549243917835, + "learning_rate": 3.853658536585366e-05, + "loss": 1.2064, + "step": 79 + }, + { + "epoch": 0.15, + "grad_norm": 0.22119482175714267, + "learning_rate": 3.9024390243902444e-05, + "loss": 1.2715, + "step": 80 + }, + { + "epoch": 0.15, + "grad_norm": 0.23047133748844142, + "learning_rate": 3.951219512195122e-05, + "loss": 1.2888, + "step": 81 + }, + { + "epoch": 0.15, + "grad_norm": 0.18741863156973176, + "learning_rate": 4e-05, + "loss": 1.248, + "step": 82 + }, + { + "epoch": 0.15, + "grad_norm": 0.1747859810629604, + "learning_rate": 4.0487804878048786e-05, + "loss": 1.1683, + "step": 83 + }, + { + "epoch": 0.15, + "grad_norm": 0.1896944798413341, + "learning_rate": 4.097560975609756e-05, + "loss": 1.2155, + "step": 84 + }, + { + "epoch": 0.16, + "grad_norm": 0.18724128114363303, + "learning_rate": 4.1463414634146346e-05, + "loss": 1.2273, + "step": 85 + }, + { + "epoch": 0.16, + "grad_norm": 0.17368125504855478, + "learning_rate": 4.195121951219513e-05, + "loss": 1.224, + "step": 86 + }, + { + "epoch": 0.16, + "grad_norm": 0.18371141013625703, + "learning_rate": 4.2439024390243905e-05, + "loss": 1.2294, + "step": 87 + }, + { + "epoch": 0.16, + "grad_norm": 0.1791029365673714, + "learning_rate": 4.292682926829269e-05, + "loss": 1.2895, + "step": 88 + }, + { + "epoch": 0.16, + "grad_norm": 0.20259974283859655, + "learning_rate": 4.341463414634147e-05, + "loss": 1.1841, + "step": 89 + }, + { + "epoch": 0.16, + "grad_norm": 0.17457456183272174, + "learning_rate": 4.390243902439025e-05, + "loss": 1.2357, + "step": 90 + }, + { + "epoch": 0.17, + "grad_norm": 0.1815824380789748, + "learning_rate": 4.439024390243903e-05, + "loss": 1.2304, + "step": 91 + }, + { + "epoch": 0.17, + "grad_norm": 0.17566480599583392, + "learning_rate": 4.4878048780487814e-05, + "loss": 1.242, + "step": 92 + }, + { + "epoch": 0.17, + "grad_norm": 0.18422975005984474, + "learning_rate": 4.536585365853658e-05, + "loss": 1.2177, + "step": 93 + }, + { + "epoch": 0.17, + "grad_norm": 0.16796781877940678, + "learning_rate": 4.5853658536585366e-05, + "loss": 1.1482, + "step": 94 + }, + { + "epoch": 0.17, + "grad_norm": 0.18636131653783305, + "learning_rate": 4.634146341463415e-05, + "loss": 1.1758, + "step": 95 + }, + { + "epoch": 0.18, + "grad_norm": 0.1823665700289814, + "learning_rate": 4.6829268292682926e-05, + "loss": 1.289, + "step": 96 + }, + { + "epoch": 0.18, + "grad_norm": 0.1719900691262439, + "learning_rate": 4.731707317073171e-05, + "loss": 1.1626, + "step": 97 + }, + { + "epoch": 0.18, + "grad_norm": 0.17937994168039778, + "learning_rate": 4.780487804878049e-05, + "loss": 1.175, + "step": 98 + }, + { + "epoch": 0.18, + "grad_norm": 0.16631851422106986, + "learning_rate": 4.829268292682927e-05, + "loss": 1.2177, + "step": 99 + }, + { + "epoch": 0.18, + "grad_norm": 0.19143696232800309, + "learning_rate": 4.878048780487805e-05, + "loss": 1.3071, + "step": 100 + }, + { + "epoch": 0.18, + "grad_norm": 0.17859506638780318, + "learning_rate": 4.9268292682926835e-05, + "loss": 1.2351, + "step": 101 + }, + { + "epoch": 0.19, + "grad_norm": 0.18381520321248196, + "learning_rate": 4.975609756097561e-05, + "loss": 1.2342, + "step": 102 + }, + { + "epoch": 0.19, + "grad_norm": 0.17968218683773912, + "learning_rate": 5.0243902439024394e-05, + "loss": 1.2074, + "step": 103 + }, + { + "epoch": 0.19, + "grad_norm": 0.18139489969339018, + "learning_rate": 5.073170731707318e-05, + "loss": 1.1558, + "step": 104 + }, + { + "epoch": 0.19, + "grad_norm": 0.17366624842514394, + "learning_rate": 5.121951219512195e-05, + "loss": 1.1897, + "step": 105 + }, + { + "epoch": 0.19, + "grad_norm": 0.16034845455223745, + "learning_rate": 5.1707317073170736e-05, + "loss": 1.179, + "step": 106 + }, + { + "epoch": 0.2, + "grad_norm": 0.17583069577827776, + "learning_rate": 5.219512195121952e-05, + "loss": 1.1856, + "step": 107 + }, + { + "epoch": 0.2, + "grad_norm": 0.1853758076989552, + "learning_rate": 5.26829268292683e-05, + "loss": 1.2072, + "step": 108 + }, + { + "epoch": 0.2, + "grad_norm": 0.19597443965936462, + "learning_rate": 5.317073170731708e-05, + "loss": 1.2271, + "step": 109 + }, + { + "epoch": 0.2, + "grad_norm": 0.1899206334098331, + "learning_rate": 5.365853658536586e-05, + "loss": 1.1961, + "step": 110 + }, + { + "epoch": 0.2, + "grad_norm": 0.17463763837757018, + "learning_rate": 5.4146341463414645e-05, + "loss": 1.2049, + "step": 111 + }, + { + "epoch": 0.2, + "grad_norm": 0.20431371701229986, + "learning_rate": 5.463414634146342e-05, + "loss": 1.2891, + "step": 112 + }, + { + "epoch": 0.21, + "grad_norm": 0.1814475107638498, + "learning_rate": 5.51219512195122e-05, + "loss": 1.2346, + "step": 113 + }, + { + "epoch": 0.21, + "grad_norm": 0.1883849423207823, + "learning_rate": 5.5609756097560974e-05, + "loss": 1.244, + "step": 114 + }, + { + "epoch": 0.21, + "grad_norm": 0.1857258128640568, + "learning_rate": 5.609756097560976e-05, + "loss": 1.2669, + "step": 115 + }, + { + "epoch": 0.21, + "grad_norm": 0.1740768514118401, + "learning_rate": 5.658536585365854e-05, + "loss": 1.2414, + "step": 116 + }, + { + "epoch": 0.21, + "grad_norm": 0.1919320335584178, + "learning_rate": 5.7073170731707317e-05, + "loss": 1.2886, + "step": 117 + }, + { + "epoch": 0.22, + "grad_norm": 0.18288775167828136, + "learning_rate": 5.75609756097561e-05, + "loss": 1.1875, + "step": 118 + }, + { + "epoch": 0.22, + "grad_norm": 0.18208588867750863, + "learning_rate": 5.804878048780488e-05, + "loss": 1.2388, + "step": 119 + }, + { + "epoch": 0.22, + "grad_norm": 0.1743260015658331, + "learning_rate": 5.853658536585366e-05, + "loss": 1.1762, + "step": 120 + }, + { + "epoch": 0.22, + "grad_norm": 0.17856046291517946, + "learning_rate": 5.902439024390244e-05, + "loss": 1.2888, + "step": 121 + }, + { + "epoch": 0.22, + "grad_norm": 0.17493794870966536, + "learning_rate": 5.9512195121951225e-05, + "loss": 1.2222, + "step": 122 + }, + { + "epoch": 0.22, + "grad_norm": 0.1909202655203384, + "learning_rate": 6.000000000000001e-05, + "loss": 1.2414, + "step": 123 + }, + { + "epoch": 0.23, + "grad_norm": 0.18345819482834988, + "learning_rate": 6.0487804878048785e-05, + "loss": 1.2756, + "step": 124 + }, + { + "epoch": 0.23, + "grad_norm": 0.2057069352956621, + "learning_rate": 6.097560975609757e-05, + "loss": 1.261, + "step": 125 + }, + { + "epoch": 0.23, + "grad_norm": 0.299775882469108, + "learning_rate": 6.146341463414634e-05, + "loss": 1.2566, + "step": 126 + }, + { + "epoch": 0.23, + "grad_norm": 0.1869687633018095, + "learning_rate": 6.195121951219513e-05, + "loss": 1.3039, + "step": 127 + }, + { + "epoch": 0.23, + "grad_norm": 0.17747149926197442, + "learning_rate": 6.243902439024391e-05, + "loss": 1.2524, + "step": 128 + }, + { + "epoch": 0.24, + "grad_norm": 0.17885157788044242, + "learning_rate": 6.29268292682927e-05, + "loss": 1.2455, + "step": 129 + }, + { + "epoch": 0.24, + "grad_norm": 0.17617298187845123, + "learning_rate": 6.341463414634148e-05, + "loss": 1.2009, + "step": 130 + }, + { + "epoch": 0.24, + "grad_norm": 0.20164176323497066, + "learning_rate": 6.390243902439025e-05, + "loss": 1.2634, + "step": 131 + }, + { + "epoch": 0.24, + "grad_norm": 0.20459903417307612, + "learning_rate": 6.439024390243903e-05, + "loss": 1.1963, + "step": 132 + }, + { + "epoch": 0.24, + "grad_norm": 0.1863755486334296, + "learning_rate": 6.487804878048781e-05, + "loss": 1.2387, + "step": 133 + }, + { + "epoch": 0.25, + "grad_norm": 0.19265866140295207, + "learning_rate": 6.536585365853658e-05, + "loss": 1.2688, + "step": 134 + }, + { + "epoch": 0.25, + "grad_norm": 0.1823425868969493, + "learning_rate": 6.585365853658536e-05, + "loss": 1.2041, + "step": 135 + }, + { + "epoch": 0.25, + "grad_norm": 0.2016853266472781, + "learning_rate": 6.634146341463415e-05, + "loss": 1.1223, + "step": 136 + }, + { + "epoch": 0.25, + "grad_norm": 0.17282675192463448, + "learning_rate": 6.682926829268293e-05, + "loss": 1.1879, + "step": 137 + }, + { + "epoch": 0.25, + "grad_norm": 0.17398811693399288, + "learning_rate": 6.731707317073171e-05, + "loss": 1.2682, + "step": 138 + }, + { + "epoch": 0.25, + "grad_norm": 0.18516916965434696, + "learning_rate": 6.78048780487805e-05, + "loss": 1.1666, + "step": 139 + }, + { + "epoch": 0.26, + "grad_norm": 0.1852213129647933, + "learning_rate": 6.829268292682927e-05, + "loss": 1.2501, + "step": 140 + }, + { + "epoch": 0.26, + "grad_norm": 0.17915948766591883, + "learning_rate": 6.878048780487805e-05, + "loss": 1.2264, + "step": 141 + }, + { + "epoch": 0.26, + "grad_norm": 0.21599939417233183, + "learning_rate": 6.926829268292683e-05, + "loss": 1.2376, + "step": 142 + }, + { + "epoch": 0.26, + "grad_norm": 0.17839304459521851, + "learning_rate": 6.975609756097562e-05, + "loss": 1.2353, + "step": 143 + }, + { + "epoch": 0.26, + "grad_norm": 0.20826913231380875, + "learning_rate": 7.02439024390244e-05, + "loss": 1.1901, + "step": 144 + }, + { + "epoch": 0.27, + "grad_norm": 0.20788894913361589, + "learning_rate": 7.073170731707318e-05, + "loss": 1.2577, + "step": 145 + }, + { + "epoch": 0.27, + "grad_norm": 0.18420055842301297, + "learning_rate": 7.121951219512195e-05, + "loss": 1.1393, + "step": 146 + }, + { + "epoch": 0.27, + "grad_norm": 0.19903048468685589, + "learning_rate": 7.170731707317073e-05, + "loss": 1.2321, + "step": 147 + }, + { + "epoch": 0.27, + "grad_norm": 0.19074116314985748, + "learning_rate": 7.219512195121952e-05, + "loss": 1.1912, + "step": 148 + }, + { + "epoch": 0.27, + "grad_norm": 0.2353816469403903, + "learning_rate": 7.26829268292683e-05, + "loss": 1.28, + "step": 149 + }, + { + "epoch": 0.27, + "grad_norm": 0.21634875684769345, + "learning_rate": 7.317073170731708e-05, + "loss": 1.3312, + "step": 150 + }, + { + "epoch": 0.28, + "grad_norm": 0.18290969006743918, + "learning_rate": 7.365853658536587e-05, + "loss": 1.2214, + "step": 151 + }, + { + "epoch": 0.28, + "grad_norm": 0.18484243897545208, + "learning_rate": 7.414634146341465e-05, + "loss": 1.1895, + "step": 152 + }, + { + "epoch": 0.28, + "grad_norm": 0.21882343112978872, + "learning_rate": 7.463414634146342e-05, + "loss": 1.2219, + "step": 153 + }, + { + "epoch": 0.28, + "grad_norm": 0.19868284379241205, + "learning_rate": 7.51219512195122e-05, + "loss": 1.2176, + "step": 154 + }, + { + "epoch": 0.28, + "grad_norm": 0.20912516312950613, + "learning_rate": 7.560975609756097e-05, + "loss": 1.242, + "step": 155 + }, + { + "epoch": 0.29, + "grad_norm": 0.23811880045549916, + "learning_rate": 7.609756097560976e-05, + "loss": 1.2838, + "step": 156 + }, + { + "epoch": 0.29, + "grad_norm": 0.19511077122033713, + "learning_rate": 7.658536585365854e-05, + "loss": 1.1594, + "step": 157 + }, + { + "epoch": 0.29, + "grad_norm": 0.20094129399534238, + "learning_rate": 7.707317073170732e-05, + "loss": 1.2966, + "step": 158 + }, + { + "epoch": 0.29, + "grad_norm": 0.19366245038292418, + "learning_rate": 7.75609756097561e-05, + "loss": 1.2246, + "step": 159 + }, + { + "epoch": 0.29, + "grad_norm": 0.19409570223867306, + "learning_rate": 7.804878048780489e-05, + "loss": 1.2312, + "step": 160 + }, + { + "epoch": 0.29, + "grad_norm": 0.2087258457033805, + "learning_rate": 7.853658536585366e-05, + "loss": 1.2169, + "step": 161 + }, + { + "epoch": 0.3, + "grad_norm": 0.18765223996270428, + "learning_rate": 7.902439024390244e-05, + "loss": 1.2383, + "step": 162 + }, + { + "epoch": 0.3, + "grad_norm": 0.20734180224147242, + "learning_rate": 7.951219512195122e-05, + "loss": 1.2587, + "step": 163 + }, + { + "epoch": 0.3, + "grad_norm": 0.24690929540287834, + "learning_rate": 8e-05, + "loss": 1.1951, + "step": 164 + }, + { + "epoch": 0.3, + "grad_norm": 0.2003538797619543, + "learning_rate": 7.999990914797545e-05, + "loss": 1.1982, + "step": 165 + }, + { + "epoch": 0.3, + "grad_norm": 0.22469075613510484, + "learning_rate": 7.99996365923145e-05, + "loss": 1.2355, + "step": 166 + }, + { + "epoch": 0.31, + "grad_norm": 0.21870100788336058, + "learning_rate": 7.999918233425526e-05, + "loss": 1.1103, + "step": 167 + }, + { + "epoch": 0.31, + "grad_norm": 0.20939989594131886, + "learning_rate": 7.999854637586122e-05, + "loss": 1.1966, + "step": 168 + }, + { + "epoch": 0.31, + "grad_norm": 0.43108211416237796, + "learning_rate": 7.999772872002132e-05, + "loss": 1.2882, + "step": 169 + }, + { + "epoch": 0.31, + "grad_norm": 0.27045413432174487, + "learning_rate": 7.999672937044984e-05, + "loss": 1.2399, + "step": 170 + }, + { + "epoch": 0.31, + "grad_norm": 0.19700483036740515, + "learning_rate": 7.999554833168642e-05, + "loss": 1.202, + "step": 171 + }, + { + "epoch": 0.31, + "grad_norm": 0.3335979493370708, + "learning_rate": 7.999418560909604e-05, + "loss": 1.1995, + "step": 172 + }, + { + "epoch": 0.32, + "grad_norm": 0.3165803974474567, + "learning_rate": 7.999264120886902e-05, + "loss": 1.1569, + "step": 173 + }, + { + "epoch": 0.32, + "grad_norm": 0.1951699080346223, + "learning_rate": 7.999091513802093e-05, + "loss": 1.1778, + "step": 174 + }, + { + "epoch": 0.32, + "grad_norm": 0.2087559121749787, + "learning_rate": 7.998900740439265e-05, + "loss": 1.1736, + "step": 175 + }, + { + "epoch": 0.32, + "grad_norm": 0.20345180977460478, + "learning_rate": 7.998691801665024e-05, + "loss": 1.2281, + "step": 176 + }, + { + "epoch": 0.32, + "grad_norm": 0.24617644827252333, + "learning_rate": 7.998464698428495e-05, + "loss": 1.2072, + "step": 177 + }, + { + "epoch": 0.33, + "grad_norm": 0.2469050959356265, + "learning_rate": 7.998219431761318e-05, + "loss": 1.2242, + "step": 178 + }, + { + "epoch": 0.33, + "grad_norm": 0.19529317748460623, + "learning_rate": 7.997956002777642e-05, + "loss": 1.2567, + "step": 179 + }, + { + "epoch": 0.33, + "grad_norm": 0.19048389491381376, + "learning_rate": 7.99767441267412e-05, + "loss": 1.2982, + "step": 180 + }, + { + "epoch": 0.33, + "grad_norm": 0.2085799116493225, + "learning_rate": 7.997374662729904e-05, + "loss": 1.1254, + "step": 181 + }, + { + "epoch": 0.33, + "grad_norm": 0.20636853256378995, + "learning_rate": 7.997056754306636e-05, + "loss": 1.2435, + "step": 182 + }, + { + "epoch": 0.33, + "grad_norm": 0.20590016382290252, + "learning_rate": 7.99672068884845e-05, + "loss": 1.2658, + "step": 183 + }, + { + "epoch": 0.34, + "grad_norm": 0.1931166169764433, + "learning_rate": 7.996366467881955e-05, + "loss": 1.1637, + "step": 184 + }, + { + "epoch": 0.34, + "grad_norm": 0.18873318157988098, + "learning_rate": 7.995994093016237e-05, + "loss": 1.1335, + "step": 185 + }, + { + "epoch": 0.34, + "grad_norm": 0.19210254625199108, + "learning_rate": 7.995603565942846e-05, + "loss": 1.1928, + "step": 186 + }, + { + "epoch": 0.34, + "grad_norm": 0.2130986479765664, + "learning_rate": 7.995194888435792e-05, + "loss": 1.2158, + "step": 187 + }, + { + "epoch": 0.34, + "grad_norm": 0.22003854501814088, + "learning_rate": 7.994768062351532e-05, + "loss": 1.2288, + "step": 188 + }, + { + "epoch": 0.35, + "grad_norm": 0.20330803191993058, + "learning_rate": 7.994323089628968e-05, + "loss": 1.2426, + "step": 189 + }, + { + "epoch": 0.35, + "grad_norm": 0.20567314642208634, + "learning_rate": 7.993859972289434e-05, + "loss": 1.2649, + "step": 190 + }, + { + "epoch": 0.35, + "grad_norm": 0.21556663727342962, + "learning_rate": 7.993378712436686e-05, + "loss": 1.2545, + "step": 191 + }, + { + "epoch": 0.35, + "grad_norm": 0.20309165469109888, + "learning_rate": 7.992879312256897e-05, + "loss": 1.3338, + "step": 192 + }, + { + "epoch": 0.35, + "grad_norm": 0.19574356669421325, + "learning_rate": 7.992361774018641e-05, + "loss": 1.278, + "step": 193 + }, + { + "epoch": 0.35, + "grad_norm": 0.2763613746722313, + "learning_rate": 7.991826100072891e-05, + "loss": 1.2571, + "step": 194 + }, + { + "epoch": 0.36, + "grad_norm": 0.19346552479915102, + "learning_rate": 7.991272292852996e-05, + "loss": 1.2027, + "step": 195 + }, + { + "epoch": 0.36, + "grad_norm": 0.2281167812123908, + "learning_rate": 7.990700354874683e-05, + "loss": 1.2586, + "step": 196 + }, + { + "epoch": 0.36, + "grad_norm": 0.19699013712137542, + "learning_rate": 7.990110288736042e-05, + "loss": 1.1371, + "step": 197 + }, + { + "epoch": 0.36, + "grad_norm": 0.21768209981475933, + "learning_rate": 7.989502097117503e-05, + "loss": 1.2522, + "step": 198 + }, + { + "epoch": 0.36, + "grad_norm": 0.21335427847754582, + "learning_rate": 7.988875782781838e-05, + "loss": 1.2437, + "step": 199 + }, + { + "epoch": 0.37, + "grad_norm": 0.21856710629066897, + "learning_rate": 7.988231348574147e-05, + "loss": 1.2135, + "step": 200 + }, + { + "epoch": 0.37, + "grad_norm": 0.20482062658774797, + "learning_rate": 7.987568797421836e-05, + "loss": 1.1755, + "step": 201 + }, + { + "epoch": 0.37, + "grad_norm": 0.2017756813960897, + "learning_rate": 7.986888132334608e-05, + "loss": 1.1699, + "step": 202 + }, + { + "epoch": 0.37, + "grad_norm": 0.20496443848153809, + "learning_rate": 7.986189356404458e-05, + "loss": 1.2125, + "step": 203 + }, + { + "epoch": 0.37, + "grad_norm": 0.2134603800558358, + "learning_rate": 7.985472472805643e-05, + "loss": 1.2391, + "step": 204 + }, + { + "epoch": 0.37, + "grad_norm": 0.2364175573420861, + "learning_rate": 7.98473748479468e-05, + "loss": 1.2384, + "step": 205 + }, + { + "epoch": 0.38, + "grad_norm": 0.1872419861598724, + "learning_rate": 7.983984395710326e-05, + "loss": 1.1457, + "step": 206 + }, + { + "epoch": 0.38, + "grad_norm": 0.28222194007095774, + "learning_rate": 7.983213208973566e-05, + "loss": 1.2952, + "step": 207 + }, + { + "epoch": 0.38, + "grad_norm": 0.1916094851162064, + "learning_rate": 7.982423928087593e-05, + "loss": 1.1763, + "step": 208 + }, + { + "epoch": 0.38, + "grad_norm": 0.18446245256166657, + "learning_rate": 7.981616556637795e-05, + "loss": 1.1863, + "step": 209 + }, + { + "epoch": 0.38, + "grad_norm": 0.195191961022491, + "learning_rate": 7.980791098291737e-05, + "loss": 1.2036, + "step": 210 + }, + { + "epoch": 0.39, + "grad_norm": 0.2652439657825496, + "learning_rate": 7.979947556799151e-05, + "loss": 1.2834, + "step": 211 + }, + { + "epoch": 0.39, + "grad_norm": 0.24308438957843412, + "learning_rate": 7.979085935991906e-05, + "loss": 1.234, + "step": 212 + }, + { + "epoch": 0.39, + "grad_norm": 0.21294701043622016, + "learning_rate": 7.978206239784004e-05, + "loss": 1.3006, + "step": 213 + }, + { + "epoch": 0.39, + "grad_norm": 0.25809277041859524, + "learning_rate": 7.977308472171553e-05, + "loss": 1.2272, + "step": 214 + }, + { + "epoch": 0.39, + "grad_norm": 0.193463860107294, + "learning_rate": 7.976392637232754e-05, + "loss": 1.2295, + "step": 215 + }, + { + "epoch": 0.4, + "grad_norm": 0.2150023760609626, + "learning_rate": 7.975458739127877e-05, + "loss": 1.2135, + "step": 216 + }, + { + "epoch": 0.4, + "grad_norm": 0.22590495955605894, + "learning_rate": 7.974506782099253e-05, + "loss": 1.2532, + "step": 217 + }, + { + "epoch": 0.4, + "grad_norm": 0.21023744668403702, + "learning_rate": 7.973536770471242e-05, + "loss": 1.2472, + "step": 218 + }, + { + "epoch": 0.4, + "grad_norm": 0.2345749799511543, + "learning_rate": 7.972548708650218e-05, + "loss": 1.1791, + "step": 219 + }, + { + "epoch": 0.4, + "grad_norm": 0.2158876734005217, + "learning_rate": 7.971542601124553e-05, + "loss": 1.2483, + "step": 220 + }, + { + "epoch": 0.4, + "grad_norm": 0.29455339949432446, + "learning_rate": 7.970518452464593e-05, + "loss": 1.2894, + "step": 221 + }, + { + "epoch": 0.41, + "grad_norm": 0.23983708730626851, + "learning_rate": 7.969476267322636e-05, + "loss": 1.271, + "step": 222 + }, + { + "epoch": 0.41, + "grad_norm": 0.1922400905426158, + "learning_rate": 7.968416050432912e-05, + "loss": 1.2139, + "step": 223 + }, + { + "epoch": 0.41, + "grad_norm": 0.2238136844422931, + "learning_rate": 7.967337806611568e-05, + "loss": 1.2655, + "step": 224 + }, + { + "epoch": 0.41, + "grad_norm": 0.21230292828267672, + "learning_rate": 7.966241540756631e-05, + "loss": 1.2406, + "step": 225 + }, + { + "epoch": 0.41, + "grad_norm": 0.26656119419070456, + "learning_rate": 7.965127257848004e-05, + "loss": 1.2595, + "step": 226 + }, + { + "epoch": 0.42, + "grad_norm": 0.22381385502992684, + "learning_rate": 7.963994962947426e-05, + "loss": 1.1737, + "step": 227 + }, + { + "epoch": 0.42, + "grad_norm": 0.20056702203994298, + "learning_rate": 7.962844661198462e-05, + "loss": 1.1969, + "step": 228 + }, + { + "epoch": 0.42, + "grad_norm": 0.20148701321526885, + "learning_rate": 7.961676357826478e-05, + "loss": 1.2151, + "step": 229 + }, + { + "epoch": 0.42, + "grad_norm": 0.20034834807028637, + "learning_rate": 7.960490058138604e-05, + "loss": 1.1455, + "step": 230 + }, + { + "epoch": 0.42, + "grad_norm": 0.21050838521846033, + "learning_rate": 7.959285767523732e-05, + "loss": 1.2223, + "step": 231 + }, + { + "epoch": 0.42, + "grad_norm": 0.20904772138969777, + "learning_rate": 7.95806349145247e-05, + "loss": 1.2534, + "step": 232 + }, + { + "epoch": 0.43, + "grad_norm": 0.20307877304792957, + "learning_rate": 7.956823235477134e-05, + "loss": 1.1352, + "step": 233 + }, + { + "epoch": 0.43, + "grad_norm": 0.20501105270897094, + "learning_rate": 7.95556500523171e-05, + "loss": 1.2031, + "step": 234 + }, + { + "epoch": 0.43, + "grad_norm": 0.19800586972038586, + "learning_rate": 7.954288806431838e-05, + "loss": 1.2567, + "step": 235 + }, + { + "epoch": 0.43, + "grad_norm": 0.2175102450594135, + "learning_rate": 7.952994644874777e-05, + "loss": 1.2538, + "step": 236 + }, + { + "epoch": 0.43, + "grad_norm": 0.22698189300067595, + "learning_rate": 7.951682526439391e-05, + "loss": 1.3088, + "step": 237 + }, + { + "epoch": 0.44, + "grad_norm": 0.19208392014975315, + "learning_rate": 7.950352457086109e-05, + "loss": 1.2336, + "step": 238 + }, + { + "epoch": 0.44, + "grad_norm": 0.27004086334319655, + "learning_rate": 7.949004442856905e-05, + "loss": 1.2012, + "step": 239 + }, + { + "epoch": 0.44, + "grad_norm": 0.23420974954538043, + "learning_rate": 7.947638489875272e-05, + "loss": 1.2244, + "step": 240 + }, + { + "epoch": 0.44, + "grad_norm": 0.20514399124802024, + "learning_rate": 7.946254604346186e-05, + "loss": 1.2548, + "step": 241 + }, + { + "epoch": 0.44, + "grad_norm": 0.19334973602372896, + "learning_rate": 7.944852792556092e-05, + "loss": 1.2104, + "step": 242 + }, + { + "epoch": 0.44, + "grad_norm": 0.1992640714537956, + "learning_rate": 7.943433060872858e-05, + "loss": 1.2628, + "step": 243 + }, + { + "epoch": 0.45, + "grad_norm": 0.203284617090413, + "learning_rate": 7.941995415745761e-05, + "loss": 1.2002, + "step": 244 + }, + { + "epoch": 0.45, + "grad_norm": 0.22795306969682058, + "learning_rate": 7.94053986370545e-05, + "loss": 1.2215, + "step": 245 + }, + { + "epoch": 0.45, + "grad_norm": 0.20789041346838505, + "learning_rate": 7.939066411363915e-05, + "loss": 1.0998, + "step": 246 + }, + { + "epoch": 0.45, + "grad_norm": 0.22354868884742066, + "learning_rate": 7.937575065414464e-05, + "loss": 1.2564, + "step": 247 + }, + { + "epoch": 0.45, + "grad_norm": 0.21176392726647736, + "learning_rate": 7.936065832631687e-05, + "loss": 1.2816, + "step": 248 + }, + { + "epoch": 0.46, + "grad_norm": 0.19967179557235587, + "learning_rate": 7.934538719871427e-05, + "loss": 1.1961, + "step": 249 + }, + { + "epoch": 0.46, + "grad_norm": 0.210819577350627, + "learning_rate": 7.932993734070747e-05, + "loss": 1.2167, + "step": 250 + }, + { + "epoch": 0.46, + "grad_norm": 0.21537794551756187, + "learning_rate": 7.931430882247903e-05, + "loss": 1.2341, + "step": 251 + }, + { + "epoch": 0.46, + "grad_norm": 0.22850872387256574, + "learning_rate": 7.929850171502304e-05, + "loss": 1.1686, + "step": 252 + }, + { + "epoch": 0.46, + "grad_norm": 0.22380366415076383, + "learning_rate": 7.928251609014493e-05, + "loss": 1.1462, + "step": 253 + }, + { + "epoch": 0.46, + "grad_norm": 0.22426923149036065, + "learning_rate": 7.926635202046102e-05, + "loss": 1.1792, + "step": 254 + }, + { + "epoch": 0.47, + "grad_norm": 0.42082703321103965, + "learning_rate": 7.925000957939822e-05, + "loss": 1.2718, + "step": 255 + }, + { + "epoch": 0.47, + "grad_norm": 0.2235432774854074, + "learning_rate": 7.92334888411937e-05, + "loss": 1.2598, + "step": 256 + }, + { + "epoch": 0.47, + "grad_norm": 0.281644028934108, + "learning_rate": 7.92167898808946e-05, + "loss": 1.2205, + "step": 257 + }, + { + "epoch": 0.47, + "grad_norm": 0.2037705143888748, + "learning_rate": 7.919991277435763e-05, + "loss": 1.1737, + "step": 258 + }, + { + "epoch": 0.47, + "grad_norm": 0.20917419230028977, + "learning_rate": 7.918285759824879e-05, + "loss": 1.2035, + "step": 259 + }, + { + "epoch": 0.48, + "grad_norm": 0.20510847570635518, + "learning_rate": 7.916562443004292e-05, + "loss": 1.2135, + "step": 260 + }, + { + "epoch": 0.48, + "grad_norm": 0.25172483071092466, + "learning_rate": 7.914821334802342e-05, + "loss": 1.2218, + "step": 261 + }, + { + "epoch": 0.48, + "grad_norm": 0.21102706700634313, + "learning_rate": 7.91306244312819e-05, + "loss": 1.1738, + "step": 262 + }, + { + "epoch": 0.48, + "grad_norm": 0.22626060872645815, + "learning_rate": 7.911285775971781e-05, + "loss": 1.238, + "step": 263 + }, + { + "epoch": 0.48, + "grad_norm": 0.22448567539778486, + "learning_rate": 7.909491341403805e-05, + "loss": 1.2404, + "step": 264 + }, + { + "epoch": 0.48, + "grad_norm": 0.2019099786139193, + "learning_rate": 7.907679147575661e-05, + "loss": 1.213, + "step": 265 + }, + { + "epoch": 0.49, + "grad_norm": 0.24307234839096267, + "learning_rate": 7.905849202719422e-05, + "loss": 1.2322, + "step": 266 + }, + { + "epoch": 0.49, + "grad_norm": 0.19801890521743487, + "learning_rate": 7.904001515147802e-05, + "loss": 1.2448, + "step": 267 + }, + { + "epoch": 0.49, + "grad_norm": 0.2102742273575385, + "learning_rate": 7.902136093254106e-05, + "loss": 1.1657, + "step": 268 + }, + { + "epoch": 0.49, + "grad_norm": 0.2173464476815016, + "learning_rate": 7.900252945512201e-05, + "loss": 1.2549, + "step": 269 + }, + { + "epoch": 0.49, + "grad_norm": 0.20957275458699595, + "learning_rate": 7.898352080476479e-05, + "loss": 1.2536, + "step": 270 + }, + { + "epoch": 0.5, + "grad_norm": 0.20691966388952363, + "learning_rate": 7.896433506781811e-05, + "loss": 1.2661, + "step": 271 + }, + { + "epoch": 0.5, + "grad_norm": 0.2276662275112648, + "learning_rate": 7.894497233143509e-05, + "loss": 1.2409, + "step": 272 + }, + { + "epoch": 0.5, + "grad_norm": 0.23854109569301263, + "learning_rate": 7.892543268357297e-05, + "loss": 1.2681, + "step": 273 + }, + { + "epoch": 0.5, + "grad_norm": 0.2233864156677627, + "learning_rate": 7.890571621299252e-05, + "loss": 1.1687, + "step": 274 + }, + { + "epoch": 0.5, + "grad_norm": 0.20114129147925475, + "learning_rate": 7.888582300925787e-05, + "loss": 1.2184, + "step": 275 + }, + { + "epoch": 0.5, + "grad_norm": 0.2154654670569462, + "learning_rate": 7.886575316273586e-05, + "loss": 1.1982, + "step": 276 + }, + { + "epoch": 0.51, + "grad_norm": 0.2292982209343639, + "learning_rate": 7.884550676459583e-05, + "loss": 1.2129, + "step": 277 + }, + { + "epoch": 0.51, + "grad_norm": 0.21302713135229548, + "learning_rate": 7.882508390680908e-05, + "loss": 1.1605, + "step": 278 + }, + { + "epoch": 0.51, + "grad_norm": 0.2123661020671048, + "learning_rate": 7.88044846821485e-05, + "loss": 1.2308, + "step": 279 + }, + { + "epoch": 0.51, + "grad_norm": 0.2080577410800404, + "learning_rate": 7.878370918418818e-05, + "loss": 1.2195, + "step": 280 + }, + { + "epoch": 0.51, + "grad_norm": 0.19663901881127385, + "learning_rate": 7.876275750730289e-05, + "loss": 1.1591, + "step": 281 + }, + { + "epoch": 0.52, + "grad_norm": 0.20534502031312163, + "learning_rate": 7.874162974666776e-05, + "loss": 1.2664, + "step": 282 + }, + { + "epoch": 0.52, + "grad_norm": 0.23240445399513837, + "learning_rate": 7.872032599825779e-05, + "loss": 1.2151, + "step": 283 + }, + { + "epoch": 0.52, + "grad_norm": 0.2672527316717507, + "learning_rate": 7.86988463588474e-05, + "loss": 1.2406, + "step": 284 + }, + { + "epoch": 0.52, + "grad_norm": 0.19893903058743695, + "learning_rate": 7.867719092601003e-05, + "loss": 1.1291, + "step": 285 + }, + { + "epoch": 0.52, + "grad_norm": 0.33275268109930917, + "learning_rate": 7.865535979811768e-05, + "loss": 1.1406, + "step": 286 + }, + { + "epoch": 0.52, + "grad_norm": 0.2373619455690358, + "learning_rate": 7.863335307434045e-05, + "loss": 1.2799, + "step": 287 + }, + { + "epoch": 0.53, + "grad_norm": 0.263235735390858, + "learning_rate": 7.861117085464612e-05, + "loss": 1.2415, + "step": 288 + }, + { + "epoch": 0.53, + "grad_norm": 0.25884281780784324, + "learning_rate": 7.858881323979965e-05, + "loss": 1.3919, + "step": 289 + }, + { + "epoch": 0.53, + "grad_norm": 0.25426288332255736, + "learning_rate": 7.85662803313628e-05, + "loss": 1.174, + "step": 290 + }, + { + "epoch": 0.53, + "grad_norm": 0.26655405527881243, + "learning_rate": 7.854357223169356e-05, + "loss": 1.2806, + "step": 291 + }, + { + "epoch": 0.53, + "grad_norm": 0.20909844432349833, + "learning_rate": 7.852068904394579e-05, + "loss": 1.2627, + "step": 292 + }, + { + "epoch": 0.54, + "grad_norm": 0.21307115068935759, + "learning_rate": 7.849763087206866e-05, + "loss": 1.1879, + "step": 293 + }, + { + "epoch": 0.54, + "grad_norm": 0.25009949471398946, + "learning_rate": 7.847439782080628e-05, + "loss": 1.2881, + "step": 294 + }, + { + "epoch": 0.54, + "grad_norm": 0.20960783418679174, + "learning_rate": 7.845098999569712e-05, + "loss": 1.2723, + "step": 295 + }, + { + "epoch": 0.54, + "grad_norm": 0.24968832437925104, + "learning_rate": 7.842740750307362e-05, + "loss": 1.2029, + "step": 296 + }, + { + "epoch": 0.54, + "grad_norm": 0.22981196585125677, + "learning_rate": 7.84036504500616e-05, + "loss": 1.1695, + "step": 297 + }, + { + "epoch": 0.55, + "grad_norm": 0.2320606844751365, + "learning_rate": 7.837971894457991e-05, + "loss": 1.2317, + "step": 298 + }, + { + "epoch": 0.55, + "grad_norm": 0.23051459673906124, + "learning_rate": 7.835561309533981e-05, + "loss": 1.2046, + "step": 299 + }, + { + "epoch": 0.55, + "grad_norm": 0.2510027231060586, + "learning_rate": 7.833133301184457e-05, + "loss": 1.199, + "step": 300 + }, + { + "epoch": 0.55, + "grad_norm": 0.23601180466018787, + "learning_rate": 7.830687880438895e-05, + "loss": 1.1755, + "step": 301 + }, + { + "epoch": 0.55, + "grad_norm": 0.24740820934385369, + "learning_rate": 7.828225058405864e-05, + "loss": 1.2054, + "step": 302 + }, + { + "epoch": 0.55, + "grad_norm": 0.23065372979111173, + "learning_rate": 7.825744846272984e-05, + "loss": 1.2066, + "step": 303 + }, + { + "epoch": 0.56, + "grad_norm": 0.22385077334838213, + "learning_rate": 7.823247255306866e-05, + "loss": 1.2147, + "step": 304 + }, + { + "epoch": 0.56, + "grad_norm": 0.42981213948386104, + "learning_rate": 7.820732296853074e-05, + "loss": 1.2314, + "step": 305 + }, + { + "epoch": 0.56, + "grad_norm": 0.21122844902751076, + "learning_rate": 7.818199982336058e-05, + "loss": 1.1462, + "step": 306 + }, + { + "epoch": 0.56, + "grad_norm": 0.23374869692118933, + "learning_rate": 7.815650323259117e-05, + "loss": 1.2051, + "step": 307 + }, + { + "epoch": 0.56, + "grad_norm": 0.21662363795962128, + "learning_rate": 7.813083331204332e-05, + "loss": 1.1575, + "step": 308 + }, + { + "epoch": 0.57, + "grad_norm": 0.2088315773384112, + "learning_rate": 7.810499017832526e-05, + "loss": 1.1316, + "step": 309 + }, + { + "epoch": 0.57, + "grad_norm": 0.2095238410730976, + "learning_rate": 7.807897394883203e-05, + "loss": 1.2087, + "step": 310 + }, + { + "epoch": 0.57, + "grad_norm": 0.22672932127256515, + "learning_rate": 7.805278474174499e-05, + "loss": 1.2512, + "step": 311 + }, + { + "epoch": 0.57, + "grad_norm": 0.21873052340922736, + "learning_rate": 7.802642267603126e-05, + "loss": 1.1909, + "step": 312 + }, + { + "epoch": 0.57, + "grad_norm": 0.219814521916342, + "learning_rate": 7.79998878714432e-05, + "loss": 1.1669, + "step": 313 + }, + { + "epoch": 0.57, + "grad_norm": 0.3049426027257317, + "learning_rate": 7.797318044851786e-05, + "loss": 1.1797, + "step": 314 + }, + { + "epoch": 0.58, + "grad_norm": 0.22309435690065985, + "learning_rate": 7.794630052857638e-05, + "loss": 1.1417, + "step": 315 + }, + { + "epoch": 0.58, + "grad_norm": 0.3891885169154885, + "learning_rate": 7.791924823372354e-05, + "loss": 1.2369, + "step": 316 + }, + { + "epoch": 0.58, + "grad_norm": 0.24780269452456372, + "learning_rate": 7.789202368684711e-05, + "loss": 1.2521, + "step": 317 + }, + { + "epoch": 0.58, + "grad_norm": 0.21660460720269362, + "learning_rate": 7.786462701161738e-05, + "loss": 1.2151, + "step": 318 + }, + { + "epoch": 0.58, + "grad_norm": 0.23635409466561857, + "learning_rate": 7.783705833248649e-05, + "loss": 1.2363, + "step": 319 + }, + { + "epoch": 0.59, + "grad_norm": 0.2616135839903218, + "learning_rate": 7.780931777468797e-05, + "loss": 1.2428, + "step": 320 + }, + { + "epoch": 0.59, + "grad_norm": 0.21461059159245083, + "learning_rate": 7.77814054642361e-05, + "loss": 1.1434, + "step": 321 + }, + { + "epoch": 0.59, + "grad_norm": 0.25348824286656163, + "learning_rate": 7.775332152792539e-05, + "loss": 1.2368, + "step": 322 + }, + { + "epoch": 0.59, + "grad_norm": 0.22275034726331247, + "learning_rate": 7.772506609332995e-05, + "loss": 1.1827, + "step": 323 + }, + { + "epoch": 0.59, + "grad_norm": 0.25030821228147526, + "learning_rate": 7.769663928880298e-05, + "loss": 1.2428, + "step": 324 + }, + { + "epoch": 0.59, + "grad_norm": 0.22251804398745534, + "learning_rate": 7.766804124347608e-05, + "loss": 1.1889, + "step": 325 + }, + { + "epoch": 0.6, + "grad_norm": 0.23381455520411995, + "learning_rate": 7.763927208725879e-05, + "loss": 1.2115, + "step": 326 + }, + { + "epoch": 0.6, + "grad_norm": 0.27341902651946226, + "learning_rate": 7.761033195083791e-05, + "loss": 1.2535, + "step": 327 + }, + { + "epoch": 0.6, + "grad_norm": 0.24862471659814522, + "learning_rate": 7.758122096567694e-05, + "loss": 1.2128, + "step": 328 + }, + { + "epoch": 0.6, + "grad_norm": 0.2251357082045494, + "learning_rate": 7.755193926401547e-05, + "loss": 1.2334, + "step": 329 + }, + { + "epoch": 0.6, + "grad_norm": 0.3173274941622932, + "learning_rate": 7.752248697886857e-05, + "loss": 1.226, + "step": 330 + }, + { + "epoch": 0.61, + "grad_norm": 0.23056440717672175, + "learning_rate": 7.74928642440263e-05, + "loss": 1.2339, + "step": 331 + }, + { + "epoch": 0.61, + "grad_norm": 0.2801507500859342, + "learning_rate": 7.746307119405286e-05, + "loss": 1.287, + "step": 332 + }, + { + "epoch": 0.61, + "grad_norm": 0.2267818430426272, + "learning_rate": 7.743310796428622e-05, + "loss": 1.1916, + "step": 333 + }, + { + "epoch": 0.61, + "grad_norm": 0.2777329160365585, + "learning_rate": 7.74029746908374e-05, + "loss": 1.252, + "step": 334 + }, + { + "epoch": 0.61, + "grad_norm": 0.25289169762353, + "learning_rate": 7.737267151058983e-05, + "loss": 1.2153, + "step": 335 + }, + { + "epoch": 0.61, + "grad_norm": 0.2424670686901653, + "learning_rate": 7.734219856119875e-05, + "loss": 1.2227, + "step": 336 + }, + { + "epoch": 0.62, + "grad_norm": 0.22747092217441645, + "learning_rate": 7.731155598109067e-05, + "loss": 1.19, + "step": 337 + }, + { + "epoch": 0.62, + "grad_norm": 0.2307810940100189, + "learning_rate": 7.728074390946257e-05, + "loss": 1.1818, + "step": 338 + }, + { + "epoch": 0.62, + "grad_norm": 0.2583402574655623, + "learning_rate": 7.724976248628142e-05, + "loss": 1.1608, + "step": 339 + }, + { + "epoch": 0.62, + "grad_norm": 0.22140209760890694, + "learning_rate": 7.721861185228347e-05, + "loss": 1.1245, + "step": 340 + }, + { + "epoch": 0.62, + "grad_norm": 0.25859310758244686, + "learning_rate": 7.718729214897362e-05, + "loss": 1.2247, + "step": 341 + }, + { + "epoch": 0.63, + "grad_norm": 0.26371179531372124, + "learning_rate": 7.715580351862482e-05, + "loss": 1.2128, + "step": 342 + }, + { + "epoch": 0.63, + "grad_norm": 0.26575541302851047, + "learning_rate": 7.712414610427733e-05, + "loss": 1.2443, + "step": 343 + }, + { + "epoch": 0.63, + "grad_norm": 0.269978305197599, + "learning_rate": 7.709232004973816e-05, + "loss": 1.2231, + "step": 344 + }, + { + "epoch": 0.63, + "grad_norm": 0.26583998705977047, + "learning_rate": 7.70603254995804e-05, + "loss": 1.2476, + "step": 345 + }, + { + "epoch": 0.63, + "grad_norm": 0.24256062164066097, + "learning_rate": 7.702816259914253e-05, + "loss": 1.2901, + "step": 346 + }, + { + "epoch": 0.63, + "grad_norm": 0.3463123472658915, + "learning_rate": 7.699583149452779e-05, + "loss": 1.3277, + "step": 347 + }, + { + "epoch": 0.64, + "grad_norm": 0.2269096590531878, + "learning_rate": 7.696333233260345e-05, + "loss": 1.2047, + "step": 348 + }, + { + "epoch": 0.64, + "grad_norm": 0.25136883001050025, + "learning_rate": 7.693066526100031e-05, + "loss": 1.1619, + "step": 349 + }, + { + "epoch": 0.64, + "grad_norm": 0.2565112571116145, + "learning_rate": 7.68978304281118e-05, + "loss": 1.2389, + "step": 350 + }, + { + "epoch": 0.64, + "grad_norm": 0.22175779550828703, + "learning_rate": 7.686482798309349e-05, + "loss": 1.2238, + "step": 351 + }, + { + "epoch": 0.64, + "grad_norm": 0.22588304332216555, + "learning_rate": 7.683165807586234e-05, + "loss": 1.174, + "step": 352 + }, + { + "epoch": 0.65, + "grad_norm": 0.24889474296529737, + "learning_rate": 7.6798320857096e-05, + "loss": 1.2366, + "step": 353 + }, + { + "epoch": 0.65, + "grad_norm": 0.27339703806525034, + "learning_rate": 7.676481647823214e-05, + "loss": 1.2356, + "step": 354 + }, + { + "epoch": 0.65, + "grad_norm": 0.23424666722888365, + "learning_rate": 7.673114509146782e-05, + "loss": 1.2089, + "step": 355 + }, + { + "epoch": 0.65, + "grad_norm": 0.27978285392461766, + "learning_rate": 7.66973068497587e-05, + "loss": 1.2609, + "step": 356 + }, + { + "epoch": 0.65, + "grad_norm": 0.2509423350138824, + "learning_rate": 7.666330190681844e-05, + "loss": 1.1777, + "step": 357 + }, + { + "epoch": 0.65, + "grad_norm": 0.23007730927468031, + "learning_rate": 7.662913041711793e-05, + "loss": 1.154, + "step": 358 + }, + { + "epoch": 0.66, + "grad_norm": 0.2438648674953112, + "learning_rate": 7.659479253588462e-05, + "loss": 1.2257, + "step": 359 + }, + { + "epoch": 0.66, + "grad_norm": 0.28816093242092233, + "learning_rate": 7.65602884191018e-05, + "loss": 1.2558, + "step": 360 + }, + { + "epoch": 0.66, + "grad_norm": 0.24972815300596035, + "learning_rate": 7.652561822350793e-05, + "loss": 1.2837, + "step": 361 + }, + { + "epoch": 0.66, + "grad_norm": 0.2543189139697063, + "learning_rate": 7.649078210659587e-05, + "loss": 1.2193, + "step": 362 + }, + { + "epoch": 0.66, + "grad_norm": 0.2237937956718952, + "learning_rate": 7.645578022661224e-05, + "loss": 1.2237, + "step": 363 + }, + { + "epoch": 0.67, + "grad_norm": 0.29742029408787396, + "learning_rate": 7.642061274255657e-05, + "loss": 1.2116, + "step": 364 + }, + { + "epoch": 0.67, + "grad_norm": 0.2462883147335493, + "learning_rate": 7.638527981418075e-05, + "loss": 1.1827, + "step": 365 + }, + { + "epoch": 0.67, + "grad_norm": 0.2647802498907096, + "learning_rate": 7.634978160198817e-05, + "loss": 1.2739, + "step": 366 + }, + { + "epoch": 0.67, + "grad_norm": 0.22360398779217264, + "learning_rate": 7.631411826723306e-05, + "loss": 1.2185, + "step": 367 + }, + { + "epoch": 0.67, + "grad_norm": 0.2635048004593543, + "learning_rate": 7.627828997191973e-05, + "loss": 1.2317, + "step": 368 + }, + { + "epoch": 0.67, + "grad_norm": 0.2764803449917684, + "learning_rate": 7.624229687880184e-05, + "loss": 1.1923, + "step": 369 + }, + { + "epoch": 0.68, + "grad_norm": 0.25724943233414527, + "learning_rate": 7.620613915138166e-05, + "loss": 1.2218, + "step": 370 + }, + { + "epoch": 0.68, + "grad_norm": 0.2858318045794755, + "learning_rate": 7.61698169539093e-05, + "loss": 1.1496, + "step": 371 + }, + { + "epoch": 0.68, + "grad_norm": 0.23547216647460364, + "learning_rate": 7.613333045138206e-05, + "loss": 1.1905, + "step": 372 + }, + { + "epoch": 0.68, + "grad_norm": 0.22984814903684375, + "learning_rate": 7.609667980954355e-05, + "loss": 1.2009, + "step": 373 + }, + { + "epoch": 0.68, + "grad_norm": 0.2551903754079084, + "learning_rate": 7.605986519488301e-05, + "loss": 1.2042, + "step": 374 + }, + { + "epoch": 0.69, + "grad_norm": 0.2508257410125616, + "learning_rate": 7.602288677463457e-05, + "loss": 1.2468, + "step": 375 + }, + { + "epoch": 0.69, + "grad_norm": 0.25324577774935964, + "learning_rate": 7.598574471677644e-05, + "loss": 1.2603, + "step": 376 + }, + { + "epoch": 0.69, + "grad_norm": 0.35888776531769967, + "learning_rate": 7.59484391900302e-05, + "loss": 1.1929, + "step": 377 + }, + { + "epoch": 0.69, + "grad_norm": 0.22048517191014724, + "learning_rate": 7.591097036385994e-05, + "loss": 1.1783, + "step": 378 + }, + { + "epoch": 0.69, + "grad_norm": 0.2781160412746083, + "learning_rate": 7.587333840847162e-05, + "loss": 1.3397, + "step": 379 + }, + { + "epoch": 0.7, + "grad_norm": 0.24033046830332258, + "learning_rate": 7.583554349481222e-05, + "loss": 1.2436, + "step": 380 + }, + { + "epoch": 0.7, + "grad_norm": 0.26413762380260003, + "learning_rate": 7.579758579456893e-05, + "loss": 1.1917, + "step": 381 + }, + { + "epoch": 0.7, + "grad_norm": 0.2390937887338632, + "learning_rate": 7.575946548016847e-05, + "loss": 1.2186, + "step": 382 + }, + { + "epoch": 0.7, + "grad_norm": 0.25131263043429275, + "learning_rate": 7.572118272477622e-05, + "loss": 1.2538, + "step": 383 + }, + { + "epoch": 0.7, + "grad_norm": 0.223974104870702, + "learning_rate": 7.568273770229546e-05, + "loss": 1.2165, + "step": 384 + }, + { + "epoch": 0.7, + "grad_norm": 0.25840356830252875, + "learning_rate": 7.564413058736663e-05, + "loss": 1.1848, + "step": 385 + }, + { + "epoch": 0.71, + "grad_norm": 0.2723156683076603, + "learning_rate": 7.560536155536641e-05, + "loss": 1.1982, + "step": 386 + }, + { + "epoch": 0.71, + "grad_norm": 0.265687427976889, + "learning_rate": 7.556643078240708e-05, + "loss": 1.231, + "step": 387 + }, + { + "epoch": 0.71, + "grad_norm": 0.25152762080976077, + "learning_rate": 7.552733844533562e-05, + "loss": 1.1974, + "step": 388 + }, + { + "epoch": 0.71, + "grad_norm": 0.2366049485053541, + "learning_rate": 7.548808472173292e-05, + "loss": 1.3119, + "step": 389 + }, + { + "epoch": 0.71, + "grad_norm": 0.22092196577077122, + "learning_rate": 7.5448669789913e-05, + "loss": 1.195, + "step": 390 + }, + { + "epoch": 0.72, + "grad_norm": 0.22667521540462374, + "learning_rate": 7.540909382892217e-05, + "loss": 1.1431, + "step": 391 + }, + { + "epoch": 0.72, + "grad_norm": 0.25432207282646513, + "learning_rate": 7.536935701853823e-05, + "loss": 1.2173, + "step": 392 + }, + { + "epoch": 0.72, + "grad_norm": 0.29950506457923864, + "learning_rate": 7.53294595392697e-05, + "loss": 1.1962, + "step": 393 + }, + { + "epoch": 0.72, + "grad_norm": 0.24735689607229913, + "learning_rate": 7.528940157235487e-05, + "loss": 1.2053, + "step": 394 + }, + { + "epoch": 0.72, + "grad_norm": 0.24394198607459663, + "learning_rate": 7.524918329976114e-05, + "loss": 1.1979, + "step": 395 + }, + { + "epoch": 0.72, + "grad_norm": 0.2630369372689188, + "learning_rate": 7.520880490418409e-05, + "loss": 1.2111, + "step": 396 + }, + { + "epoch": 0.73, + "grad_norm": 0.26275028416291457, + "learning_rate": 7.516826656904664e-05, + "loss": 1.2133, + "step": 397 + }, + { + "epoch": 0.73, + "grad_norm": 0.23938074620956928, + "learning_rate": 7.512756847849831e-05, + "loss": 1.1355, + "step": 398 + }, + { + "epoch": 0.73, + "grad_norm": 0.3724960610098138, + "learning_rate": 7.508671081741428e-05, + "loss": 1.2572, + "step": 399 + }, + { + "epoch": 0.73, + "grad_norm": 0.24161685847894723, + "learning_rate": 7.504569377139462e-05, + "loss": 1.1706, + "step": 400 + }, + { + "epoch": 0.73, + "grad_norm": 0.26121591322670523, + "learning_rate": 7.50045175267634e-05, + "loss": 1.2135, + "step": 401 + }, + { + "epoch": 0.74, + "grad_norm": 0.2465579498164775, + "learning_rate": 7.496318227056788e-05, + "loss": 1.1641, + "step": 402 + }, + { + "epoch": 0.74, + "grad_norm": 0.2556288696122787, + "learning_rate": 7.492168819057767e-05, + "loss": 1.2939, + "step": 403 + }, + { + "epoch": 0.74, + "grad_norm": 0.261481216336303, + "learning_rate": 7.488003547528382e-05, + "loss": 1.2026, + "step": 404 + }, + { + "epoch": 0.74, + "grad_norm": 0.2389415135676362, + "learning_rate": 7.483822431389799e-05, + "loss": 1.2131, + "step": 405 + }, + { + "epoch": 0.74, + "grad_norm": 0.2559201956627192, + "learning_rate": 7.479625489635162e-05, + "loss": 1.1246, + "step": 406 + }, + { + "epoch": 0.74, + "grad_norm": 0.27127932491822604, + "learning_rate": 7.475412741329504e-05, + "loss": 1.2429, + "step": 407 + }, + { + "epoch": 0.75, + "grad_norm": 0.27006004008695594, + "learning_rate": 7.47118420560966e-05, + "loss": 1.2388, + "step": 408 + }, + { + "epoch": 0.75, + "grad_norm": 0.23716823297200537, + "learning_rate": 7.466939901684182e-05, + "loss": 1.1264, + "step": 409 + }, + { + "epoch": 0.75, + "grad_norm": 0.2885373898669248, + "learning_rate": 7.462679848833252e-05, + "loss": 1.2786, + "step": 410 + }, + { + "epoch": 0.75, + "grad_norm": 0.49215227598639927, + "learning_rate": 7.458404066408588e-05, + "loss": 1.2386, + "step": 411 + }, + { + "epoch": 0.75, + "grad_norm": 0.24235735604947403, + "learning_rate": 7.454112573833368e-05, + "loss": 1.1423, + "step": 412 + }, + { + "epoch": 0.76, + "grad_norm": 0.2584614748054343, + "learning_rate": 7.449805390602127e-05, + "loss": 1.2669, + "step": 413 + }, + { + "epoch": 0.76, + "grad_norm": 0.23806123085998873, + "learning_rate": 7.445482536280684e-05, + "loss": 1.1763, + "step": 414 + }, + { + "epoch": 0.76, + "grad_norm": 0.24459517607786851, + "learning_rate": 7.441144030506043e-05, + "loss": 1.198, + "step": 415 + }, + { + "epoch": 0.76, + "grad_norm": 0.25801616402700395, + "learning_rate": 7.436789892986304e-05, + "loss": 1.2136, + "step": 416 + }, + { + "epoch": 0.76, + "grad_norm": 0.2814819942392514, + "learning_rate": 7.432420143500578e-05, + "loss": 1.2398, + "step": 417 + }, + { + "epoch": 0.76, + "grad_norm": 0.22134709322606153, + "learning_rate": 7.428034801898893e-05, + "loss": 1.1592, + "step": 418 + }, + { + "epoch": 0.77, + "grad_norm": 0.2899677536995633, + "learning_rate": 7.42363388810211e-05, + "loss": 1.2296, + "step": 419 + }, + { + "epoch": 0.77, + "grad_norm": 0.24005943230262294, + "learning_rate": 7.419217422101822e-05, + "loss": 1.2223, + "step": 420 + }, + { + "epoch": 0.77, + "grad_norm": 0.26417562369496167, + "learning_rate": 7.414785423960275e-05, + "loss": 1.2261, + "step": 421 + }, + { + "epoch": 0.77, + "grad_norm": 0.2580815883535521, + "learning_rate": 7.410337913810271e-05, + "loss": 1.2021, + "step": 422 + }, + { + "epoch": 0.77, + "grad_norm": 0.25242217589496435, + "learning_rate": 7.405874911855071e-05, + "loss": 1.239, + "step": 423 + }, + { + "epoch": 0.78, + "grad_norm": 0.21991733999839932, + "learning_rate": 7.401396438368315e-05, + "loss": 1.1716, + "step": 424 + }, + { + "epoch": 0.78, + "grad_norm": 0.40116538322720213, + "learning_rate": 7.396902513693924e-05, + "loss": 1.2773, + "step": 425 + }, + { + "epoch": 0.78, + "grad_norm": 0.277333939455099, + "learning_rate": 7.392393158246002e-05, + "loss": 1.2574, + "step": 426 + }, + { + "epoch": 0.78, + "grad_norm": 0.27146087746385755, + "learning_rate": 7.387868392508756e-05, + "loss": 1.2243, + "step": 427 + }, + { + "epoch": 0.78, + "grad_norm": 0.255881055620786, + "learning_rate": 7.38332823703639e-05, + "loss": 1.223, + "step": 428 + }, + { + "epoch": 0.78, + "grad_norm": 0.24807364856677255, + "learning_rate": 7.378772712453021e-05, + "loss": 1.1985, + "step": 429 + }, + { + "epoch": 0.79, + "grad_norm": 0.25746257617764423, + "learning_rate": 7.37420183945258e-05, + "loss": 1.2502, + "step": 430 + }, + { + "epoch": 0.79, + "grad_norm": 0.28851991049982234, + "learning_rate": 7.369615638798722e-05, + "loss": 1.2535, + "step": 431 + }, + { + "epoch": 0.79, + "grad_norm": 0.24113389811604363, + "learning_rate": 7.365014131324725e-05, + "loss": 1.2227, + "step": 432 + }, + { + "epoch": 0.79, + "grad_norm": 0.2414465151257969, + "learning_rate": 7.360397337933405e-05, + "loss": 1.1884, + "step": 433 + }, + { + "epoch": 0.79, + "grad_norm": 0.2735463134699831, + "learning_rate": 7.355765279597011e-05, + "loss": 1.2756, + "step": 434 + }, + { + "epoch": 0.8, + "grad_norm": 0.2588437452987293, + "learning_rate": 7.351117977357139e-05, + "loss": 1.2108, + "step": 435 + }, + { + "epoch": 0.8, + "grad_norm": 0.26573294117796553, + "learning_rate": 7.346455452324629e-05, + "loss": 1.1821, + "step": 436 + }, + { + "epoch": 0.8, + "grad_norm": 0.2555476577827304, + "learning_rate": 7.341777725679473e-05, + "loss": 1.1937, + "step": 437 + }, + { + "epoch": 0.8, + "grad_norm": 0.2867704132108098, + "learning_rate": 7.337084818670716e-05, + "loss": 1.2272, + "step": 438 + }, + { + "epoch": 0.8, + "grad_norm": 0.27726678115981157, + "learning_rate": 7.332376752616367e-05, + "loss": 1.2331, + "step": 439 + }, + { + "epoch": 0.8, + "grad_norm": 0.26955338021079955, + "learning_rate": 7.32765354890329e-05, + "loss": 1.1731, + "step": 440 + }, + { + "epoch": 0.81, + "grad_norm": 0.25250321202536524, + "learning_rate": 7.322915228987116e-05, + "loss": 1.2653, + "step": 441 + }, + { + "epoch": 0.81, + "grad_norm": 0.24748844179765395, + "learning_rate": 7.318161814392143e-05, + "loss": 1.24, + "step": 442 + }, + { + "epoch": 0.81, + "grad_norm": 0.28177805247356325, + "learning_rate": 7.313393326711239e-05, + "loss": 1.185, + "step": 443 + }, + { + "epoch": 0.81, + "grad_norm": 0.24093242000396312, + "learning_rate": 7.30860978760574e-05, + "loss": 1.1994, + "step": 444 + }, + { + "epoch": 0.81, + "grad_norm": 0.26277803901457075, + "learning_rate": 7.30381121880536e-05, + "loss": 1.212, + "step": 445 + }, + { + "epoch": 0.82, + "grad_norm": 0.2506524258682433, + "learning_rate": 7.298997642108079e-05, + "loss": 1.2421, + "step": 446 + }, + { + "epoch": 0.82, + "grad_norm": 0.2840599700015824, + "learning_rate": 7.294169079380061e-05, + "loss": 1.1818, + "step": 447 + }, + { + "epoch": 0.82, + "grad_norm": 0.24892184038117549, + "learning_rate": 7.289325552555538e-05, + "loss": 1.1916, + "step": 448 + }, + { + "epoch": 0.82, + "grad_norm": 0.2700898428541357, + "learning_rate": 7.284467083636722e-05, + "loss": 1.2517, + "step": 449 + }, + { + "epoch": 0.82, + "grad_norm": 0.2617848546539419, + "learning_rate": 7.279593694693698e-05, + "loss": 1.2063, + "step": 450 + }, + { + "epoch": 0.82, + "grad_norm": 0.2698278585334131, + "learning_rate": 7.274705407864332e-05, + "loss": 1.194, + "step": 451 + }, + { + "epoch": 0.83, + "grad_norm": 0.23678313024953834, + "learning_rate": 7.26980224535416e-05, + "loss": 1.2349, + "step": 452 + }, + { + "epoch": 0.83, + "grad_norm": 0.24851875792002978, + "learning_rate": 7.264884229436293e-05, + "loss": 1.1758, + "step": 453 + }, + { + "epoch": 0.83, + "grad_norm": 0.24122080121681125, + "learning_rate": 7.259951382451318e-05, + "loss": 1.1962, + "step": 454 + }, + { + "epoch": 0.83, + "grad_norm": 0.22741322959884405, + "learning_rate": 7.25500372680719e-05, + "loss": 1.1702, + "step": 455 + }, + { + "epoch": 0.83, + "grad_norm": 0.2297475610861458, + "learning_rate": 7.250041284979137e-05, + "loss": 1.1466, + "step": 456 + }, + { + "epoch": 0.84, + "grad_norm": 0.3057605989721467, + "learning_rate": 7.245064079509553e-05, + "loss": 1.246, + "step": 457 + }, + { + "epoch": 0.84, + "grad_norm": 0.2719638501597136, + "learning_rate": 7.240072133007899e-05, + "loss": 1.2184, + "step": 458 + }, + { + "epoch": 0.84, + "grad_norm": 0.2436807816414479, + "learning_rate": 7.235065468150593e-05, + "loss": 1.2324, + "step": 459 + }, + { + "epoch": 0.84, + "grad_norm": 0.23436349430255515, + "learning_rate": 7.23004410768092e-05, + "loss": 1.1813, + "step": 460 + }, + { + "epoch": 0.84, + "grad_norm": 0.2398940990211377, + "learning_rate": 7.22500807440892e-05, + "loss": 1.1924, + "step": 461 + }, + { + "epoch": 0.84, + "grad_norm": 0.2605716625062531, + "learning_rate": 7.219957391211281e-05, + "loss": 1.182, + "step": 462 + }, + { + "epoch": 0.85, + "grad_norm": 0.260462524570941, + "learning_rate": 7.214892081031244e-05, + "loss": 1.2136, + "step": 463 + }, + { + "epoch": 0.85, + "grad_norm": 0.21979766512306334, + "learning_rate": 7.209812166878491e-05, + "loss": 1.2066, + "step": 464 + }, + { + "epoch": 0.85, + "grad_norm": 0.23324453647530663, + "learning_rate": 7.204717671829051e-05, + "loss": 1.1657, + "step": 465 + }, + { + "epoch": 0.85, + "grad_norm": 0.2529434935507481, + "learning_rate": 7.199608619025177e-05, + "loss": 1.2093, + "step": 466 + }, + { + "epoch": 0.85, + "grad_norm": 0.25371701891720116, + "learning_rate": 7.194485031675265e-05, + "loss": 1.2225, + "step": 467 + }, + { + "epoch": 0.86, + "grad_norm": 0.23272423066292103, + "learning_rate": 7.189346933053725e-05, + "loss": 1.1721, + "step": 468 + }, + { + "epoch": 0.86, + "grad_norm": 0.25122928735587546, + "learning_rate": 7.184194346500892e-05, + "loss": 1.2537, + "step": 469 + }, + { + "epoch": 0.86, + "grad_norm": 0.2159270875490409, + "learning_rate": 7.179027295422913e-05, + "loss": 1.197, + "step": 470 + }, + { + "epoch": 0.86, + "grad_norm": 0.2633111059076544, + "learning_rate": 7.173845803291636e-05, + "loss": 1.1721, + "step": 471 + }, + { + "epoch": 0.86, + "grad_norm": 0.30555936322098703, + "learning_rate": 7.168649893644517e-05, + "loss": 1.3011, + "step": 472 + }, + { + "epoch": 0.87, + "grad_norm": 0.23492670111453726, + "learning_rate": 7.163439590084502e-05, + "loss": 1.1601, + "step": 473 + }, + { + "epoch": 0.87, + "grad_norm": 0.26602734263721806, + "learning_rate": 7.158214916279923e-05, + "loss": 1.2808, + "step": 474 + }, + { + "epoch": 0.87, + "grad_norm": 0.3182695007856262, + "learning_rate": 7.152975895964386e-05, + "loss": 1.2967, + "step": 475 + }, + { + "epoch": 0.87, + "grad_norm": 0.2785021674736721, + "learning_rate": 7.147722552936673e-05, + "loss": 1.1789, + "step": 476 + }, + { + "epoch": 0.87, + "grad_norm": 0.279474303138652, + "learning_rate": 7.142454911060627e-05, + "loss": 1.2596, + "step": 477 + }, + { + "epoch": 0.87, + "grad_norm": 0.2556980144910755, + "learning_rate": 7.137172994265044e-05, + "loss": 1.2426, + "step": 478 + }, + { + "epoch": 0.88, + "grad_norm": 0.3311256331993533, + "learning_rate": 7.131876826543565e-05, + "loss": 1.2059, + "step": 479 + }, + { + "epoch": 0.88, + "grad_norm": 0.26467296197775253, + "learning_rate": 7.12656643195457e-05, + "loss": 1.2482, + "step": 480 + }, + { + "epoch": 0.88, + "grad_norm": 0.27444885274652553, + "learning_rate": 7.121241834621064e-05, + "loss": 1.2528, + "step": 481 + }, + { + "epoch": 0.88, + "grad_norm": 0.2572283861115396, + "learning_rate": 7.115903058730567e-05, + "loss": 1.1849, + "step": 482 + }, + { + "epoch": 0.88, + "grad_norm": 0.2677065778235683, + "learning_rate": 7.11055012853501e-05, + "loss": 1.2011, + "step": 483 + }, + { + "epoch": 0.89, + "grad_norm": 0.29470622036742816, + "learning_rate": 7.105183068350619e-05, + "loss": 1.2398, + "step": 484 + }, + { + "epoch": 0.89, + "grad_norm": 0.27609230248969197, + "learning_rate": 7.099801902557811e-05, + "loss": 1.2259, + "step": 485 + }, + { + "epoch": 0.89, + "grad_norm": 0.24248634168099284, + "learning_rate": 7.094406655601073e-05, + "loss": 1.2282, + "step": 486 + }, + { + "epoch": 0.89, + "grad_norm": 0.2765941767688746, + "learning_rate": 7.088997351988865e-05, + "loss": 1.2319, + "step": 487 + }, + { + "epoch": 0.89, + "grad_norm": 0.29347776909858947, + "learning_rate": 7.083574016293493e-05, + "loss": 1.1765, + "step": 488 + }, + { + "epoch": 0.89, + "grad_norm": 0.285370295424537, + "learning_rate": 7.078136673151008e-05, + "loss": 1.26, + "step": 489 + }, + { + "epoch": 0.9, + "grad_norm": 0.29408734903836536, + "learning_rate": 7.072685347261093e-05, + "loss": 1.226, + "step": 490 + }, + { + "epoch": 0.9, + "grad_norm": 0.27437470239205813, + "learning_rate": 7.067220063386947e-05, + "loss": 1.1976, + "step": 491 + }, + { + "epoch": 0.9, + "grad_norm": 0.2680770258777871, + "learning_rate": 7.061740846355176e-05, + "loss": 1.1915, + "step": 492 + }, + { + "epoch": 0.9, + "grad_norm": 0.27200362879502954, + "learning_rate": 7.056247721055678e-05, + "loss": 1.2002, + "step": 493 + }, + { + "epoch": 0.9, + "grad_norm": 0.2637811092577037, + "learning_rate": 7.050740712441528e-05, + "loss": 1.287, + "step": 494 + }, + { + "epoch": 0.91, + "grad_norm": 0.24657959209271266, + "learning_rate": 7.045219845528875e-05, + "loss": 1.2284, + "step": 495 + }, + { + "epoch": 0.91, + "grad_norm": 0.25311992110358666, + "learning_rate": 7.039685145396812e-05, + "loss": 1.1616, + "step": 496 + }, + { + "epoch": 0.91, + "grad_norm": 0.2564633694193358, + "learning_rate": 7.034136637187275e-05, + "loss": 1.2067, + "step": 497 + }, + { + "epoch": 0.91, + "grad_norm": 0.2446797651174144, + "learning_rate": 7.028574346104926e-05, + "loss": 1.2284, + "step": 498 + }, + { + "epoch": 0.91, + "grad_norm": 0.2592751463399255, + "learning_rate": 7.022998297417034e-05, + "loss": 1.2371, + "step": 499 + }, + { + "epoch": 0.91, + "grad_norm": 0.2500713943206808, + "learning_rate": 7.017408516453365e-05, + "loss": 1.1061, + "step": 500 + } + ], + "logging_steps": 1.0, + "max_steps": 1638, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 50, + "total_flos": 518357992341504.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-500/training_args.bin b/checkpoint-500/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..c5d2416a3b70bb5260978ec9996f00154a724ba7 --- /dev/null +++ b/checkpoint-500/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b22e8f9d51a16d03a2c506fa3d1eafa8f4b1ae992992c2086a4d435ffd97387e +size 6712 diff --git a/checkpoint-500/zero_to_fp32.py b/checkpoint-500/zero_to_fp32.py new file mode 100755 index 0000000000000000000000000000000000000000..24cc342e78d1a006c782b3a4cd68d9ce786d8fd8 --- /dev/null +++ b/checkpoint-500/zero_to_fp32.py @@ -0,0 +1,604 @@ +#!/usr/bin/env python + +# Copyright (c) Microsoft Corporation. +# SPDX-License-Identifier: Apache-2.0 + +# DeepSpeed Team + +# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets +# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in +# the future. Once extracted, the weights don't require DeepSpeed and can be used in any +# application. +# +# example: python zero_to_fp32.py . pytorch_model.bin + +import argparse +import torch +import glob +import math +import os +import re +from collections import OrderedDict +from dataclasses import dataclass + +# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with +# DeepSpeed data structures it has to be available in the current python environment. +from deepspeed.utils import logger +from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS, + FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES, + FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS) + + +@dataclass +class zero_model_state: + buffers: dict() + param_shapes: dict() + shared_params: list + ds_version: int + frozen_param_shapes: dict() + frozen_param_fragments: dict() + + +debug = 0 + +# load to cpu +device = torch.device('cpu') + + +def atoi(text): + return int(text) if text.isdigit() else text + + +def natural_keys(text): + ''' + alist.sort(key=natural_keys) sorts in human order + http://nedbatchelder.com/blog/200712/human_sorting.html + (See Toothy's implementation in the comments) + ''' + return [atoi(c) for c in re.split(r'(\d+)', text)] + + +def get_model_state_file(checkpoint_dir, zero_stage): + if not os.path.isdir(checkpoint_dir): + raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist") + + # there should be only one file + if zero_stage <= 2: + file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt") + elif zero_stage == 3: + file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt") + + if not os.path.exists(file): + raise FileNotFoundError(f"can't find model states file at '{file}'") + + return file + + +def get_checkpoint_files(checkpoint_dir, glob_pattern): + # XXX: need to test that this simple glob rule works for multi-node setup too + ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys) + + if len(ckpt_files) == 0: + raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'") + + return ckpt_files + + +def get_optim_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt") + + +def get_model_state_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_model_states.pt") + + +def parse_model_states(files): + zero_model_states = [] + for file in files: + state_dict = torch.load(file, map_location=device) + + if BUFFER_NAMES not in state_dict: + raise ValueError(f"{file} is not a model state checkpoint") + buffer_names = state_dict[BUFFER_NAMES] + if debug: + print("Found buffers:", buffer_names) + + # recover just the buffers while restoring them to fp32 if they were saved in fp16 + buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names} + param_shapes = state_dict[PARAM_SHAPES] + + # collect parameters that are included in param_shapes + param_names = [] + for s in param_shapes: + for name in s.keys(): + param_names.append(name) + + # update with frozen parameters + frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None) + if frozen_param_shapes is not None: + if debug: + print(f"Found frozen_param_shapes: {frozen_param_shapes}") + param_names += list(frozen_param_shapes.keys()) + + # handle shared params + shared_params = [[k, v] for k, v in state_dict["shared_params"].items()] + + ds_version = state_dict.get(DS_VERSION, None) + + frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None) + + z_model_state = zero_model_state(buffers=buffers, + param_shapes=param_shapes, + shared_params=shared_params, + ds_version=ds_version, + frozen_param_shapes=frozen_param_shapes, + frozen_param_fragments=frozen_param_fragments) + zero_model_states.append(z_model_state) + + return zero_model_states + + +def parse_optim_states(files, ds_checkpoint_dir): + + total_files = len(files) + state_dicts = [] + for f in files: + state_dict = torch.load(f, map_location=device) + # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights + # and also handle the case where it was already removed by another helper script + state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None) + state_dicts.append(state_dict) + + if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]: + raise ValueError(f"{files[0]} is not a zero checkpoint") + zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE] + world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT] + + # For ZeRO-2 each param group can have different partition_count as data parallelism for expert + # parameters can be different from data parallelism for non-expert parameters. So we can just + # use the max of the partition_count to get the dp world_size. + + if type(world_size) is list: + world_size = max(world_size) + + if world_size != total_files: + raise ValueError( + f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. " + "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes." + ) + + # the groups are named differently in each stage + if zero_stage <= 2: + fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS + elif zero_stage == 3: + fp32_groups_key = FP32_FLAT_GROUPS + else: + raise ValueError(f"unknown zero stage {zero_stage}") + + if zero_stage <= 2: + fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))] + elif zero_stage == 3: + # if there is more than one param group, there will be multiple flattened tensors - one + # flattened tensor per group - for simplicity merge them into a single tensor + # + # XXX: could make the script more memory efficient for when there are multiple groups - it + # will require matching the sub-lists of param_shapes for each param group flattened tensor + + fp32_flat_groups = [ + torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts)) + ] + + return zero_stage, world_size, fp32_flat_groups + + +def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters): + """ + Returns fp32 state_dict reconstructed from ds checkpoint + + Args: + - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are) + + """ + print(f"Processing zero checkpoint '{ds_checkpoint_dir}'") + + optim_files = get_optim_files(ds_checkpoint_dir) + zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir) + print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}") + + model_files = get_model_state_files(ds_checkpoint_dir) + + zero_model_states = parse_model_states(model_files) + print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}') + + if zero_stage <= 2: + return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + elif zero_stage == 3: + return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + + +def _zero2_merge_frozen_params(state_dict, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + frozen_param_fragments = zero_model_states[0].frozen_param_fragments + + if debug: + num_elem = sum(s.numel() for s in frozen_param_shapes.values()) + print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in frozen_param_fragments.values()]) + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + state_dict[name] = frozen_param_fragments[name] + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _has_callable(obj, fn): + attr = getattr(obj, fn, None) + return callable(attr) + + +def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + + # Reconstruction protocol: + # + # XXX: document this + + if debug: + for i in range(world_size): + for j in range(len(fp32_flat_groups[0])): + print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}") + + # XXX: memory usage doubles here (zero2) + num_param_groups = len(fp32_flat_groups[0]) + merged_single_partition_of_fp32_groups = [] + for i in range(num_param_groups): + merged_partitions = [sd[i] for sd in fp32_flat_groups] + full_single_fp32_vector = torch.cat(merged_partitions, 0) + merged_single_partition_of_fp32_groups.append(full_single_fp32_vector) + avail_numel = sum( + [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups]) + + if debug: + wanted_params = sum([len(shapes) for shapes in param_shapes]) + wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes]) + # not asserting if there is a mismatch due to possible padding + print(f"Have {avail_numel} numels to process.") + print(f"Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + total_numel = 0 + total_params = 0 + for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups): + offset = 0 + avail_numel = full_single_fp32_vector.numel() + for name, shape in shapes.items(): + + unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape) + total_numel += unpartitioned_numel + total_params += 1 + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape) + offset += unpartitioned_numel + + # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and + # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex + # paddings performed in the code it's almost impossible to predict the exact numbers w/o the + # live optimizer object, so we are checking that the numbers are within the right range + align_to = 2 * world_size + + def zero2_align(x): + return align_to * math.ceil(x / align_to) + + if debug: + print(f"original offset={offset}, avail_numel={avail_numel}") + + offset = zero2_align(offset) + avail_numel = zero2_align(avail_numel) + + if debug: + print(f"aligned offset={offset}, avail_numel={avail_numel}") + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero2_merge_frozen_params(state_dict, zero_model_states) + + _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def zero3_partitioned_param_info(unpartitioned_numel, world_size): + remainder = unpartitioned_numel % world_size + padding_numel = (world_size - remainder) if remainder else 0 + partitioned_numel = math.ceil(unpartitioned_numel / world_size) + return partitioned_numel, padding_numel + + +def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + if debug: + for i in range(world_size): + num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values()) + print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in zero_model_states[0].frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states) + state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape) + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + avail_numel = fp32_flat_groups[0].numel() * world_size + # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each + # param, re-consolidating each param, while dealing with padding if any + + # merge list of dicts, preserving order + param_shapes = {k: v for d in param_shapes for k, v in d.items()} + + if debug: + for i in range(world_size): + print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}") + + wanted_params = len(param_shapes) + wanted_numel = sum(shape.numel() for shape in param_shapes.values()) + # not asserting if there is a mismatch due to possible padding + avail_numel = fp32_flat_groups[0].numel() * world_size + print(f"Trainable params: Have {avail_numel} numels to process.") + print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + offset = 0 + total_numel = 0 + total_params = 0 + for name, shape in param_shapes.items(): + + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + total_params += 1 + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + # XXX: memory usage doubles here + state_dict[name] = torch.cat( + tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)), + 0).narrow(0, 0, unpartitioned_numel).view(shape) + offset += partitioned_numel + + offset *= world_size + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero3_merge_frozen_params(state_dict, world_size, zero_model_states) + + _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with + ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example + via a model hub. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + + Returns: + - pytorch ``state_dict`` + + Note: this approach may not work if your application doesn't have sufficient free CPU memory and + you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with + the checkpoint. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + # do the training and checkpoint saving + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu + model = model.cpu() # move to cpu + model.load_state_dict(state_dict) + # submit to model hub or save the model to share with others + + In this example the ``model`` will no longer be usable in the deepspeed context of the same + application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead. + + """ + if tag is None: + latest_path = os.path.join(checkpoint_dir, 'latest') + if os.path.isfile(latest_path): + with open(latest_path, 'r') as fd: + tag = fd.read().strip() + else: + raise ValueError(f"Unable to find 'latest' file at {latest_path}") + + ds_checkpoint_dir = os.path.join(checkpoint_dir, tag) + + if not os.path.isdir(ds_checkpoint_dir): + raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist") + + return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters) + + +def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be + loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + """ + + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters) + print(f"Saving fp32 state dict to {output_file}") + torch.save(state_dict, output_file) + + +def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None): + """ + 1. Put the provided model to cpu + 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` + 3. Load it into the provided model + + Args: + - ``model``: the model object to update + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + + Returns: + - ``model`: modified model + + Make sure you have plenty of CPU memory available before you call this function. If you don't + have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it + conveniently placed for you in the checkpoint folder. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint + model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir) + # submit to model hub or save the model to share with others + + Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context + of the same application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + """ + logger.info(f"Extracting fp32 weights") + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag) + + logger.info(f"Overwriting model with fp32 weights") + model = model.cpu() + model.load_state_dict(state_dict, strict=False) + + return model + + +if __name__ == "__main__": + + parser = argparse.ArgumentParser() + parser.add_argument("checkpoint_dir", + type=str, + help="path to the desired checkpoint folder, e.g., path/checkpoint-12") + parser.add_argument( + "output_file", + type=str, + help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)") + parser.add_argument("-t", + "--tag", + type=str, + default=None, + help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1") + parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters") + parser.add_argument("-d", "--debug", action='store_true', help="enable debug") + args = parser.parse_args() + + debug = args.debug + + convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, + args.output_file, + tag=args.tag, + exclude_frozen_parameters=args.exclude_frozen_parameters) diff --git a/checkpoint-550/README.md b/checkpoint-550/README.md new file mode 100644 index 0000000000000000000000000000000000000000..16b1eacdd9353dec380a08ee77ce6ed5ab50f12e --- /dev/null +++ b/checkpoint-550/README.md @@ -0,0 +1,202 @@ +--- +library_name: peft +base_model: gotzmann/uni +--- + +# Model Card for Model ID + + + + + +## Model Details + +### Model Description + + + + + +- **Developed by:** [More Information Needed] +- **Funded by [optional]:** [More Information Needed] +- **Shared by [optional]:** [More Information Needed] +- **Model type:** [More Information Needed] +- **Language(s) (NLP):** [More Information Needed] +- **License:** [More Information Needed] +- **Finetuned from model [optional]:** [More Information Needed] + +### Model Sources [optional] + + + +- **Repository:** [More Information Needed] +- **Paper [optional]:** [More Information Needed] +- **Demo [optional]:** [More Information Needed] + +## Uses + + + +### Direct Use + + + +[More Information Needed] + +### Downstream Use [optional] + + + +[More Information Needed] + +### Out-of-Scope Use + + + +[More Information Needed] + +## Bias, Risks, and Limitations + + + +[More Information Needed] + +### Recommendations + + + +Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations. + +## How to Get Started with the Model + +Use the code below to get started with the model. + +[More Information Needed] + +## Training Details + +### Training Data + + + +[More Information Needed] + +### Training Procedure + + + +#### Preprocessing [optional] + +[More Information Needed] + + +#### Training Hyperparameters + +- **Training regime:** [More Information Needed] + +#### Speeds, Sizes, Times [optional] + + + +[More Information Needed] + +## Evaluation + + + +### Testing Data, Factors & Metrics + +#### Testing Data + + + +[More Information Needed] + +#### Factors + + + +[More Information Needed] + +#### Metrics + + + +[More Information Needed] + +### Results + +[More Information Needed] + +#### Summary + + + +## Model Examination [optional] + + + +[More Information Needed] + +## Environmental Impact + + + +Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700). + +- **Hardware Type:** [More Information Needed] +- **Hours used:** [More Information Needed] +- **Cloud Provider:** [More Information Needed] +- **Compute Region:** [More Information Needed] +- **Carbon Emitted:** [More Information Needed] + +## Technical Specifications [optional] + +### Model Architecture and Objective + +[More Information Needed] + +### Compute Infrastructure + +[More Information Needed] + +#### Hardware + +[More Information Needed] + +#### Software + +[More Information Needed] + +## Citation [optional] + + + +**BibTeX:** + +[More Information Needed] + +**APA:** + +[More Information Needed] + +## Glossary [optional] + + + +[More Information Needed] + +## More Information [optional] + +[More Information Needed] + +## Model Card Authors [optional] + +[More Information Needed] + +## Model Card Contact + +[More Information Needed] +### Framework versions + +- PEFT 0.10.0 \ No newline at end of file diff --git a/checkpoint-550/adapter_config.json b/checkpoint-550/adapter_config.json new file mode 100644 index 0000000000000000000000000000000000000000..3cd6dba5d79f7ca21fd4ad465cbbcac1e0960476 --- /dev/null +++ b/checkpoint-550/adapter_config.json @@ -0,0 +1,31 @@ +{ + "alpha_pattern": {}, + "auto_mapping": null, + "base_model_name_or_path": "gotzmann/uni", + "bias": "none", + "fan_in_fan_out": false, + "inference_mode": true, + "init_lora_weights": true, + "layer_replication": null, + "layers_pattern": null, + "layers_to_transform": null, + "loftq_config": {}, + "lora_alpha": 128, + "lora_dropout": 0.0, + "megatron_config": null, + "megatron_core": "megatron.core", + "modules_to_save": null, + "peft_type": "LORA", + "r": 128, + "rank_pattern": {}, + "revision": null, + "target_modules": [ + "v_proj", + "k_proj", + "q_proj", + "o_proj" + ], + "task_type": "CAUSAL_LM", + "use_dora": false, + "use_rslora": true +} \ No newline at end of file diff --git a/checkpoint-550/adapter_model.safetensors b/checkpoint-550/adapter_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..32e1954bb1c649a08bf8ea73cff1f174ddf0dec4 --- /dev/null +++ b/checkpoint-550/adapter_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8c75b39d45cca7d206d977667bc6f8c9c1466310d7e45d2b24582dd05c09d3db +size 1048664848 diff --git a/checkpoint-550/global_step550/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt b/checkpoint-550/global_step550/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..f39a5d0af0a7ea2a0629a459487b48598c4e557d --- /dev/null +++ b/checkpoint-550/global_step550/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:c9bdb8e9714818d43e310ae5399805ac80d7ae6269611e90f9dd23963ed30838 +size 787270042 diff --git a/checkpoint-550/global_step550/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt b/checkpoint-550/global_step550/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..a45ae9086bd58f91c4bd3de98e1620f82546bd96 --- /dev/null +++ b/checkpoint-550/global_step550/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:e0443646b7690f9304417958fb5a414b64baa76379152e0bd516ede2965a97e9 +size 787270042 diff --git a/checkpoint-550/global_step550/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt b/checkpoint-550/global_step550/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..20dcedc5e8c8745891d355d6f378e2ebe0bee10b --- /dev/null +++ b/checkpoint-550/global_step550/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:afc4a95e25b4f44f2ff1bd820133c35ecf8aa5ec9050f73dfb4fad574492bedf +size 787270042 diff --git a/checkpoint-550/global_step550/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt b/checkpoint-550/global_step550/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..b3b2e351f6d2c7e59d7afc9750152b8132f0f618 --- /dev/null +++ b/checkpoint-550/global_step550/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:992e4ad48b550c9b90f88b6618771e8506ca25504d1fcd4e6eed468584bded16 +size 787270042 diff --git a/checkpoint-550/global_step550/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt b/checkpoint-550/global_step550/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..22b535e2f0e78cfbed91c70f3186b86314b0b443 --- /dev/null +++ b/checkpoint-550/global_step550/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6e9ef5d326419cdae9d907d5de6e3f99eb044ac2cfbb99807e487457d17fd8ca +size 787270042 diff --git a/checkpoint-550/global_step550/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt b/checkpoint-550/global_step550/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..1f3ea5f34070ed1b88839aacea0ac9d236af9ff6 --- /dev/null +++ b/checkpoint-550/global_step550/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4b99e005d2dba98532c74b684aceb1c3a39f3e07367a7e13c1200dec9743c797 +size 787270042 diff --git a/checkpoint-550/global_step550/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt b/checkpoint-550/global_step550/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..ea803c68de7912923a9a4de8756d3c25af7dcd9b --- /dev/null +++ b/checkpoint-550/global_step550/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:c269f1869bf87711b60c46947dccc4cd8681ea1928ef78648755f7f6e4038bec +size 787270042 diff --git a/checkpoint-550/global_step550/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt b/checkpoint-550/global_step550/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..94ba0996ed37a37b32c5677b71d859f850668300 --- /dev/null +++ b/checkpoint-550/global_step550/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:e4d47a45eba702ef60e7bc54a1cd392064cdc931b040e08d10c82aaefabf8729 +size 787270042 diff --git a/checkpoint-550/global_step550/zero_pp_rank_0_mp_rank_00_model_states.pt b/checkpoint-550/global_step550/zero_pp_rank_0_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..95dea22e460e244a4bd3bdcd488d690e16af38f8 --- /dev/null +++ b/checkpoint-550/global_step550/zero_pp_rank_0_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:447754814077d8171e639808eaec7f09a69b51aaa33d141f25fc81cac7eb9680 +size 653742 diff --git a/checkpoint-550/global_step550/zero_pp_rank_1_mp_rank_00_model_states.pt b/checkpoint-550/global_step550/zero_pp_rank_1_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..73bfdcc01b8ed843ca4c6087e88e721e12b7c260 --- /dev/null +++ b/checkpoint-550/global_step550/zero_pp_rank_1_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:96179433ffa678357050191f70db8b1315e460028f80e7047d21a0e3353e0244 +size 653742 diff --git a/checkpoint-550/global_step550/zero_pp_rank_2_mp_rank_00_model_states.pt b/checkpoint-550/global_step550/zero_pp_rank_2_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..efb5294b9ff666cc57531d42fba0b54f2fe93715 --- /dev/null +++ b/checkpoint-550/global_step550/zero_pp_rank_2_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1f1fd925d4de9ef7d7d59ca38c304f25d64f0234d3f4daba18404e56c8d0f2de +size 653742 diff --git a/checkpoint-550/global_step550/zero_pp_rank_3_mp_rank_00_model_states.pt b/checkpoint-550/global_step550/zero_pp_rank_3_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..599cf68642f398ddbfba68b01d8778b09e49f71b --- /dev/null +++ b/checkpoint-550/global_step550/zero_pp_rank_3_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:da3108e83cd4db592cb1e8a8cd99164b398f2ba3aee6e7e26083d5ebc5b43ebe +size 653742 diff --git a/checkpoint-550/global_step550/zero_pp_rank_4_mp_rank_00_model_states.pt b/checkpoint-550/global_step550/zero_pp_rank_4_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..5863f4f0d1ffc776e22c872b2d435f5fe5e3b94e --- /dev/null +++ b/checkpoint-550/global_step550/zero_pp_rank_4_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b884d28b156e88efe0190688ca8b0c7d7a3a1eca22d067534d3bd88f8732d592 +size 653742 diff --git a/checkpoint-550/global_step550/zero_pp_rank_5_mp_rank_00_model_states.pt b/checkpoint-550/global_step550/zero_pp_rank_5_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..e0b71021f810289c008896de21ea62e594b08b56 --- /dev/null +++ b/checkpoint-550/global_step550/zero_pp_rank_5_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:924dc44d269e8843dc537a87bca260b55d6843a6ef03a6a40e699d6b251fabf3 +size 653742 diff --git a/checkpoint-550/global_step550/zero_pp_rank_6_mp_rank_00_model_states.pt b/checkpoint-550/global_step550/zero_pp_rank_6_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..89d7e71fcabfc2848909f4b712f402481d021154 --- /dev/null +++ b/checkpoint-550/global_step550/zero_pp_rank_6_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2f0a2ba01f539e197fda6de338350c7aafecb693b871f33faa40c627a772b6ce +size 653742 diff --git a/checkpoint-550/global_step550/zero_pp_rank_7_mp_rank_00_model_states.pt b/checkpoint-550/global_step550/zero_pp_rank_7_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..a7ee13bd1ecdb8b4a1696dae6d8fdb272170e7a7 --- /dev/null +++ b/checkpoint-550/global_step550/zero_pp_rank_7_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a18f4449b3c2850948d30da549ad064e9d34b2f009393c309722cb30f55e4d97 +size 653742 diff --git a/checkpoint-550/latest b/checkpoint-550/latest new file mode 100644 index 0000000000000000000000000000000000000000..1606c8674d0d1cc86edce34c7f47c11b57f13e09 --- /dev/null +++ b/checkpoint-550/latest @@ -0,0 +1 @@ +global_step550 \ No newline at end of file diff --git a/checkpoint-550/rng_state_0.pth b/checkpoint-550/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..4e5b7e2ec90fdb824c8932464c1d9068330655a7 --- /dev/null +++ b/checkpoint-550/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:36d2a2034ebb05cb71c510897f2795b31164e50f17b270bc25d2be3ad9a17b22 +size 15984 diff --git a/checkpoint-550/rng_state_1.pth b/checkpoint-550/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..7d8d7722fc72cab6d492b76cb99c8177dcc47544 --- /dev/null +++ b/checkpoint-550/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:060dfdb1c49102cbdc8868a6031e68787601b4ccd782f3fb9b137e20c1fd2c7a +size 15984 diff --git a/checkpoint-550/rng_state_2.pth b/checkpoint-550/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..3c9f84eff30cfa9ea1feedaf262d61fb12e4cba7 --- /dev/null +++ b/checkpoint-550/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:af01895cb66e616591f2e4baa8dcd8151530eab133c73571ccb31c74f35422ce +size 15984 diff --git a/checkpoint-550/rng_state_3.pth b/checkpoint-550/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..6eebfb928f8e91eff0ea1645a20b5aa4465c705b --- /dev/null +++ b/checkpoint-550/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:677921992b1e0cef3aee776f245975003d22f51d9bd6ed20f248ded1deb72fa9 +size 15984 diff --git a/checkpoint-550/rng_state_4.pth b/checkpoint-550/rng_state_4.pth new file mode 100644 index 0000000000000000000000000000000000000000..0866030a266c6d003cc378a9418a723f69e8ab99 --- /dev/null +++ b/checkpoint-550/rng_state_4.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d69353c629541c690c5471f8ec05fdab2bfecf3d37afaa436bc45939da6db68f +size 15984 diff --git a/checkpoint-550/rng_state_5.pth b/checkpoint-550/rng_state_5.pth new file mode 100644 index 0000000000000000000000000000000000000000..554638d77107f832d7aa51c61645ee2d6c48a36d --- /dev/null +++ b/checkpoint-550/rng_state_5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8e40ba6668cc03c9162c68a933d164bf38ae2d196a9a6fec03ae615491201185 +size 15984 diff --git a/checkpoint-550/rng_state_6.pth b/checkpoint-550/rng_state_6.pth new file mode 100644 index 0000000000000000000000000000000000000000..964331b65172a1bcac03e4673415fa787f724268 --- /dev/null +++ b/checkpoint-550/rng_state_6.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:870968fea834e24b2e099cf3e4fe1e3fb8caf38d8f8e5b790d7d47386d4d05f5 +size 15984 diff --git a/checkpoint-550/rng_state_7.pth b/checkpoint-550/rng_state_7.pth new file mode 100644 index 0000000000000000000000000000000000000000..cd4754d65217d0f9d1f2d3334397df7a8a079652 --- /dev/null +++ b/checkpoint-550/rng_state_7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e9e19618bee7c6ef43256fea25abe19bca88535eb1e7dc213cde8929ae4e8180 +size 15984 diff --git a/checkpoint-550/scheduler.pt b/checkpoint-550/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..ee995c910a44d9ae50344b80840b95e1f818da04 --- /dev/null +++ b/checkpoint-550/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e0e7272088a458885697bbc37f90655326bd1ad16b6e603840724e6ba896f59f +size 1064 diff --git a/checkpoint-550/special_tokens_map.json b/checkpoint-550/special_tokens_map.json new file mode 100644 index 0000000000000000000000000000000000000000..72ecfeeb7e14d244c936169d2ed139eeae235ef1 --- /dev/null +++ b/checkpoint-550/special_tokens_map.json @@ -0,0 +1,24 @@ +{ + "bos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "eos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "pad_token": "", + "unk_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + } +} diff --git a/checkpoint-550/tokenizer.model b/checkpoint-550/tokenizer.model new file mode 100644 index 0000000000000000000000000000000000000000..6c00c742ce03c627d6cd5b795984876fa49fa899 --- /dev/null +++ b/checkpoint-550/tokenizer.model @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347 +size 499723 diff --git a/checkpoint-550/tokenizer_config.json b/checkpoint-550/tokenizer_config.json new file mode 100644 index 0000000000000000000000000000000000000000..bb5a9f09d8c0f3c32c66fc6118fe5c76c5c6fd90 --- /dev/null +++ b/checkpoint-550/tokenizer_config.json @@ -0,0 +1,45 @@ +{ + "add_bos_token": true, + "add_eos_token": false, + "add_prefix_space": false, + "added_tokens_decoder": { + "0": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "1": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "2": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + } + }, + "bos_token": "", + "chat_template": "{% if messages[0]['role'] == 'system' %}{% set system_message = messages[0]['content'] %}{% endif %}{% if system_message is defined %}{{ '' + '### System:\\n\\n' + system_message }}{% endif %}{% for message in messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ '\\n\\n### Human:\\n\\n' + content }}{% elif message['role'] == 'assistant' %}{{ '\\n\\n### Assistant:\\n\\n' + content + '' }}{% endif %}{% endfor %}", + "clean_up_tokenization_spaces": false, + "eos_token": "", + "legacy": false, + "model_max_length": 1000000000000000019884624838656, + "pad_token": "", + "padding_side": "right", + "sp_model_kwargs": {}, + "spaces_between_special_tokens": false, + "split_special_tokens": false, + "tokenizer_class": "LlamaTokenizer", + "unk_token": "", + "use_default_system_prompt": false +} diff --git a/checkpoint-550/trainer_state.json b/checkpoint-550/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..3d8108b8aa12bbcb297f8976c18b1a479ebe992c --- /dev/null +++ b/checkpoint-550/trainer_state.json @@ -0,0 +1,3871 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 1.005944215820759, + "eval_steps": 500, + "global_step": 550, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "grad_norm": 0.849355824164473, + "learning_rate": 4.878048780487805e-07, + "loss": 1.3655, + "step": 1 + }, + { + "epoch": 0.0, + "grad_norm": 10.01567518957158, + "learning_rate": 9.75609756097561e-07, + "loss": 1.5767, + "step": 2 + }, + { + "epoch": 0.01, + "grad_norm": 0.6466000875559635, + "learning_rate": 1.4634146341463414e-06, + "loss": 1.3913, + "step": 3 + }, + { + "epoch": 0.01, + "grad_norm": 0.6644565932010504, + "learning_rate": 1.951219512195122e-06, + "loss": 1.3218, + "step": 4 + }, + { + "epoch": 0.01, + "grad_norm": 0.571354207588475, + "learning_rate": 2.4390243902439027e-06, + "loss": 1.3597, + "step": 5 + }, + { + "epoch": 0.01, + "grad_norm": 0.31036262839244955, + "learning_rate": 2.926829268292683e-06, + "loss": 1.2832, + "step": 6 + }, + { + "epoch": 0.01, + "grad_norm": 0.2622135027188184, + "learning_rate": 3.414634146341464e-06, + "loss": 1.2161, + "step": 7 + }, + { + "epoch": 0.01, + "grad_norm": 0.296824630261661, + "learning_rate": 3.902439024390244e-06, + "loss": 1.2985, + "step": 8 + }, + { + "epoch": 0.02, + "grad_norm": 0.2557267467361569, + "learning_rate": 4.390243902439025e-06, + "loss": 1.3175, + "step": 9 + }, + { + "epoch": 0.02, + "grad_norm": 0.23418939513890769, + "learning_rate": 4.8780487804878055e-06, + "loss": 1.2617, + "step": 10 + }, + { + "epoch": 0.02, + "grad_norm": 0.2364760983285843, + "learning_rate": 5.365853658536586e-06, + "loss": 1.3103, + "step": 11 + }, + { + "epoch": 0.02, + "grad_norm": 0.23893034721889, + "learning_rate": 5.853658536585366e-06, + "loss": 1.2405, + "step": 12 + }, + { + "epoch": 0.02, + "grad_norm": 0.25563593295485887, + "learning_rate": 6.341463414634147e-06, + "loss": 1.2831, + "step": 13 + }, + { + "epoch": 0.03, + "grad_norm": 0.23239975352661665, + "learning_rate": 6.829268292682928e-06, + "loss": 1.3125, + "step": 14 + }, + { + "epoch": 0.03, + "grad_norm": 0.3092813858209507, + "learning_rate": 7.317073170731707e-06, + "loss": 1.2422, + "step": 15 + }, + { + "epoch": 0.03, + "grad_norm": 0.282563380367434, + "learning_rate": 7.804878048780489e-06, + "loss": 1.2453, + "step": 16 + }, + { + "epoch": 0.03, + "grad_norm": 0.22065680088315018, + "learning_rate": 8.292682926829268e-06, + "loss": 1.2491, + "step": 17 + }, + { + "epoch": 0.03, + "grad_norm": 0.22777800877980184, + "learning_rate": 8.78048780487805e-06, + "loss": 1.2655, + "step": 18 + }, + { + "epoch": 0.03, + "grad_norm": 0.22145212540177928, + "learning_rate": 9.268292682926831e-06, + "loss": 1.2413, + "step": 19 + }, + { + "epoch": 0.04, + "grad_norm": 0.22482351883112714, + "learning_rate": 9.756097560975611e-06, + "loss": 1.2653, + "step": 20 + }, + { + "epoch": 0.04, + "grad_norm": 0.20823080508385733, + "learning_rate": 1.024390243902439e-05, + "loss": 1.2374, + "step": 21 + }, + { + "epoch": 0.04, + "grad_norm": 0.26025492562935737, + "learning_rate": 1.0731707317073172e-05, + "loss": 1.2065, + "step": 22 + }, + { + "epoch": 0.04, + "grad_norm": 0.2150252124176173, + "learning_rate": 1.1219512195121953e-05, + "loss": 1.2782, + "step": 23 + }, + { + "epoch": 0.04, + "grad_norm": 0.2505915177425618, + "learning_rate": 1.1707317073170731e-05, + "loss": 1.2742, + "step": 24 + }, + { + "epoch": 0.05, + "grad_norm": 0.20129223044786942, + "learning_rate": 1.2195121951219513e-05, + "loss": 1.3366, + "step": 25 + }, + { + "epoch": 0.05, + "grad_norm": 0.1973508510397107, + "learning_rate": 1.2682926829268294e-05, + "loss": 1.2476, + "step": 26 + }, + { + "epoch": 0.05, + "grad_norm": 0.27103325392437194, + "learning_rate": 1.3170731707317076e-05, + "loss": 1.2325, + "step": 27 + }, + { + "epoch": 0.05, + "grad_norm": 0.17954976411006285, + "learning_rate": 1.3658536585365855e-05, + "loss": 1.2523, + "step": 28 + }, + { + "epoch": 0.05, + "grad_norm": 0.22216997851088888, + "learning_rate": 1.4146341463414635e-05, + "loss": 1.3297, + "step": 29 + }, + { + "epoch": 0.05, + "grad_norm": 0.2071458864548587, + "learning_rate": 1.4634146341463415e-05, + "loss": 1.2127, + "step": 30 + }, + { + "epoch": 0.06, + "grad_norm": 0.18039422081622164, + "learning_rate": 1.5121951219512196e-05, + "loss": 1.2509, + "step": 31 + }, + { + "epoch": 0.06, + "grad_norm": 0.18631254372974412, + "learning_rate": 1.5609756097560978e-05, + "loss": 1.2247, + "step": 32 + }, + { + "epoch": 0.06, + "grad_norm": 0.18843872523649827, + "learning_rate": 1.6097560975609757e-05, + "loss": 1.195, + "step": 33 + }, + { + "epoch": 0.06, + "grad_norm": 0.2163847267778325, + "learning_rate": 1.6585365853658537e-05, + "loss": 1.2179, + "step": 34 + }, + { + "epoch": 0.06, + "grad_norm": 0.19687688475496104, + "learning_rate": 1.7073170731707317e-05, + "loss": 1.2763, + "step": 35 + }, + { + "epoch": 0.07, + "grad_norm": 0.20409643064887947, + "learning_rate": 1.75609756097561e-05, + "loss": 1.253, + "step": 36 + }, + { + "epoch": 0.07, + "grad_norm": 0.1879182661759335, + "learning_rate": 1.804878048780488e-05, + "loss": 1.2586, + "step": 37 + }, + { + "epoch": 0.07, + "grad_norm": 0.19400648948514373, + "learning_rate": 1.8536585365853663e-05, + "loss": 1.2154, + "step": 38 + }, + { + "epoch": 0.07, + "grad_norm": 0.1878879343148452, + "learning_rate": 1.902439024390244e-05, + "loss": 1.2304, + "step": 39 + }, + { + "epoch": 0.07, + "grad_norm": 0.17687475469924052, + "learning_rate": 1.9512195121951222e-05, + "loss": 1.2351, + "step": 40 + }, + { + "epoch": 0.07, + "grad_norm": 0.18223935625384885, + "learning_rate": 2e-05, + "loss": 1.2222, + "step": 41 + }, + { + "epoch": 0.08, + "grad_norm": 0.1943061629408338, + "learning_rate": 2.048780487804878e-05, + "loss": 1.2044, + "step": 42 + }, + { + "epoch": 0.08, + "grad_norm": 0.17027514338700078, + "learning_rate": 2.0975609756097564e-05, + "loss": 1.1548, + "step": 43 + }, + { + "epoch": 0.08, + "grad_norm": 0.18553769630586192, + "learning_rate": 2.1463414634146344e-05, + "loss": 1.2721, + "step": 44 + }, + { + "epoch": 0.08, + "grad_norm": 0.19732826914228765, + "learning_rate": 2.1951219512195124e-05, + "loss": 1.3097, + "step": 45 + }, + { + "epoch": 0.08, + "grad_norm": 0.18714230986631472, + "learning_rate": 2.2439024390243907e-05, + "loss": 1.2662, + "step": 46 + }, + { + "epoch": 0.09, + "grad_norm": 0.19988987568002223, + "learning_rate": 2.2926829268292683e-05, + "loss": 1.2904, + "step": 47 + }, + { + "epoch": 0.09, + "grad_norm": 0.17744650133390918, + "learning_rate": 2.3414634146341463e-05, + "loss": 1.1825, + "step": 48 + }, + { + "epoch": 0.09, + "grad_norm": 0.16576734763834533, + "learning_rate": 2.3902439024390246e-05, + "loss": 1.1858, + "step": 49 + }, + { + "epoch": 0.09, + "grad_norm": 0.179591794065527, + "learning_rate": 2.4390243902439026e-05, + "loss": 1.2711, + "step": 50 + }, + { + "epoch": 0.09, + "grad_norm": 0.17923464471176911, + "learning_rate": 2.4878048780487805e-05, + "loss": 1.2289, + "step": 51 + }, + { + "epoch": 0.1, + "grad_norm": 0.18991742907836837, + "learning_rate": 2.536585365853659e-05, + "loss": 1.3097, + "step": 52 + }, + { + "epoch": 0.1, + "grad_norm": 0.19849796137254636, + "learning_rate": 2.5853658536585368e-05, + "loss": 1.2489, + "step": 53 + }, + { + "epoch": 0.1, + "grad_norm": 0.17452371110976383, + "learning_rate": 2.634146341463415e-05, + "loss": 1.2461, + "step": 54 + }, + { + "epoch": 0.1, + "grad_norm": 0.17671022353085036, + "learning_rate": 2.682926829268293e-05, + "loss": 1.153, + "step": 55 + }, + { + "epoch": 0.1, + "grad_norm": 0.36820559192096686, + "learning_rate": 2.731707317073171e-05, + "loss": 1.2431, + "step": 56 + }, + { + "epoch": 0.1, + "grad_norm": 0.20331468526494198, + "learning_rate": 2.7804878048780487e-05, + "loss": 1.2575, + "step": 57 + }, + { + "epoch": 0.11, + "grad_norm": 0.2402486598118377, + "learning_rate": 2.829268292682927e-05, + "loss": 1.2538, + "step": 58 + }, + { + "epoch": 0.11, + "grad_norm": 0.2549409484173144, + "learning_rate": 2.878048780487805e-05, + "loss": 1.2065, + "step": 59 + }, + { + "epoch": 0.11, + "grad_norm": 0.2053105349872685, + "learning_rate": 2.926829268292683e-05, + "loss": 1.2094, + "step": 60 + }, + { + "epoch": 0.11, + "grad_norm": 0.17971910872957886, + "learning_rate": 2.9756097560975613e-05, + "loss": 1.228, + "step": 61 + }, + { + "epoch": 0.11, + "grad_norm": 0.1885853654992973, + "learning_rate": 3.0243902439024392e-05, + "loss": 1.2286, + "step": 62 + }, + { + "epoch": 0.12, + "grad_norm": 0.1848524571968613, + "learning_rate": 3.073170731707317e-05, + "loss": 1.2718, + "step": 63 + }, + { + "epoch": 0.12, + "grad_norm": 0.18734105883548513, + "learning_rate": 3.1219512195121955e-05, + "loss": 1.2357, + "step": 64 + }, + { + "epoch": 0.12, + "grad_norm": 0.17774668052121825, + "learning_rate": 3.170731707317074e-05, + "loss": 1.1509, + "step": 65 + }, + { + "epoch": 0.12, + "grad_norm": 0.17890968008080646, + "learning_rate": 3.2195121951219514e-05, + "loss": 1.1924, + "step": 66 + }, + { + "epoch": 0.12, + "grad_norm": 0.18249273371332375, + "learning_rate": 3.268292682926829e-05, + "loss": 1.2545, + "step": 67 + }, + { + "epoch": 0.12, + "grad_norm": 0.21064122671902577, + "learning_rate": 3.3170731707317074e-05, + "loss": 1.2832, + "step": 68 + }, + { + "epoch": 0.13, + "grad_norm": 0.1820064171955093, + "learning_rate": 3.365853658536586e-05, + "loss": 1.2071, + "step": 69 + }, + { + "epoch": 0.13, + "grad_norm": 0.16996662800553433, + "learning_rate": 3.414634146341463e-05, + "loss": 1.2073, + "step": 70 + }, + { + "epoch": 0.13, + "grad_norm": 0.1618669302922445, + "learning_rate": 3.4634146341463416e-05, + "loss": 1.1289, + "step": 71 + }, + { + "epoch": 0.13, + "grad_norm": 0.18948744950985544, + "learning_rate": 3.51219512195122e-05, + "loss": 1.2915, + "step": 72 + }, + { + "epoch": 0.13, + "grad_norm": 0.18326143691603383, + "learning_rate": 3.5609756097560976e-05, + "loss": 1.2238, + "step": 73 + }, + { + "epoch": 0.14, + "grad_norm": 0.17410704510700503, + "learning_rate": 3.609756097560976e-05, + "loss": 1.1784, + "step": 74 + }, + { + "epoch": 0.14, + "grad_norm": 0.1983667344995625, + "learning_rate": 3.658536585365854e-05, + "loss": 1.2452, + "step": 75 + }, + { + "epoch": 0.14, + "grad_norm": 0.3416310763369357, + "learning_rate": 3.7073170731707325e-05, + "loss": 1.1972, + "step": 76 + }, + { + "epoch": 0.14, + "grad_norm": 0.2776466983511955, + "learning_rate": 3.75609756097561e-05, + "loss": 1.3121, + "step": 77 + }, + { + "epoch": 0.14, + "grad_norm": 0.20026129636576834, + "learning_rate": 3.804878048780488e-05, + "loss": 1.2436, + "step": 78 + }, + { + "epoch": 0.14, + "grad_norm": 0.21064549243917835, + "learning_rate": 3.853658536585366e-05, + "loss": 1.2064, + "step": 79 + }, + { + "epoch": 0.15, + "grad_norm": 0.22119482175714267, + "learning_rate": 3.9024390243902444e-05, + "loss": 1.2715, + "step": 80 + }, + { + "epoch": 0.15, + "grad_norm": 0.23047133748844142, + "learning_rate": 3.951219512195122e-05, + "loss": 1.2888, + "step": 81 + }, + { + "epoch": 0.15, + "grad_norm": 0.18741863156973176, + "learning_rate": 4e-05, + "loss": 1.248, + "step": 82 + }, + { + "epoch": 0.15, + "grad_norm": 0.1747859810629604, + "learning_rate": 4.0487804878048786e-05, + "loss": 1.1683, + "step": 83 + }, + { + "epoch": 0.15, + "grad_norm": 0.1896944798413341, + "learning_rate": 4.097560975609756e-05, + "loss": 1.2155, + "step": 84 + }, + { + "epoch": 0.16, + "grad_norm": 0.18724128114363303, + "learning_rate": 4.1463414634146346e-05, + "loss": 1.2273, + "step": 85 + }, + { + "epoch": 0.16, + "grad_norm": 0.17368125504855478, + "learning_rate": 4.195121951219513e-05, + "loss": 1.224, + "step": 86 + }, + { + "epoch": 0.16, + "grad_norm": 0.18371141013625703, + "learning_rate": 4.2439024390243905e-05, + "loss": 1.2294, + "step": 87 + }, + { + "epoch": 0.16, + "grad_norm": 0.1791029365673714, + "learning_rate": 4.292682926829269e-05, + "loss": 1.2895, + "step": 88 + }, + { + "epoch": 0.16, + "grad_norm": 0.20259974283859655, + "learning_rate": 4.341463414634147e-05, + "loss": 1.1841, + "step": 89 + }, + { + "epoch": 0.16, + "grad_norm": 0.17457456183272174, + "learning_rate": 4.390243902439025e-05, + "loss": 1.2357, + "step": 90 + }, + { + "epoch": 0.17, + "grad_norm": 0.1815824380789748, + "learning_rate": 4.439024390243903e-05, + "loss": 1.2304, + "step": 91 + }, + { + "epoch": 0.17, + "grad_norm": 0.17566480599583392, + "learning_rate": 4.4878048780487814e-05, + "loss": 1.242, + "step": 92 + }, + { + "epoch": 0.17, + "grad_norm": 0.18422975005984474, + "learning_rate": 4.536585365853658e-05, + "loss": 1.2177, + "step": 93 + }, + { + "epoch": 0.17, + "grad_norm": 0.16796781877940678, + "learning_rate": 4.5853658536585366e-05, + "loss": 1.1482, + "step": 94 + }, + { + "epoch": 0.17, + "grad_norm": 0.18636131653783305, + "learning_rate": 4.634146341463415e-05, + "loss": 1.1758, + "step": 95 + }, + { + "epoch": 0.18, + "grad_norm": 0.1823665700289814, + "learning_rate": 4.6829268292682926e-05, + "loss": 1.289, + "step": 96 + }, + { + "epoch": 0.18, + "grad_norm": 0.1719900691262439, + "learning_rate": 4.731707317073171e-05, + "loss": 1.1626, + "step": 97 + }, + { + "epoch": 0.18, + "grad_norm": 0.17937994168039778, + "learning_rate": 4.780487804878049e-05, + "loss": 1.175, + "step": 98 + }, + { + "epoch": 0.18, + "grad_norm": 0.16631851422106986, + "learning_rate": 4.829268292682927e-05, + "loss": 1.2177, + "step": 99 + }, + { + "epoch": 0.18, + "grad_norm": 0.19143696232800309, + "learning_rate": 4.878048780487805e-05, + "loss": 1.3071, + "step": 100 + }, + { + "epoch": 0.18, + "grad_norm": 0.17859506638780318, + "learning_rate": 4.9268292682926835e-05, + "loss": 1.2351, + "step": 101 + }, + { + "epoch": 0.19, + "grad_norm": 0.18381520321248196, + "learning_rate": 4.975609756097561e-05, + "loss": 1.2342, + "step": 102 + }, + { + "epoch": 0.19, + "grad_norm": 0.17968218683773912, + "learning_rate": 5.0243902439024394e-05, + "loss": 1.2074, + "step": 103 + }, + { + "epoch": 0.19, + "grad_norm": 0.18139489969339018, + "learning_rate": 5.073170731707318e-05, + "loss": 1.1558, + "step": 104 + }, + { + "epoch": 0.19, + "grad_norm": 0.17366624842514394, + "learning_rate": 5.121951219512195e-05, + "loss": 1.1897, + "step": 105 + }, + { + "epoch": 0.19, + "grad_norm": 0.16034845455223745, + "learning_rate": 5.1707317073170736e-05, + "loss": 1.179, + "step": 106 + }, + { + "epoch": 0.2, + "grad_norm": 0.17583069577827776, + "learning_rate": 5.219512195121952e-05, + "loss": 1.1856, + "step": 107 + }, + { + "epoch": 0.2, + "grad_norm": 0.1853758076989552, + "learning_rate": 5.26829268292683e-05, + "loss": 1.2072, + "step": 108 + }, + { + "epoch": 0.2, + "grad_norm": 0.19597443965936462, + "learning_rate": 5.317073170731708e-05, + "loss": 1.2271, + "step": 109 + }, + { + "epoch": 0.2, + "grad_norm": 0.1899206334098331, + "learning_rate": 5.365853658536586e-05, + "loss": 1.1961, + "step": 110 + }, + { + "epoch": 0.2, + "grad_norm": 0.17463763837757018, + "learning_rate": 5.4146341463414645e-05, + "loss": 1.2049, + "step": 111 + }, + { + "epoch": 0.2, + "grad_norm": 0.20431371701229986, + "learning_rate": 5.463414634146342e-05, + "loss": 1.2891, + "step": 112 + }, + { + "epoch": 0.21, + "grad_norm": 0.1814475107638498, + "learning_rate": 5.51219512195122e-05, + "loss": 1.2346, + "step": 113 + }, + { + "epoch": 0.21, + "grad_norm": 0.1883849423207823, + "learning_rate": 5.5609756097560974e-05, + "loss": 1.244, + "step": 114 + }, + { + "epoch": 0.21, + "grad_norm": 0.1857258128640568, + "learning_rate": 5.609756097560976e-05, + "loss": 1.2669, + "step": 115 + }, + { + "epoch": 0.21, + "grad_norm": 0.1740768514118401, + "learning_rate": 5.658536585365854e-05, + "loss": 1.2414, + "step": 116 + }, + { + "epoch": 0.21, + "grad_norm": 0.1919320335584178, + "learning_rate": 5.7073170731707317e-05, + "loss": 1.2886, + "step": 117 + }, + { + "epoch": 0.22, + "grad_norm": 0.18288775167828136, + "learning_rate": 5.75609756097561e-05, + "loss": 1.1875, + "step": 118 + }, + { + "epoch": 0.22, + "grad_norm": 0.18208588867750863, + "learning_rate": 5.804878048780488e-05, + "loss": 1.2388, + "step": 119 + }, + { + "epoch": 0.22, + "grad_norm": 0.1743260015658331, + "learning_rate": 5.853658536585366e-05, + "loss": 1.1762, + "step": 120 + }, + { + "epoch": 0.22, + "grad_norm": 0.17856046291517946, + "learning_rate": 5.902439024390244e-05, + "loss": 1.2888, + "step": 121 + }, + { + "epoch": 0.22, + "grad_norm": 0.17493794870966536, + "learning_rate": 5.9512195121951225e-05, + "loss": 1.2222, + "step": 122 + }, + { + "epoch": 0.22, + "grad_norm": 0.1909202655203384, + "learning_rate": 6.000000000000001e-05, + "loss": 1.2414, + "step": 123 + }, + { + "epoch": 0.23, + "grad_norm": 0.18345819482834988, + "learning_rate": 6.0487804878048785e-05, + "loss": 1.2756, + "step": 124 + }, + { + "epoch": 0.23, + "grad_norm": 0.2057069352956621, + "learning_rate": 6.097560975609757e-05, + "loss": 1.261, + "step": 125 + }, + { + "epoch": 0.23, + "grad_norm": 0.299775882469108, + "learning_rate": 6.146341463414634e-05, + "loss": 1.2566, + "step": 126 + }, + { + "epoch": 0.23, + "grad_norm": 0.1869687633018095, + "learning_rate": 6.195121951219513e-05, + "loss": 1.3039, + "step": 127 + }, + { + "epoch": 0.23, + "grad_norm": 0.17747149926197442, + "learning_rate": 6.243902439024391e-05, + "loss": 1.2524, + "step": 128 + }, + { + "epoch": 0.24, + "grad_norm": 0.17885157788044242, + "learning_rate": 6.29268292682927e-05, + "loss": 1.2455, + "step": 129 + }, + { + "epoch": 0.24, + "grad_norm": 0.17617298187845123, + "learning_rate": 6.341463414634148e-05, + "loss": 1.2009, + "step": 130 + }, + { + "epoch": 0.24, + "grad_norm": 0.20164176323497066, + "learning_rate": 6.390243902439025e-05, + "loss": 1.2634, + "step": 131 + }, + { + "epoch": 0.24, + "grad_norm": 0.20459903417307612, + "learning_rate": 6.439024390243903e-05, + "loss": 1.1963, + "step": 132 + }, + { + "epoch": 0.24, + "grad_norm": 0.1863755486334296, + "learning_rate": 6.487804878048781e-05, + "loss": 1.2387, + "step": 133 + }, + { + "epoch": 0.25, + "grad_norm": 0.19265866140295207, + "learning_rate": 6.536585365853658e-05, + "loss": 1.2688, + "step": 134 + }, + { + "epoch": 0.25, + "grad_norm": 0.1823425868969493, + "learning_rate": 6.585365853658536e-05, + "loss": 1.2041, + "step": 135 + }, + { + "epoch": 0.25, + "grad_norm": 0.2016853266472781, + "learning_rate": 6.634146341463415e-05, + "loss": 1.1223, + "step": 136 + }, + { + "epoch": 0.25, + "grad_norm": 0.17282675192463448, + "learning_rate": 6.682926829268293e-05, + "loss": 1.1879, + "step": 137 + }, + { + "epoch": 0.25, + "grad_norm": 0.17398811693399288, + "learning_rate": 6.731707317073171e-05, + "loss": 1.2682, + "step": 138 + }, + { + "epoch": 0.25, + "grad_norm": 0.18516916965434696, + "learning_rate": 6.78048780487805e-05, + "loss": 1.1666, + "step": 139 + }, + { + "epoch": 0.26, + "grad_norm": 0.1852213129647933, + "learning_rate": 6.829268292682927e-05, + "loss": 1.2501, + "step": 140 + }, + { + "epoch": 0.26, + "grad_norm": 0.17915948766591883, + "learning_rate": 6.878048780487805e-05, + "loss": 1.2264, + "step": 141 + }, + { + "epoch": 0.26, + "grad_norm": 0.21599939417233183, + "learning_rate": 6.926829268292683e-05, + "loss": 1.2376, + "step": 142 + }, + { + "epoch": 0.26, + "grad_norm": 0.17839304459521851, + "learning_rate": 6.975609756097562e-05, + "loss": 1.2353, + "step": 143 + }, + { + "epoch": 0.26, + "grad_norm": 0.20826913231380875, + "learning_rate": 7.02439024390244e-05, + "loss": 1.1901, + "step": 144 + }, + { + "epoch": 0.27, + "grad_norm": 0.20788894913361589, + "learning_rate": 7.073170731707318e-05, + "loss": 1.2577, + "step": 145 + }, + { + "epoch": 0.27, + "grad_norm": 0.18420055842301297, + "learning_rate": 7.121951219512195e-05, + "loss": 1.1393, + "step": 146 + }, + { + "epoch": 0.27, + "grad_norm": 0.19903048468685589, + "learning_rate": 7.170731707317073e-05, + "loss": 1.2321, + "step": 147 + }, + { + "epoch": 0.27, + "grad_norm": 0.19074116314985748, + "learning_rate": 7.219512195121952e-05, + "loss": 1.1912, + "step": 148 + }, + { + "epoch": 0.27, + "grad_norm": 0.2353816469403903, + "learning_rate": 7.26829268292683e-05, + "loss": 1.28, + "step": 149 + }, + { + "epoch": 0.27, + "grad_norm": 0.21634875684769345, + "learning_rate": 7.317073170731708e-05, + "loss": 1.3312, + "step": 150 + }, + { + "epoch": 0.28, + "grad_norm": 0.18290969006743918, + "learning_rate": 7.365853658536587e-05, + "loss": 1.2214, + "step": 151 + }, + { + "epoch": 0.28, + "grad_norm": 0.18484243897545208, + "learning_rate": 7.414634146341465e-05, + "loss": 1.1895, + "step": 152 + }, + { + "epoch": 0.28, + "grad_norm": 0.21882343112978872, + "learning_rate": 7.463414634146342e-05, + "loss": 1.2219, + "step": 153 + }, + { + "epoch": 0.28, + "grad_norm": 0.19868284379241205, + "learning_rate": 7.51219512195122e-05, + "loss": 1.2176, + "step": 154 + }, + { + "epoch": 0.28, + "grad_norm": 0.20912516312950613, + "learning_rate": 7.560975609756097e-05, + "loss": 1.242, + "step": 155 + }, + { + "epoch": 0.29, + "grad_norm": 0.23811880045549916, + "learning_rate": 7.609756097560976e-05, + "loss": 1.2838, + "step": 156 + }, + { + "epoch": 0.29, + "grad_norm": 0.19511077122033713, + "learning_rate": 7.658536585365854e-05, + "loss": 1.1594, + "step": 157 + }, + { + "epoch": 0.29, + "grad_norm": 0.20094129399534238, + "learning_rate": 7.707317073170732e-05, + "loss": 1.2966, + "step": 158 + }, + { + "epoch": 0.29, + "grad_norm": 0.19366245038292418, + "learning_rate": 7.75609756097561e-05, + "loss": 1.2246, + "step": 159 + }, + { + "epoch": 0.29, + "grad_norm": 0.19409570223867306, + "learning_rate": 7.804878048780489e-05, + "loss": 1.2312, + "step": 160 + }, + { + "epoch": 0.29, + "grad_norm": 0.2087258457033805, + "learning_rate": 7.853658536585366e-05, + "loss": 1.2169, + "step": 161 + }, + { + "epoch": 0.3, + "grad_norm": 0.18765223996270428, + "learning_rate": 7.902439024390244e-05, + "loss": 1.2383, + "step": 162 + }, + { + "epoch": 0.3, + "grad_norm": 0.20734180224147242, + "learning_rate": 7.951219512195122e-05, + "loss": 1.2587, + "step": 163 + }, + { + "epoch": 0.3, + "grad_norm": 0.24690929540287834, + "learning_rate": 8e-05, + "loss": 1.1951, + "step": 164 + }, + { + "epoch": 0.3, + "grad_norm": 0.2003538797619543, + "learning_rate": 7.999990914797545e-05, + "loss": 1.1982, + "step": 165 + }, + { + "epoch": 0.3, + "grad_norm": 0.22469075613510484, + "learning_rate": 7.99996365923145e-05, + "loss": 1.2355, + "step": 166 + }, + { + "epoch": 0.31, + "grad_norm": 0.21870100788336058, + "learning_rate": 7.999918233425526e-05, + "loss": 1.1103, + "step": 167 + }, + { + "epoch": 0.31, + "grad_norm": 0.20939989594131886, + "learning_rate": 7.999854637586122e-05, + "loss": 1.1966, + "step": 168 + }, + { + "epoch": 0.31, + "grad_norm": 0.43108211416237796, + "learning_rate": 7.999772872002132e-05, + "loss": 1.2882, + "step": 169 + }, + { + "epoch": 0.31, + "grad_norm": 0.27045413432174487, + "learning_rate": 7.999672937044984e-05, + "loss": 1.2399, + "step": 170 + }, + { + "epoch": 0.31, + "grad_norm": 0.19700483036740515, + "learning_rate": 7.999554833168642e-05, + "loss": 1.202, + "step": 171 + }, + { + "epoch": 0.31, + "grad_norm": 0.3335979493370708, + "learning_rate": 7.999418560909604e-05, + "loss": 1.1995, + "step": 172 + }, + { + "epoch": 0.32, + "grad_norm": 0.3165803974474567, + "learning_rate": 7.999264120886902e-05, + "loss": 1.1569, + "step": 173 + }, + { + "epoch": 0.32, + "grad_norm": 0.1951699080346223, + "learning_rate": 7.999091513802093e-05, + "loss": 1.1778, + "step": 174 + }, + { + "epoch": 0.32, + "grad_norm": 0.2087559121749787, + "learning_rate": 7.998900740439265e-05, + "loss": 1.1736, + "step": 175 + }, + { + "epoch": 0.32, + "grad_norm": 0.20345180977460478, + "learning_rate": 7.998691801665024e-05, + "loss": 1.2281, + "step": 176 + }, + { + "epoch": 0.32, + "grad_norm": 0.24617644827252333, + "learning_rate": 7.998464698428495e-05, + "loss": 1.2072, + "step": 177 + }, + { + "epoch": 0.33, + "grad_norm": 0.2469050959356265, + "learning_rate": 7.998219431761318e-05, + "loss": 1.2242, + "step": 178 + }, + { + "epoch": 0.33, + "grad_norm": 0.19529317748460623, + "learning_rate": 7.997956002777642e-05, + "loss": 1.2567, + "step": 179 + }, + { + "epoch": 0.33, + "grad_norm": 0.19048389491381376, + "learning_rate": 7.99767441267412e-05, + "loss": 1.2982, + "step": 180 + }, + { + "epoch": 0.33, + "grad_norm": 0.2085799116493225, + "learning_rate": 7.997374662729904e-05, + "loss": 1.1254, + "step": 181 + }, + { + "epoch": 0.33, + "grad_norm": 0.20636853256378995, + "learning_rate": 7.997056754306636e-05, + "loss": 1.2435, + "step": 182 + }, + { + "epoch": 0.33, + "grad_norm": 0.20590016382290252, + "learning_rate": 7.99672068884845e-05, + "loss": 1.2658, + "step": 183 + }, + { + "epoch": 0.34, + "grad_norm": 0.1931166169764433, + "learning_rate": 7.996366467881955e-05, + "loss": 1.1637, + "step": 184 + }, + { + "epoch": 0.34, + "grad_norm": 0.18873318157988098, + "learning_rate": 7.995994093016237e-05, + "loss": 1.1335, + "step": 185 + }, + { + "epoch": 0.34, + "grad_norm": 0.19210254625199108, + "learning_rate": 7.995603565942846e-05, + "loss": 1.1928, + "step": 186 + }, + { + "epoch": 0.34, + "grad_norm": 0.2130986479765664, + "learning_rate": 7.995194888435792e-05, + "loss": 1.2158, + "step": 187 + }, + { + "epoch": 0.34, + "grad_norm": 0.22003854501814088, + "learning_rate": 7.994768062351532e-05, + "loss": 1.2288, + "step": 188 + }, + { + "epoch": 0.35, + "grad_norm": 0.20330803191993058, + "learning_rate": 7.994323089628968e-05, + "loss": 1.2426, + "step": 189 + }, + { + "epoch": 0.35, + "grad_norm": 0.20567314642208634, + "learning_rate": 7.993859972289434e-05, + "loss": 1.2649, + "step": 190 + }, + { + "epoch": 0.35, + "grad_norm": 0.21556663727342962, + "learning_rate": 7.993378712436686e-05, + "loss": 1.2545, + "step": 191 + }, + { + "epoch": 0.35, + "grad_norm": 0.20309165469109888, + "learning_rate": 7.992879312256897e-05, + "loss": 1.3338, + "step": 192 + }, + { + "epoch": 0.35, + "grad_norm": 0.19574356669421325, + "learning_rate": 7.992361774018641e-05, + "loss": 1.278, + "step": 193 + }, + { + "epoch": 0.35, + "grad_norm": 0.2763613746722313, + "learning_rate": 7.991826100072891e-05, + "loss": 1.2571, + "step": 194 + }, + { + "epoch": 0.36, + "grad_norm": 0.19346552479915102, + "learning_rate": 7.991272292852996e-05, + "loss": 1.2027, + "step": 195 + }, + { + "epoch": 0.36, + "grad_norm": 0.2281167812123908, + "learning_rate": 7.990700354874683e-05, + "loss": 1.2586, + "step": 196 + }, + { + "epoch": 0.36, + "grad_norm": 0.19699013712137542, + "learning_rate": 7.990110288736042e-05, + "loss": 1.1371, + "step": 197 + }, + { + "epoch": 0.36, + "grad_norm": 0.21768209981475933, + "learning_rate": 7.989502097117503e-05, + "loss": 1.2522, + "step": 198 + }, + { + "epoch": 0.36, + "grad_norm": 0.21335427847754582, + "learning_rate": 7.988875782781838e-05, + "loss": 1.2437, + "step": 199 + }, + { + "epoch": 0.37, + "grad_norm": 0.21856710629066897, + "learning_rate": 7.988231348574147e-05, + "loss": 1.2135, + "step": 200 + }, + { + "epoch": 0.37, + "grad_norm": 0.20482062658774797, + "learning_rate": 7.987568797421836e-05, + "loss": 1.1755, + "step": 201 + }, + { + "epoch": 0.37, + "grad_norm": 0.2017756813960897, + "learning_rate": 7.986888132334608e-05, + "loss": 1.1699, + "step": 202 + }, + { + "epoch": 0.37, + "grad_norm": 0.20496443848153809, + "learning_rate": 7.986189356404458e-05, + "loss": 1.2125, + "step": 203 + }, + { + "epoch": 0.37, + "grad_norm": 0.2134603800558358, + "learning_rate": 7.985472472805643e-05, + "loss": 1.2391, + "step": 204 + }, + { + "epoch": 0.37, + "grad_norm": 0.2364175573420861, + "learning_rate": 7.98473748479468e-05, + "loss": 1.2384, + "step": 205 + }, + { + "epoch": 0.38, + "grad_norm": 0.1872419861598724, + "learning_rate": 7.983984395710326e-05, + "loss": 1.1457, + "step": 206 + }, + { + "epoch": 0.38, + "grad_norm": 0.28222194007095774, + "learning_rate": 7.983213208973566e-05, + "loss": 1.2952, + "step": 207 + }, + { + "epoch": 0.38, + "grad_norm": 0.1916094851162064, + "learning_rate": 7.982423928087593e-05, + "loss": 1.1763, + "step": 208 + }, + { + "epoch": 0.38, + "grad_norm": 0.18446245256166657, + "learning_rate": 7.981616556637795e-05, + "loss": 1.1863, + "step": 209 + }, + { + "epoch": 0.38, + "grad_norm": 0.195191961022491, + "learning_rate": 7.980791098291737e-05, + "loss": 1.2036, + "step": 210 + }, + { + "epoch": 0.39, + "grad_norm": 0.2652439657825496, + "learning_rate": 7.979947556799151e-05, + "loss": 1.2834, + "step": 211 + }, + { + "epoch": 0.39, + "grad_norm": 0.24308438957843412, + "learning_rate": 7.979085935991906e-05, + "loss": 1.234, + "step": 212 + }, + { + "epoch": 0.39, + "grad_norm": 0.21294701043622016, + "learning_rate": 7.978206239784004e-05, + "loss": 1.3006, + "step": 213 + }, + { + "epoch": 0.39, + "grad_norm": 0.25809277041859524, + "learning_rate": 7.977308472171553e-05, + "loss": 1.2272, + "step": 214 + }, + { + "epoch": 0.39, + "grad_norm": 0.193463860107294, + "learning_rate": 7.976392637232754e-05, + "loss": 1.2295, + "step": 215 + }, + { + "epoch": 0.4, + "grad_norm": 0.2150023760609626, + "learning_rate": 7.975458739127877e-05, + "loss": 1.2135, + "step": 216 + }, + { + "epoch": 0.4, + "grad_norm": 0.22590495955605894, + "learning_rate": 7.974506782099253e-05, + "loss": 1.2532, + "step": 217 + }, + { + "epoch": 0.4, + "grad_norm": 0.21023744668403702, + "learning_rate": 7.973536770471242e-05, + "loss": 1.2472, + "step": 218 + }, + { + "epoch": 0.4, + "grad_norm": 0.2345749799511543, + "learning_rate": 7.972548708650218e-05, + "loss": 1.1791, + "step": 219 + }, + { + "epoch": 0.4, + "grad_norm": 0.2158876734005217, + "learning_rate": 7.971542601124553e-05, + "loss": 1.2483, + "step": 220 + }, + { + "epoch": 0.4, + "grad_norm": 0.29455339949432446, + "learning_rate": 7.970518452464593e-05, + "loss": 1.2894, + "step": 221 + }, + { + "epoch": 0.41, + "grad_norm": 0.23983708730626851, + "learning_rate": 7.969476267322636e-05, + "loss": 1.271, + "step": 222 + }, + { + "epoch": 0.41, + "grad_norm": 0.1922400905426158, + "learning_rate": 7.968416050432912e-05, + "loss": 1.2139, + "step": 223 + }, + { + "epoch": 0.41, + "grad_norm": 0.2238136844422931, + "learning_rate": 7.967337806611568e-05, + "loss": 1.2655, + "step": 224 + }, + { + "epoch": 0.41, + "grad_norm": 0.21230292828267672, + "learning_rate": 7.966241540756631e-05, + "loss": 1.2406, + "step": 225 + }, + { + "epoch": 0.41, + "grad_norm": 0.26656119419070456, + "learning_rate": 7.965127257848004e-05, + "loss": 1.2595, + "step": 226 + }, + { + "epoch": 0.42, + "grad_norm": 0.22381385502992684, + "learning_rate": 7.963994962947426e-05, + "loss": 1.1737, + "step": 227 + }, + { + "epoch": 0.42, + "grad_norm": 0.20056702203994298, + "learning_rate": 7.962844661198462e-05, + "loss": 1.1969, + "step": 228 + }, + { + "epoch": 0.42, + "grad_norm": 0.20148701321526885, + "learning_rate": 7.961676357826478e-05, + "loss": 1.2151, + "step": 229 + }, + { + "epoch": 0.42, + "grad_norm": 0.20034834807028637, + "learning_rate": 7.960490058138604e-05, + "loss": 1.1455, + "step": 230 + }, + { + "epoch": 0.42, + "grad_norm": 0.21050838521846033, + "learning_rate": 7.959285767523732e-05, + "loss": 1.2223, + "step": 231 + }, + { + "epoch": 0.42, + "grad_norm": 0.20904772138969777, + "learning_rate": 7.95806349145247e-05, + "loss": 1.2534, + "step": 232 + }, + { + "epoch": 0.43, + "grad_norm": 0.20307877304792957, + "learning_rate": 7.956823235477134e-05, + "loss": 1.1352, + "step": 233 + }, + { + "epoch": 0.43, + "grad_norm": 0.20501105270897094, + "learning_rate": 7.95556500523171e-05, + "loss": 1.2031, + "step": 234 + }, + { + "epoch": 0.43, + "grad_norm": 0.19800586972038586, + "learning_rate": 7.954288806431838e-05, + "loss": 1.2567, + "step": 235 + }, + { + "epoch": 0.43, + "grad_norm": 0.2175102450594135, + "learning_rate": 7.952994644874777e-05, + "loss": 1.2538, + "step": 236 + }, + { + "epoch": 0.43, + "grad_norm": 0.22698189300067595, + "learning_rate": 7.951682526439391e-05, + "loss": 1.3088, + "step": 237 + }, + { + "epoch": 0.44, + "grad_norm": 0.19208392014975315, + "learning_rate": 7.950352457086109e-05, + "loss": 1.2336, + "step": 238 + }, + { + "epoch": 0.44, + "grad_norm": 0.27004086334319655, + "learning_rate": 7.949004442856905e-05, + "loss": 1.2012, + "step": 239 + }, + { + "epoch": 0.44, + "grad_norm": 0.23420974954538043, + "learning_rate": 7.947638489875272e-05, + "loss": 1.2244, + "step": 240 + }, + { + "epoch": 0.44, + "grad_norm": 0.20514399124802024, + "learning_rate": 7.946254604346186e-05, + "loss": 1.2548, + "step": 241 + }, + { + "epoch": 0.44, + "grad_norm": 0.19334973602372896, + "learning_rate": 7.944852792556092e-05, + "loss": 1.2104, + "step": 242 + }, + { + "epoch": 0.44, + "grad_norm": 0.1992640714537956, + "learning_rate": 7.943433060872858e-05, + "loss": 1.2628, + "step": 243 + }, + { + "epoch": 0.45, + "grad_norm": 0.203284617090413, + "learning_rate": 7.941995415745761e-05, + "loss": 1.2002, + "step": 244 + }, + { + "epoch": 0.45, + "grad_norm": 0.22795306969682058, + "learning_rate": 7.94053986370545e-05, + "loss": 1.2215, + "step": 245 + }, + { + "epoch": 0.45, + "grad_norm": 0.20789041346838505, + "learning_rate": 7.939066411363915e-05, + "loss": 1.0998, + "step": 246 + }, + { + "epoch": 0.45, + "grad_norm": 0.22354868884742066, + "learning_rate": 7.937575065414464e-05, + "loss": 1.2564, + "step": 247 + }, + { + "epoch": 0.45, + "grad_norm": 0.21176392726647736, + "learning_rate": 7.936065832631687e-05, + "loss": 1.2816, + "step": 248 + }, + { + "epoch": 0.46, + "grad_norm": 0.19967179557235587, + "learning_rate": 7.934538719871427e-05, + "loss": 1.1961, + "step": 249 + }, + { + "epoch": 0.46, + "grad_norm": 0.210819577350627, + "learning_rate": 7.932993734070747e-05, + "loss": 1.2167, + "step": 250 + }, + { + "epoch": 0.46, + "grad_norm": 0.21537794551756187, + "learning_rate": 7.931430882247903e-05, + "loss": 1.2341, + "step": 251 + }, + { + "epoch": 0.46, + "grad_norm": 0.22850872387256574, + "learning_rate": 7.929850171502304e-05, + "loss": 1.1686, + "step": 252 + }, + { + "epoch": 0.46, + "grad_norm": 0.22380366415076383, + "learning_rate": 7.928251609014493e-05, + "loss": 1.1462, + "step": 253 + }, + { + "epoch": 0.46, + "grad_norm": 0.22426923149036065, + "learning_rate": 7.926635202046102e-05, + "loss": 1.1792, + "step": 254 + }, + { + "epoch": 0.47, + "grad_norm": 0.42082703321103965, + "learning_rate": 7.925000957939822e-05, + "loss": 1.2718, + "step": 255 + }, + { + "epoch": 0.47, + "grad_norm": 0.2235432774854074, + "learning_rate": 7.92334888411937e-05, + "loss": 1.2598, + "step": 256 + }, + { + "epoch": 0.47, + "grad_norm": 0.281644028934108, + "learning_rate": 7.92167898808946e-05, + "loss": 1.2205, + "step": 257 + }, + { + "epoch": 0.47, + "grad_norm": 0.2037705143888748, + "learning_rate": 7.919991277435763e-05, + "loss": 1.1737, + "step": 258 + }, + { + "epoch": 0.47, + "grad_norm": 0.20917419230028977, + "learning_rate": 7.918285759824879e-05, + "loss": 1.2035, + "step": 259 + }, + { + "epoch": 0.48, + "grad_norm": 0.20510847570635518, + "learning_rate": 7.916562443004292e-05, + "loss": 1.2135, + "step": 260 + }, + { + "epoch": 0.48, + "grad_norm": 0.25172483071092466, + "learning_rate": 7.914821334802342e-05, + "loss": 1.2218, + "step": 261 + }, + { + "epoch": 0.48, + "grad_norm": 0.21102706700634313, + "learning_rate": 7.91306244312819e-05, + "loss": 1.1738, + "step": 262 + }, + { + "epoch": 0.48, + "grad_norm": 0.22626060872645815, + "learning_rate": 7.911285775971781e-05, + "loss": 1.238, + "step": 263 + }, + { + "epoch": 0.48, + "grad_norm": 0.22448567539778486, + "learning_rate": 7.909491341403805e-05, + "loss": 1.2404, + "step": 264 + }, + { + "epoch": 0.48, + "grad_norm": 0.2019099786139193, + "learning_rate": 7.907679147575661e-05, + "loss": 1.213, + "step": 265 + }, + { + "epoch": 0.49, + "grad_norm": 0.24307234839096267, + "learning_rate": 7.905849202719422e-05, + "loss": 1.2322, + "step": 266 + }, + { + "epoch": 0.49, + "grad_norm": 0.19801890521743487, + "learning_rate": 7.904001515147802e-05, + "loss": 1.2448, + "step": 267 + }, + { + "epoch": 0.49, + "grad_norm": 0.2102742273575385, + "learning_rate": 7.902136093254106e-05, + "loss": 1.1657, + "step": 268 + }, + { + "epoch": 0.49, + "grad_norm": 0.2173464476815016, + "learning_rate": 7.900252945512201e-05, + "loss": 1.2549, + "step": 269 + }, + { + "epoch": 0.49, + "grad_norm": 0.20957275458699595, + "learning_rate": 7.898352080476479e-05, + "loss": 1.2536, + "step": 270 + }, + { + "epoch": 0.5, + "grad_norm": 0.20691966388952363, + "learning_rate": 7.896433506781811e-05, + "loss": 1.2661, + "step": 271 + }, + { + "epoch": 0.5, + "grad_norm": 0.2276662275112648, + "learning_rate": 7.894497233143509e-05, + "loss": 1.2409, + "step": 272 + }, + { + "epoch": 0.5, + "grad_norm": 0.23854109569301263, + "learning_rate": 7.892543268357297e-05, + "loss": 1.2681, + "step": 273 + }, + { + "epoch": 0.5, + "grad_norm": 0.2233864156677627, + "learning_rate": 7.890571621299252e-05, + "loss": 1.1687, + "step": 274 + }, + { + "epoch": 0.5, + "grad_norm": 0.20114129147925475, + "learning_rate": 7.888582300925787e-05, + "loss": 1.2184, + "step": 275 + }, + { + "epoch": 0.5, + "grad_norm": 0.2154654670569462, + "learning_rate": 7.886575316273586e-05, + "loss": 1.1982, + "step": 276 + }, + { + "epoch": 0.51, + "grad_norm": 0.2292982209343639, + "learning_rate": 7.884550676459583e-05, + "loss": 1.2129, + "step": 277 + }, + { + "epoch": 0.51, + "grad_norm": 0.21302713135229548, + "learning_rate": 7.882508390680908e-05, + "loss": 1.1605, + "step": 278 + }, + { + "epoch": 0.51, + "grad_norm": 0.2123661020671048, + "learning_rate": 7.88044846821485e-05, + "loss": 1.2308, + "step": 279 + }, + { + "epoch": 0.51, + "grad_norm": 0.2080577410800404, + "learning_rate": 7.878370918418818e-05, + "loss": 1.2195, + "step": 280 + }, + { + "epoch": 0.51, + "grad_norm": 0.19663901881127385, + "learning_rate": 7.876275750730289e-05, + "loss": 1.1591, + "step": 281 + }, + { + "epoch": 0.52, + "grad_norm": 0.20534502031312163, + "learning_rate": 7.874162974666776e-05, + "loss": 1.2664, + "step": 282 + }, + { + "epoch": 0.52, + "grad_norm": 0.23240445399513837, + "learning_rate": 7.872032599825779e-05, + "loss": 1.2151, + "step": 283 + }, + { + "epoch": 0.52, + "grad_norm": 0.2672527316717507, + "learning_rate": 7.86988463588474e-05, + "loss": 1.2406, + "step": 284 + }, + { + "epoch": 0.52, + "grad_norm": 0.19893903058743695, + "learning_rate": 7.867719092601003e-05, + "loss": 1.1291, + "step": 285 + }, + { + "epoch": 0.52, + "grad_norm": 0.33275268109930917, + "learning_rate": 7.865535979811768e-05, + "loss": 1.1406, + "step": 286 + }, + { + "epoch": 0.52, + "grad_norm": 0.2373619455690358, + "learning_rate": 7.863335307434045e-05, + "loss": 1.2799, + "step": 287 + }, + { + "epoch": 0.53, + "grad_norm": 0.263235735390858, + "learning_rate": 7.861117085464612e-05, + "loss": 1.2415, + "step": 288 + }, + { + "epoch": 0.53, + "grad_norm": 0.25884281780784324, + "learning_rate": 7.858881323979965e-05, + "loss": 1.3919, + "step": 289 + }, + { + "epoch": 0.53, + "grad_norm": 0.25426288332255736, + "learning_rate": 7.85662803313628e-05, + "loss": 1.174, + "step": 290 + }, + { + "epoch": 0.53, + "grad_norm": 0.26655405527881243, + "learning_rate": 7.854357223169356e-05, + "loss": 1.2806, + "step": 291 + }, + { + "epoch": 0.53, + "grad_norm": 0.20909844432349833, + "learning_rate": 7.852068904394579e-05, + "loss": 1.2627, + "step": 292 + }, + { + "epoch": 0.54, + "grad_norm": 0.21307115068935759, + "learning_rate": 7.849763087206866e-05, + "loss": 1.1879, + "step": 293 + }, + { + "epoch": 0.54, + "grad_norm": 0.25009949471398946, + "learning_rate": 7.847439782080628e-05, + "loss": 1.2881, + "step": 294 + }, + { + "epoch": 0.54, + "grad_norm": 0.20960783418679174, + "learning_rate": 7.845098999569712e-05, + "loss": 1.2723, + "step": 295 + }, + { + "epoch": 0.54, + "grad_norm": 0.24968832437925104, + "learning_rate": 7.842740750307362e-05, + "loss": 1.2029, + "step": 296 + }, + { + "epoch": 0.54, + "grad_norm": 0.22981196585125677, + "learning_rate": 7.84036504500616e-05, + "loss": 1.1695, + "step": 297 + }, + { + "epoch": 0.55, + "grad_norm": 0.2320606844751365, + "learning_rate": 7.837971894457991e-05, + "loss": 1.2317, + "step": 298 + }, + { + "epoch": 0.55, + "grad_norm": 0.23051459673906124, + "learning_rate": 7.835561309533981e-05, + "loss": 1.2046, + "step": 299 + }, + { + "epoch": 0.55, + "grad_norm": 0.2510027231060586, + "learning_rate": 7.833133301184457e-05, + "loss": 1.199, + "step": 300 + }, + { + "epoch": 0.55, + "grad_norm": 0.23601180466018787, + "learning_rate": 7.830687880438895e-05, + "loss": 1.1755, + "step": 301 + }, + { + "epoch": 0.55, + "grad_norm": 0.24740820934385369, + "learning_rate": 7.828225058405864e-05, + "loss": 1.2054, + "step": 302 + }, + { + "epoch": 0.55, + "grad_norm": 0.23065372979111173, + "learning_rate": 7.825744846272984e-05, + "loss": 1.2066, + "step": 303 + }, + { + "epoch": 0.56, + "grad_norm": 0.22385077334838213, + "learning_rate": 7.823247255306866e-05, + "loss": 1.2147, + "step": 304 + }, + { + "epoch": 0.56, + "grad_norm": 0.42981213948386104, + "learning_rate": 7.820732296853074e-05, + "loss": 1.2314, + "step": 305 + }, + { + "epoch": 0.56, + "grad_norm": 0.21122844902751076, + "learning_rate": 7.818199982336058e-05, + "loss": 1.1462, + "step": 306 + }, + { + "epoch": 0.56, + "grad_norm": 0.23374869692118933, + "learning_rate": 7.815650323259117e-05, + "loss": 1.2051, + "step": 307 + }, + { + "epoch": 0.56, + "grad_norm": 0.21662363795962128, + "learning_rate": 7.813083331204332e-05, + "loss": 1.1575, + "step": 308 + }, + { + "epoch": 0.57, + "grad_norm": 0.2088315773384112, + "learning_rate": 7.810499017832526e-05, + "loss": 1.1316, + "step": 309 + }, + { + "epoch": 0.57, + "grad_norm": 0.2095238410730976, + "learning_rate": 7.807897394883203e-05, + "loss": 1.2087, + "step": 310 + }, + { + "epoch": 0.57, + "grad_norm": 0.22672932127256515, + "learning_rate": 7.805278474174499e-05, + "loss": 1.2512, + "step": 311 + }, + { + "epoch": 0.57, + "grad_norm": 0.21873052340922736, + "learning_rate": 7.802642267603126e-05, + "loss": 1.1909, + "step": 312 + }, + { + "epoch": 0.57, + "grad_norm": 0.219814521916342, + "learning_rate": 7.79998878714432e-05, + "loss": 1.1669, + "step": 313 + }, + { + "epoch": 0.57, + "grad_norm": 0.3049426027257317, + "learning_rate": 7.797318044851786e-05, + "loss": 1.1797, + "step": 314 + }, + { + "epoch": 0.58, + "grad_norm": 0.22309435690065985, + "learning_rate": 7.794630052857638e-05, + "loss": 1.1417, + "step": 315 + }, + { + "epoch": 0.58, + "grad_norm": 0.3891885169154885, + "learning_rate": 7.791924823372354e-05, + "loss": 1.2369, + "step": 316 + }, + { + "epoch": 0.58, + "grad_norm": 0.24780269452456372, + "learning_rate": 7.789202368684711e-05, + "loss": 1.2521, + "step": 317 + }, + { + "epoch": 0.58, + "grad_norm": 0.21660460720269362, + "learning_rate": 7.786462701161738e-05, + "loss": 1.2151, + "step": 318 + }, + { + "epoch": 0.58, + "grad_norm": 0.23635409466561857, + "learning_rate": 7.783705833248649e-05, + "loss": 1.2363, + "step": 319 + }, + { + "epoch": 0.59, + "grad_norm": 0.2616135839903218, + "learning_rate": 7.780931777468797e-05, + "loss": 1.2428, + "step": 320 + }, + { + "epoch": 0.59, + "grad_norm": 0.21461059159245083, + "learning_rate": 7.77814054642361e-05, + "loss": 1.1434, + "step": 321 + }, + { + "epoch": 0.59, + "grad_norm": 0.25348824286656163, + "learning_rate": 7.775332152792539e-05, + "loss": 1.2368, + "step": 322 + }, + { + "epoch": 0.59, + "grad_norm": 0.22275034726331247, + "learning_rate": 7.772506609332995e-05, + "loss": 1.1827, + "step": 323 + }, + { + "epoch": 0.59, + "grad_norm": 0.25030821228147526, + "learning_rate": 7.769663928880298e-05, + "loss": 1.2428, + "step": 324 + }, + { + "epoch": 0.59, + "grad_norm": 0.22251804398745534, + "learning_rate": 7.766804124347608e-05, + "loss": 1.1889, + "step": 325 + }, + { + "epoch": 0.6, + "grad_norm": 0.23381455520411995, + "learning_rate": 7.763927208725879e-05, + "loss": 1.2115, + "step": 326 + }, + { + "epoch": 0.6, + "grad_norm": 0.27341902651946226, + "learning_rate": 7.761033195083791e-05, + "loss": 1.2535, + "step": 327 + }, + { + "epoch": 0.6, + "grad_norm": 0.24862471659814522, + "learning_rate": 7.758122096567694e-05, + "loss": 1.2128, + "step": 328 + }, + { + "epoch": 0.6, + "grad_norm": 0.2251357082045494, + "learning_rate": 7.755193926401547e-05, + "loss": 1.2334, + "step": 329 + }, + { + "epoch": 0.6, + "grad_norm": 0.3173274941622932, + "learning_rate": 7.752248697886857e-05, + "loss": 1.226, + "step": 330 + }, + { + "epoch": 0.61, + "grad_norm": 0.23056440717672175, + "learning_rate": 7.74928642440263e-05, + "loss": 1.2339, + "step": 331 + }, + { + "epoch": 0.61, + "grad_norm": 0.2801507500859342, + "learning_rate": 7.746307119405286e-05, + "loss": 1.287, + "step": 332 + }, + { + "epoch": 0.61, + "grad_norm": 0.2267818430426272, + "learning_rate": 7.743310796428622e-05, + "loss": 1.1916, + "step": 333 + }, + { + "epoch": 0.61, + "grad_norm": 0.2777329160365585, + "learning_rate": 7.74029746908374e-05, + "loss": 1.252, + "step": 334 + }, + { + "epoch": 0.61, + "grad_norm": 0.25289169762353, + "learning_rate": 7.737267151058983e-05, + "loss": 1.2153, + "step": 335 + }, + { + "epoch": 0.61, + "grad_norm": 0.2424670686901653, + "learning_rate": 7.734219856119875e-05, + "loss": 1.2227, + "step": 336 + }, + { + "epoch": 0.62, + "grad_norm": 0.22747092217441645, + "learning_rate": 7.731155598109067e-05, + "loss": 1.19, + "step": 337 + }, + { + "epoch": 0.62, + "grad_norm": 0.2307810940100189, + "learning_rate": 7.728074390946257e-05, + "loss": 1.1818, + "step": 338 + }, + { + "epoch": 0.62, + "grad_norm": 0.2583402574655623, + "learning_rate": 7.724976248628142e-05, + "loss": 1.1608, + "step": 339 + }, + { + "epoch": 0.62, + "grad_norm": 0.22140209760890694, + "learning_rate": 7.721861185228347e-05, + "loss": 1.1245, + "step": 340 + }, + { + "epoch": 0.62, + "grad_norm": 0.25859310758244686, + "learning_rate": 7.718729214897362e-05, + "loss": 1.2247, + "step": 341 + }, + { + "epoch": 0.63, + "grad_norm": 0.26371179531372124, + "learning_rate": 7.715580351862482e-05, + "loss": 1.2128, + "step": 342 + }, + { + "epoch": 0.63, + "grad_norm": 0.26575541302851047, + "learning_rate": 7.712414610427733e-05, + "loss": 1.2443, + "step": 343 + }, + { + "epoch": 0.63, + "grad_norm": 0.269978305197599, + "learning_rate": 7.709232004973816e-05, + "loss": 1.2231, + "step": 344 + }, + { + "epoch": 0.63, + "grad_norm": 0.26583998705977047, + "learning_rate": 7.70603254995804e-05, + "loss": 1.2476, + "step": 345 + }, + { + "epoch": 0.63, + "grad_norm": 0.24256062164066097, + "learning_rate": 7.702816259914253e-05, + "loss": 1.2901, + "step": 346 + }, + { + "epoch": 0.63, + "grad_norm": 0.3463123472658915, + "learning_rate": 7.699583149452779e-05, + "loss": 1.3277, + "step": 347 + }, + { + "epoch": 0.64, + "grad_norm": 0.2269096590531878, + "learning_rate": 7.696333233260345e-05, + "loss": 1.2047, + "step": 348 + }, + { + "epoch": 0.64, + "grad_norm": 0.25136883001050025, + "learning_rate": 7.693066526100031e-05, + "loss": 1.1619, + "step": 349 + }, + { + "epoch": 0.64, + "grad_norm": 0.2565112571116145, + "learning_rate": 7.68978304281118e-05, + "loss": 1.2389, + "step": 350 + }, + { + "epoch": 0.64, + "grad_norm": 0.22175779550828703, + "learning_rate": 7.686482798309349e-05, + "loss": 1.2238, + "step": 351 + }, + { + "epoch": 0.64, + "grad_norm": 0.22588304332216555, + "learning_rate": 7.683165807586234e-05, + "loss": 1.174, + "step": 352 + }, + { + "epoch": 0.65, + "grad_norm": 0.24889474296529737, + "learning_rate": 7.6798320857096e-05, + "loss": 1.2366, + "step": 353 + }, + { + "epoch": 0.65, + "grad_norm": 0.27339703806525034, + "learning_rate": 7.676481647823214e-05, + "loss": 1.2356, + "step": 354 + }, + { + "epoch": 0.65, + "grad_norm": 0.23424666722888365, + "learning_rate": 7.673114509146782e-05, + "loss": 1.2089, + "step": 355 + }, + { + "epoch": 0.65, + "grad_norm": 0.27978285392461766, + "learning_rate": 7.66973068497587e-05, + "loss": 1.2609, + "step": 356 + }, + { + "epoch": 0.65, + "grad_norm": 0.2509423350138824, + "learning_rate": 7.666330190681844e-05, + "loss": 1.1777, + "step": 357 + }, + { + "epoch": 0.65, + "grad_norm": 0.23007730927468031, + "learning_rate": 7.662913041711793e-05, + "loss": 1.154, + "step": 358 + }, + { + "epoch": 0.66, + "grad_norm": 0.2438648674953112, + "learning_rate": 7.659479253588462e-05, + "loss": 1.2257, + "step": 359 + }, + { + "epoch": 0.66, + "grad_norm": 0.28816093242092233, + "learning_rate": 7.65602884191018e-05, + "loss": 1.2558, + "step": 360 + }, + { + "epoch": 0.66, + "grad_norm": 0.24972815300596035, + "learning_rate": 7.652561822350793e-05, + "loss": 1.2837, + "step": 361 + }, + { + "epoch": 0.66, + "grad_norm": 0.2543189139697063, + "learning_rate": 7.649078210659587e-05, + "loss": 1.2193, + "step": 362 + }, + { + "epoch": 0.66, + "grad_norm": 0.2237937956718952, + "learning_rate": 7.645578022661224e-05, + "loss": 1.2237, + "step": 363 + }, + { + "epoch": 0.67, + "grad_norm": 0.29742029408787396, + "learning_rate": 7.642061274255657e-05, + "loss": 1.2116, + "step": 364 + }, + { + "epoch": 0.67, + "grad_norm": 0.2462883147335493, + "learning_rate": 7.638527981418075e-05, + "loss": 1.1827, + "step": 365 + }, + { + "epoch": 0.67, + "grad_norm": 0.2647802498907096, + "learning_rate": 7.634978160198817e-05, + "loss": 1.2739, + "step": 366 + }, + { + "epoch": 0.67, + "grad_norm": 0.22360398779217264, + "learning_rate": 7.631411826723306e-05, + "loss": 1.2185, + "step": 367 + }, + { + "epoch": 0.67, + "grad_norm": 0.2635048004593543, + "learning_rate": 7.627828997191973e-05, + "loss": 1.2317, + "step": 368 + }, + { + "epoch": 0.67, + "grad_norm": 0.2764803449917684, + "learning_rate": 7.624229687880184e-05, + "loss": 1.1923, + "step": 369 + }, + { + "epoch": 0.68, + "grad_norm": 0.25724943233414527, + "learning_rate": 7.620613915138166e-05, + "loss": 1.2218, + "step": 370 + }, + { + "epoch": 0.68, + "grad_norm": 0.2858318045794755, + "learning_rate": 7.61698169539093e-05, + "loss": 1.1496, + "step": 371 + }, + { + "epoch": 0.68, + "grad_norm": 0.23547216647460364, + "learning_rate": 7.613333045138206e-05, + "loss": 1.1905, + "step": 372 + }, + { + "epoch": 0.68, + "grad_norm": 0.22984814903684375, + "learning_rate": 7.609667980954355e-05, + "loss": 1.2009, + "step": 373 + }, + { + "epoch": 0.68, + "grad_norm": 0.2551903754079084, + "learning_rate": 7.605986519488301e-05, + "loss": 1.2042, + "step": 374 + }, + { + "epoch": 0.69, + "grad_norm": 0.2508257410125616, + "learning_rate": 7.602288677463457e-05, + "loss": 1.2468, + "step": 375 + }, + { + "epoch": 0.69, + "grad_norm": 0.25324577774935964, + "learning_rate": 7.598574471677644e-05, + "loss": 1.2603, + "step": 376 + }, + { + "epoch": 0.69, + "grad_norm": 0.35888776531769967, + "learning_rate": 7.59484391900302e-05, + "loss": 1.1929, + "step": 377 + }, + { + "epoch": 0.69, + "grad_norm": 0.22048517191014724, + "learning_rate": 7.591097036385994e-05, + "loss": 1.1783, + "step": 378 + }, + { + "epoch": 0.69, + "grad_norm": 0.2781160412746083, + "learning_rate": 7.587333840847162e-05, + "loss": 1.3397, + "step": 379 + }, + { + "epoch": 0.7, + "grad_norm": 0.24033046830332258, + "learning_rate": 7.583554349481222e-05, + "loss": 1.2436, + "step": 380 + }, + { + "epoch": 0.7, + "grad_norm": 0.26413762380260003, + "learning_rate": 7.579758579456893e-05, + "loss": 1.1917, + "step": 381 + }, + { + "epoch": 0.7, + "grad_norm": 0.2390937887338632, + "learning_rate": 7.575946548016847e-05, + "loss": 1.2186, + "step": 382 + }, + { + "epoch": 0.7, + "grad_norm": 0.25131263043429275, + "learning_rate": 7.572118272477622e-05, + "loss": 1.2538, + "step": 383 + }, + { + "epoch": 0.7, + "grad_norm": 0.223974104870702, + "learning_rate": 7.568273770229546e-05, + "loss": 1.2165, + "step": 384 + }, + { + "epoch": 0.7, + "grad_norm": 0.25840356830252875, + "learning_rate": 7.564413058736663e-05, + "loss": 1.1848, + "step": 385 + }, + { + "epoch": 0.71, + "grad_norm": 0.2723156683076603, + "learning_rate": 7.560536155536641e-05, + "loss": 1.1982, + "step": 386 + }, + { + "epoch": 0.71, + "grad_norm": 0.265687427976889, + "learning_rate": 7.556643078240708e-05, + "loss": 1.231, + "step": 387 + }, + { + "epoch": 0.71, + "grad_norm": 0.25152762080976077, + "learning_rate": 7.552733844533562e-05, + "loss": 1.1974, + "step": 388 + }, + { + "epoch": 0.71, + "grad_norm": 0.2366049485053541, + "learning_rate": 7.548808472173292e-05, + "loss": 1.3119, + "step": 389 + }, + { + "epoch": 0.71, + "grad_norm": 0.22092196577077122, + "learning_rate": 7.5448669789913e-05, + "loss": 1.195, + "step": 390 + }, + { + "epoch": 0.72, + "grad_norm": 0.22667521540462374, + "learning_rate": 7.540909382892217e-05, + "loss": 1.1431, + "step": 391 + }, + { + "epoch": 0.72, + "grad_norm": 0.25432207282646513, + "learning_rate": 7.536935701853823e-05, + "loss": 1.2173, + "step": 392 + }, + { + "epoch": 0.72, + "grad_norm": 0.29950506457923864, + "learning_rate": 7.53294595392697e-05, + "loss": 1.1962, + "step": 393 + }, + { + "epoch": 0.72, + "grad_norm": 0.24735689607229913, + "learning_rate": 7.528940157235487e-05, + "loss": 1.2053, + "step": 394 + }, + { + "epoch": 0.72, + "grad_norm": 0.24394198607459663, + "learning_rate": 7.524918329976114e-05, + "loss": 1.1979, + "step": 395 + }, + { + "epoch": 0.72, + "grad_norm": 0.2630369372689188, + "learning_rate": 7.520880490418409e-05, + "loss": 1.2111, + "step": 396 + }, + { + "epoch": 0.73, + "grad_norm": 0.26275028416291457, + "learning_rate": 7.516826656904664e-05, + "loss": 1.2133, + "step": 397 + }, + { + "epoch": 0.73, + "grad_norm": 0.23938074620956928, + "learning_rate": 7.512756847849831e-05, + "loss": 1.1355, + "step": 398 + }, + { + "epoch": 0.73, + "grad_norm": 0.3724960610098138, + "learning_rate": 7.508671081741428e-05, + "loss": 1.2572, + "step": 399 + }, + { + "epoch": 0.73, + "grad_norm": 0.24161685847894723, + "learning_rate": 7.504569377139462e-05, + "loss": 1.1706, + "step": 400 + }, + { + "epoch": 0.73, + "grad_norm": 0.26121591322670523, + "learning_rate": 7.50045175267634e-05, + "loss": 1.2135, + "step": 401 + }, + { + "epoch": 0.74, + "grad_norm": 0.2465579498164775, + "learning_rate": 7.496318227056788e-05, + "loss": 1.1641, + "step": 402 + }, + { + "epoch": 0.74, + "grad_norm": 0.2556288696122787, + "learning_rate": 7.492168819057767e-05, + "loss": 1.2939, + "step": 403 + }, + { + "epoch": 0.74, + "grad_norm": 0.261481216336303, + "learning_rate": 7.488003547528382e-05, + "loss": 1.2026, + "step": 404 + }, + { + "epoch": 0.74, + "grad_norm": 0.2389415135676362, + "learning_rate": 7.483822431389799e-05, + "loss": 1.2131, + "step": 405 + }, + { + "epoch": 0.74, + "grad_norm": 0.2559201956627192, + "learning_rate": 7.479625489635162e-05, + "loss": 1.1246, + "step": 406 + }, + { + "epoch": 0.74, + "grad_norm": 0.27127932491822604, + "learning_rate": 7.475412741329504e-05, + "loss": 1.2429, + "step": 407 + }, + { + "epoch": 0.75, + "grad_norm": 0.27006004008695594, + "learning_rate": 7.47118420560966e-05, + "loss": 1.2388, + "step": 408 + }, + { + "epoch": 0.75, + "grad_norm": 0.23716823297200537, + "learning_rate": 7.466939901684182e-05, + "loss": 1.1264, + "step": 409 + }, + { + "epoch": 0.75, + "grad_norm": 0.2885373898669248, + "learning_rate": 7.462679848833252e-05, + "loss": 1.2786, + "step": 410 + }, + { + "epoch": 0.75, + "grad_norm": 0.49215227598639927, + "learning_rate": 7.458404066408588e-05, + "loss": 1.2386, + "step": 411 + }, + { + "epoch": 0.75, + "grad_norm": 0.24235735604947403, + "learning_rate": 7.454112573833368e-05, + "loss": 1.1423, + "step": 412 + }, + { + "epoch": 0.76, + "grad_norm": 0.2584614748054343, + "learning_rate": 7.449805390602127e-05, + "loss": 1.2669, + "step": 413 + }, + { + "epoch": 0.76, + "grad_norm": 0.23806123085998873, + "learning_rate": 7.445482536280684e-05, + "loss": 1.1763, + "step": 414 + }, + { + "epoch": 0.76, + "grad_norm": 0.24459517607786851, + "learning_rate": 7.441144030506043e-05, + "loss": 1.198, + "step": 415 + }, + { + "epoch": 0.76, + "grad_norm": 0.25801616402700395, + "learning_rate": 7.436789892986304e-05, + "loss": 1.2136, + "step": 416 + }, + { + "epoch": 0.76, + "grad_norm": 0.2814819942392514, + "learning_rate": 7.432420143500578e-05, + "loss": 1.2398, + "step": 417 + }, + { + "epoch": 0.76, + "grad_norm": 0.22134709322606153, + "learning_rate": 7.428034801898893e-05, + "loss": 1.1592, + "step": 418 + }, + { + "epoch": 0.77, + "grad_norm": 0.2899677536995633, + "learning_rate": 7.42363388810211e-05, + "loss": 1.2296, + "step": 419 + }, + { + "epoch": 0.77, + "grad_norm": 0.24005943230262294, + "learning_rate": 7.419217422101822e-05, + "loss": 1.2223, + "step": 420 + }, + { + "epoch": 0.77, + "grad_norm": 0.26417562369496167, + "learning_rate": 7.414785423960275e-05, + "loss": 1.2261, + "step": 421 + }, + { + "epoch": 0.77, + "grad_norm": 0.2580815883535521, + "learning_rate": 7.410337913810271e-05, + "loss": 1.2021, + "step": 422 + }, + { + "epoch": 0.77, + "grad_norm": 0.25242217589496435, + "learning_rate": 7.405874911855071e-05, + "loss": 1.239, + "step": 423 + }, + { + "epoch": 0.78, + "grad_norm": 0.21991733999839932, + "learning_rate": 7.401396438368315e-05, + "loss": 1.1716, + "step": 424 + }, + { + "epoch": 0.78, + "grad_norm": 0.40116538322720213, + "learning_rate": 7.396902513693924e-05, + "loss": 1.2773, + "step": 425 + }, + { + "epoch": 0.78, + "grad_norm": 0.277333939455099, + "learning_rate": 7.392393158246002e-05, + "loss": 1.2574, + "step": 426 + }, + { + "epoch": 0.78, + "grad_norm": 0.27146087746385755, + "learning_rate": 7.387868392508756e-05, + "loss": 1.2243, + "step": 427 + }, + { + "epoch": 0.78, + "grad_norm": 0.255881055620786, + "learning_rate": 7.38332823703639e-05, + "loss": 1.223, + "step": 428 + }, + { + "epoch": 0.78, + "grad_norm": 0.24807364856677255, + "learning_rate": 7.378772712453021e-05, + "loss": 1.1985, + "step": 429 + }, + { + "epoch": 0.79, + "grad_norm": 0.25746257617764423, + "learning_rate": 7.37420183945258e-05, + "loss": 1.2502, + "step": 430 + }, + { + "epoch": 0.79, + "grad_norm": 0.28851991049982234, + "learning_rate": 7.369615638798722e-05, + "loss": 1.2535, + "step": 431 + }, + { + "epoch": 0.79, + "grad_norm": 0.24113389811604363, + "learning_rate": 7.365014131324725e-05, + "loss": 1.2227, + "step": 432 + }, + { + "epoch": 0.79, + "grad_norm": 0.2414465151257969, + "learning_rate": 7.360397337933405e-05, + "loss": 1.1884, + "step": 433 + }, + { + "epoch": 0.79, + "grad_norm": 0.2735463134699831, + "learning_rate": 7.355765279597011e-05, + "loss": 1.2756, + "step": 434 + }, + { + "epoch": 0.8, + "grad_norm": 0.2588437452987293, + "learning_rate": 7.351117977357139e-05, + "loss": 1.2108, + "step": 435 + }, + { + "epoch": 0.8, + "grad_norm": 0.26573294117796553, + "learning_rate": 7.346455452324629e-05, + "loss": 1.1821, + "step": 436 + }, + { + "epoch": 0.8, + "grad_norm": 0.2555476577827304, + "learning_rate": 7.341777725679473e-05, + "loss": 1.1937, + "step": 437 + }, + { + "epoch": 0.8, + "grad_norm": 0.2867704132108098, + "learning_rate": 7.337084818670716e-05, + "loss": 1.2272, + "step": 438 + }, + { + "epoch": 0.8, + "grad_norm": 0.27726678115981157, + "learning_rate": 7.332376752616367e-05, + "loss": 1.2331, + "step": 439 + }, + { + "epoch": 0.8, + "grad_norm": 0.26955338021079955, + "learning_rate": 7.32765354890329e-05, + "loss": 1.1731, + "step": 440 + }, + { + "epoch": 0.81, + "grad_norm": 0.25250321202536524, + "learning_rate": 7.322915228987116e-05, + "loss": 1.2653, + "step": 441 + }, + { + "epoch": 0.81, + "grad_norm": 0.24748844179765395, + "learning_rate": 7.318161814392143e-05, + "loss": 1.24, + "step": 442 + }, + { + "epoch": 0.81, + "grad_norm": 0.28177805247356325, + "learning_rate": 7.313393326711239e-05, + "loss": 1.185, + "step": 443 + }, + { + "epoch": 0.81, + "grad_norm": 0.24093242000396312, + "learning_rate": 7.30860978760574e-05, + "loss": 1.1994, + "step": 444 + }, + { + "epoch": 0.81, + "grad_norm": 0.26277803901457075, + "learning_rate": 7.30381121880536e-05, + "loss": 1.212, + "step": 445 + }, + { + "epoch": 0.82, + "grad_norm": 0.2506524258682433, + "learning_rate": 7.298997642108079e-05, + "loss": 1.2421, + "step": 446 + }, + { + "epoch": 0.82, + "grad_norm": 0.2840599700015824, + "learning_rate": 7.294169079380061e-05, + "loss": 1.1818, + "step": 447 + }, + { + "epoch": 0.82, + "grad_norm": 0.24892184038117549, + "learning_rate": 7.289325552555538e-05, + "loss": 1.1916, + "step": 448 + }, + { + "epoch": 0.82, + "grad_norm": 0.2700898428541357, + "learning_rate": 7.284467083636722e-05, + "loss": 1.2517, + "step": 449 + }, + { + "epoch": 0.82, + "grad_norm": 0.2617848546539419, + "learning_rate": 7.279593694693698e-05, + "loss": 1.2063, + "step": 450 + }, + { + "epoch": 0.82, + "grad_norm": 0.2698278585334131, + "learning_rate": 7.274705407864332e-05, + "loss": 1.194, + "step": 451 + }, + { + "epoch": 0.83, + "grad_norm": 0.23678313024953834, + "learning_rate": 7.26980224535416e-05, + "loss": 1.2349, + "step": 452 + }, + { + "epoch": 0.83, + "grad_norm": 0.24851875792002978, + "learning_rate": 7.264884229436293e-05, + "loss": 1.1758, + "step": 453 + }, + { + "epoch": 0.83, + "grad_norm": 0.24122080121681125, + "learning_rate": 7.259951382451318e-05, + "loss": 1.1962, + "step": 454 + }, + { + "epoch": 0.83, + "grad_norm": 0.22741322959884405, + "learning_rate": 7.25500372680719e-05, + "loss": 1.1702, + "step": 455 + }, + { + "epoch": 0.83, + "grad_norm": 0.2297475610861458, + "learning_rate": 7.250041284979137e-05, + "loss": 1.1466, + "step": 456 + }, + { + "epoch": 0.84, + "grad_norm": 0.3057605989721467, + "learning_rate": 7.245064079509553e-05, + "loss": 1.246, + "step": 457 + }, + { + "epoch": 0.84, + "grad_norm": 0.2719638501597136, + "learning_rate": 7.240072133007899e-05, + "loss": 1.2184, + "step": 458 + }, + { + "epoch": 0.84, + "grad_norm": 0.2436807816414479, + "learning_rate": 7.235065468150593e-05, + "loss": 1.2324, + "step": 459 + }, + { + "epoch": 0.84, + "grad_norm": 0.23436349430255515, + "learning_rate": 7.23004410768092e-05, + "loss": 1.1813, + "step": 460 + }, + { + "epoch": 0.84, + "grad_norm": 0.2398940990211377, + "learning_rate": 7.22500807440892e-05, + "loss": 1.1924, + "step": 461 + }, + { + "epoch": 0.84, + "grad_norm": 0.2605716625062531, + "learning_rate": 7.219957391211281e-05, + "loss": 1.182, + "step": 462 + }, + { + "epoch": 0.85, + "grad_norm": 0.260462524570941, + "learning_rate": 7.214892081031244e-05, + "loss": 1.2136, + "step": 463 + }, + { + "epoch": 0.85, + "grad_norm": 0.21979766512306334, + "learning_rate": 7.209812166878491e-05, + "loss": 1.2066, + "step": 464 + }, + { + "epoch": 0.85, + "grad_norm": 0.23324453647530663, + "learning_rate": 7.204717671829051e-05, + "loss": 1.1657, + "step": 465 + }, + { + "epoch": 0.85, + "grad_norm": 0.2529434935507481, + "learning_rate": 7.199608619025177e-05, + "loss": 1.2093, + "step": 466 + }, + { + "epoch": 0.85, + "grad_norm": 0.25371701891720116, + "learning_rate": 7.194485031675265e-05, + "loss": 1.2225, + "step": 467 + }, + { + "epoch": 0.86, + "grad_norm": 0.23272423066292103, + "learning_rate": 7.189346933053725e-05, + "loss": 1.1721, + "step": 468 + }, + { + "epoch": 0.86, + "grad_norm": 0.25122928735587546, + "learning_rate": 7.184194346500892e-05, + "loss": 1.2537, + "step": 469 + }, + { + "epoch": 0.86, + "grad_norm": 0.2159270875490409, + "learning_rate": 7.179027295422913e-05, + "loss": 1.197, + "step": 470 + }, + { + "epoch": 0.86, + "grad_norm": 0.2633111059076544, + "learning_rate": 7.173845803291636e-05, + "loss": 1.1721, + "step": 471 + }, + { + "epoch": 0.86, + "grad_norm": 0.30555936322098703, + "learning_rate": 7.168649893644517e-05, + "loss": 1.3011, + "step": 472 + }, + { + "epoch": 0.87, + "grad_norm": 0.23492670111453726, + "learning_rate": 7.163439590084502e-05, + "loss": 1.1601, + "step": 473 + }, + { + "epoch": 0.87, + "grad_norm": 0.26602734263721806, + "learning_rate": 7.158214916279923e-05, + "loss": 1.2808, + "step": 474 + }, + { + "epoch": 0.87, + "grad_norm": 0.3182695007856262, + "learning_rate": 7.152975895964386e-05, + "loss": 1.2967, + "step": 475 + }, + { + "epoch": 0.87, + "grad_norm": 0.2785021674736721, + "learning_rate": 7.147722552936673e-05, + "loss": 1.1789, + "step": 476 + }, + { + "epoch": 0.87, + "grad_norm": 0.279474303138652, + "learning_rate": 7.142454911060627e-05, + "loss": 1.2596, + "step": 477 + }, + { + "epoch": 0.87, + "grad_norm": 0.2556980144910755, + "learning_rate": 7.137172994265044e-05, + "loss": 1.2426, + "step": 478 + }, + { + "epoch": 0.88, + "grad_norm": 0.3311256331993533, + "learning_rate": 7.131876826543565e-05, + "loss": 1.2059, + "step": 479 + }, + { + "epoch": 0.88, + "grad_norm": 0.26467296197775253, + "learning_rate": 7.12656643195457e-05, + "loss": 1.2482, + "step": 480 + }, + { + "epoch": 0.88, + "grad_norm": 0.27444885274652553, + "learning_rate": 7.121241834621064e-05, + "loss": 1.2528, + "step": 481 + }, + { + "epoch": 0.88, + "grad_norm": 0.2572283861115396, + "learning_rate": 7.115903058730567e-05, + "loss": 1.1849, + "step": 482 + }, + { + "epoch": 0.88, + "grad_norm": 0.2677065778235683, + "learning_rate": 7.11055012853501e-05, + "loss": 1.2011, + "step": 483 + }, + { + "epoch": 0.89, + "grad_norm": 0.29470622036742816, + "learning_rate": 7.105183068350619e-05, + "loss": 1.2398, + "step": 484 + }, + { + "epoch": 0.89, + "grad_norm": 0.27609230248969197, + "learning_rate": 7.099801902557811e-05, + "loss": 1.2259, + "step": 485 + }, + { + "epoch": 0.89, + "grad_norm": 0.24248634168099284, + "learning_rate": 7.094406655601073e-05, + "loss": 1.2282, + "step": 486 + }, + { + "epoch": 0.89, + "grad_norm": 0.2765941767688746, + "learning_rate": 7.088997351988865e-05, + "loss": 1.2319, + "step": 487 + }, + { + "epoch": 0.89, + "grad_norm": 0.29347776909858947, + "learning_rate": 7.083574016293493e-05, + "loss": 1.1765, + "step": 488 + }, + { + "epoch": 0.89, + "grad_norm": 0.285370295424537, + "learning_rate": 7.078136673151008e-05, + "loss": 1.26, + "step": 489 + }, + { + "epoch": 0.9, + "grad_norm": 0.29408734903836536, + "learning_rate": 7.072685347261093e-05, + "loss": 1.226, + "step": 490 + }, + { + "epoch": 0.9, + "grad_norm": 0.27437470239205813, + "learning_rate": 7.067220063386947e-05, + "loss": 1.1976, + "step": 491 + }, + { + "epoch": 0.9, + "grad_norm": 0.2680770258777871, + "learning_rate": 7.061740846355176e-05, + "loss": 1.1915, + "step": 492 + }, + { + "epoch": 0.9, + "grad_norm": 0.27200362879502954, + "learning_rate": 7.056247721055678e-05, + "loss": 1.2002, + "step": 493 + }, + { + "epoch": 0.9, + "grad_norm": 0.2637811092577037, + "learning_rate": 7.050740712441528e-05, + "loss": 1.287, + "step": 494 + }, + { + "epoch": 0.91, + "grad_norm": 0.24657959209271266, + "learning_rate": 7.045219845528875e-05, + "loss": 1.2284, + "step": 495 + }, + { + "epoch": 0.91, + "grad_norm": 0.25311992110358666, + "learning_rate": 7.039685145396812e-05, + "loss": 1.1616, + "step": 496 + }, + { + "epoch": 0.91, + "grad_norm": 0.2564633694193358, + "learning_rate": 7.034136637187275e-05, + "loss": 1.2067, + "step": 497 + }, + { + "epoch": 0.91, + "grad_norm": 0.2446797651174144, + "learning_rate": 7.028574346104926e-05, + "loss": 1.2284, + "step": 498 + }, + { + "epoch": 0.91, + "grad_norm": 0.2592751463399255, + "learning_rate": 7.022998297417034e-05, + "loss": 1.2371, + "step": 499 + }, + { + "epoch": 0.91, + "grad_norm": 0.2500713943206808, + "learning_rate": 7.017408516453365e-05, + "loss": 1.1061, + "step": 500 + }, + { + "epoch": 0.92, + "grad_norm": 0.2812266276040743, + "learning_rate": 7.011805028606064e-05, + "loss": 1.1949, + "step": 501 + }, + { + "epoch": 0.92, + "grad_norm": 0.298829667668083, + "learning_rate": 7.006187859329544e-05, + "loss": 1.2313, + "step": 502 + }, + { + "epoch": 0.92, + "grad_norm": 0.26518768159745104, + "learning_rate": 7.000557034140361e-05, + "loss": 1.2246, + "step": 503 + }, + { + "epoch": 0.92, + "grad_norm": 0.3037280360760458, + "learning_rate": 6.994912578617113e-05, + "loss": 1.1617, + "step": 504 + }, + { + "epoch": 0.92, + "grad_norm": 0.2726903109255714, + "learning_rate": 6.989254518400309e-05, + "loss": 1.2415, + "step": 505 + }, + { + "epoch": 0.93, + "grad_norm": 0.25568082003046966, + "learning_rate": 6.98358287919226e-05, + "loss": 1.1817, + "step": 506 + }, + { + "epoch": 0.93, + "grad_norm": 0.25633294893705044, + "learning_rate": 6.97789768675696e-05, + "loss": 1.2149, + "step": 507 + }, + { + "epoch": 0.93, + "grad_norm": 0.28291439435087123, + "learning_rate": 6.972198966919972e-05, + "loss": 1.1578, + "step": 508 + }, + { + "epoch": 0.93, + "grad_norm": 0.27195184756655516, + "learning_rate": 6.966486745568308e-05, + "loss": 1.2355, + "step": 509 + }, + { + "epoch": 0.93, + "grad_norm": 0.239159568376005, + "learning_rate": 6.960761048650312e-05, + "loss": 1.1688, + "step": 510 + }, + { + "epoch": 0.93, + "grad_norm": 0.22961475425949177, + "learning_rate": 6.955021902175543e-05, + "loss": 1.2094, + "step": 511 + }, + { + "epoch": 0.94, + "grad_norm": 0.27443773600741117, + "learning_rate": 6.949269332214651e-05, + "loss": 1.2559, + "step": 512 + }, + { + "epoch": 0.94, + "grad_norm": 0.26230551832002097, + "learning_rate": 6.94350336489927e-05, + "loss": 1.2121, + "step": 513 + }, + { + "epoch": 0.94, + "grad_norm": 0.2716742985303849, + "learning_rate": 6.937724026421892e-05, + "loss": 1.2444, + "step": 514 + }, + { + "epoch": 0.94, + "grad_norm": 0.2537850139439542, + "learning_rate": 6.931931343035742e-05, + "loss": 1.1327, + "step": 515 + }, + { + "epoch": 0.94, + "grad_norm": 0.28599587967496826, + "learning_rate": 6.926125341054676e-05, + "loss": 1.2236, + "step": 516 + }, + { + "epoch": 0.95, + "grad_norm": 0.26780654378470103, + "learning_rate": 6.920306046853043e-05, + "loss": 1.2295, + "step": 517 + }, + { + "epoch": 0.95, + "grad_norm": 0.23606296888412015, + "learning_rate": 6.914473486865577e-05, + "loss": 1.1543, + "step": 518 + }, + { + "epoch": 0.95, + "grad_norm": 0.34976881174240837, + "learning_rate": 6.90862768758727e-05, + "loss": 1.2067, + "step": 519 + }, + { + "epoch": 0.95, + "grad_norm": 0.2481257873494882, + "learning_rate": 6.902768675573258e-05, + "loss": 1.2188, + "step": 520 + }, + { + "epoch": 0.95, + "grad_norm": 0.2996395778117021, + "learning_rate": 6.896896477438699e-05, + "loss": 1.2326, + "step": 521 + }, + { + "epoch": 0.95, + "grad_norm": 0.8839768816333193, + "learning_rate": 6.891011119858643e-05, + "loss": 1.2435, + "step": 522 + }, + { + "epoch": 0.96, + "grad_norm": 0.2851882482058998, + "learning_rate": 6.885112629567927e-05, + "loss": 1.2644, + "step": 523 + }, + { + "epoch": 0.96, + "grad_norm": 0.2813663482913699, + "learning_rate": 6.879201033361035e-05, + "loss": 1.2309, + "step": 524 + }, + { + "epoch": 0.96, + "grad_norm": 0.3257551560135454, + "learning_rate": 6.873276358091996e-05, + "loss": 1.2755, + "step": 525 + }, + { + "epoch": 0.96, + "grad_norm": 0.28930479952494365, + "learning_rate": 6.867338630674247e-05, + "loss": 1.1962, + "step": 526 + }, + { + "epoch": 0.96, + "grad_norm": 0.3077462996938649, + "learning_rate": 6.861387878080511e-05, + "loss": 1.2402, + "step": 527 + }, + { + "epoch": 0.97, + "grad_norm": 0.2848900193452761, + "learning_rate": 6.855424127342688e-05, + "loss": 1.2748, + "step": 528 + }, + { + "epoch": 0.97, + "grad_norm": 0.4765938812802202, + "learning_rate": 6.849447405551718e-05, + "loss": 1.2226, + "step": 529 + }, + { + "epoch": 0.97, + "grad_norm": 0.53184473292579, + "learning_rate": 6.843457739857467e-05, + "loss": 1.2347, + "step": 530 + }, + { + "epoch": 0.97, + "grad_norm": 0.6416239346492343, + "learning_rate": 6.837455157468596e-05, + "loss": 1.2429, + "step": 531 + }, + { + "epoch": 0.97, + "grad_norm": 0.3188092712502773, + "learning_rate": 6.831439685652442e-05, + "loss": 1.216, + "step": 532 + }, + { + "epoch": 0.97, + "grad_norm": 0.3527495731006385, + "learning_rate": 6.825411351734895e-05, + "loss": 1.1682, + "step": 533 + }, + { + "epoch": 0.98, + "grad_norm": 0.29603753744741856, + "learning_rate": 6.819370183100274e-05, + "loss": 1.1434, + "step": 534 + }, + { + "epoch": 0.98, + "grad_norm": 0.5252450389976622, + "learning_rate": 6.813316207191198e-05, + "loss": 1.1943, + "step": 535 + }, + { + "epoch": 0.98, + "grad_norm": 0.32999419558659937, + "learning_rate": 6.807249451508466e-05, + "loss": 1.192, + "step": 536 + }, + { + "epoch": 0.98, + "grad_norm": 0.3650175469778724, + "learning_rate": 6.801169943610929e-05, + "loss": 1.2141, + "step": 537 + }, + { + "epoch": 0.98, + "grad_norm": 1.0643532150783557, + "learning_rate": 6.795077711115368e-05, + "loss": 1.2253, + "step": 538 + }, + { + "epoch": 0.99, + "grad_norm": 0.5041310609130145, + "learning_rate": 6.788972781696363e-05, + "loss": 1.278, + "step": 539 + }, + { + "epoch": 0.99, + "grad_norm": 0.5123058164360991, + "learning_rate": 6.782855183086177e-05, + "loss": 1.2231, + "step": 540 + }, + { + "epoch": 0.99, + "grad_norm": 0.3533015702394419, + "learning_rate": 6.776724943074619e-05, + "loss": 1.2072, + "step": 541 + }, + { + "epoch": 0.99, + "grad_norm": 0.30253964625417207, + "learning_rate": 6.770582089508927e-05, + "loss": 1.1382, + "step": 542 + }, + { + "epoch": 0.99, + "grad_norm": 0.348991618828202, + "learning_rate": 6.764426650293633e-05, + "loss": 1.2079, + "step": 543 + }, + { + "epoch": 0.99, + "grad_norm": 0.46017440578788743, + "learning_rate": 6.758258653390444e-05, + "loss": 1.1813, + "step": 544 + }, + { + "epoch": 1.0, + "grad_norm": 0.31962101755594885, + "learning_rate": 6.75207812681811e-05, + "loss": 1.1339, + "step": 545 + }, + { + "epoch": 1.0, + "grad_norm": 0.37092024548285923, + "learning_rate": 6.745885098652298e-05, + "loss": 1.2591, + "step": 546 + }, + { + "epoch": 1.0, + "grad_norm": 0.32347106450715835, + "learning_rate": 6.739679597025466e-05, + "loss": 1.2017, + "step": 547 + }, + { + "epoch": 1.0, + "grad_norm": 0.39250187112342494, + "learning_rate": 6.733461650126733e-05, + "loss": 1.0933, + "step": 548 + }, + { + "epoch": 1.0, + "grad_norm": 0.473522452217324, + "learning_rate": 6.727231286201752e-05, + "loss": 1.1124, + "step": 549 + }, + { + "epoch": 1.01, + "grad_norm": 0.4809062179622052, + "learning_rate": 6.720988533552582e-05, + "loss": 1.1585, + "step": 550 + } + ], + "logging_steps": 1.0, + "max_steps": 1638, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 50, + "total_flos": 570219722440704.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-550/training_args.bin b/checkpoint-550/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..c5d2416a3b70bb5260978ec9996f00154a724ba7 --- /dev/null +++ b/checkpoint-550/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b22e8f9d51a16d03a2c506fa3d1eafa8f4b1ae992992c2086a4d435ffd97387e +size 6712 diff --git a/checkpoint-550/zero_to_fp32.py b/checkpoint-550/zero_to_fp32.py new file mode 100755 index 0000000000000000000000000000000000000000..24cc342e78d1a006c782b3a4cd68d9ce786d8fd8 --- /dev/null +++ b/checkpoint-550/zero_to_fp32.py @@ -0,0 +1,604 @@ +#!/usr/bin/env python + +# Copyright (c) Microsoft Corporation. +# SPDX-License-Identifier: Apache-2.0 + +# DeepSpeed Team + +# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets +# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in +# the future. Once extracted, the weights don't require DeepSpeed and can be used in any +# application. +# +# example: python zero_to_fp32.py . pytorch_model.bin + +import argparse +import torch +import glob +import math +import os +import re +from collections import OrderedDict +from dataclasses import dataclass + +# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with +# DeepSpeed data structures it has to be available in the current python environment. +from deepspeed.utils import logger +from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS, + FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES, + FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS) + + +@dataclass +class zero_model_state: + buffers: dict() + param_shapes: dict() + shared_params: list + ds_version: int + frozen_param_shapes: dict() + frozen_param_fragments: dict() + + +debug = 0 + +# load to cpu +device = torch.device('cpu') + + +def atoi(text): + return int(text) if text.isdigit() else text + + +def natural_keys(text): + ''' + alist.sort(key=natural_keys) sorts in human order + http://nedbatchelder.com/blog/200712/human_sorting.html + (See Toothy's implementation in the comments) + ''' + return [atoi(c) for c in re.split(r'(\d+)', text)] + + +def get_model_state_file(checkpoint_dir, zero_stage): + if not os.path.isdir(checkpoint_dir): + raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist") + + # there should be only one file + if zero_stage <= 2: + file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt") + elif zero_stage == 3: + file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt") + + if not os.path.exists(file): + raise FileNotFoundError(f"can't find model states file at '{file}'") + + return file + + +def get_checkpoint_files(checkpoint_dir, glob_pattern): + # XXX: need to test that this simple glob rule works for multi-node setup too + ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys) + + if len(ckpt_files) == 0: + raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'") + + return ckpt_files + + +def get_optim_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt") + + +def get_model_state_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_model_states.pt") + + +def parse_model_states(files): + zero_model_states = [] + for file in files: + state_dict = torch.load(file, map_location=device) + + if BUFFER_NAMES not in state_dict: + raise ValueError(f"{file} is not a model state checkpoint") + buffer_names = state_dict[BUFFER_NAMES] + if debug: + print("Found buffers:", buffer_names) + + # recover just the buffers while restoring them to fp32 if they were saved in fp16 + buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names} + param_shapes = state_dict[PARAM_SHAPES] + + # collect parameters that are included in param_shapes + param_names = [] + for s in param_shapes: + for name in s.keys(): + param_names.append(name) + + # update with frozen parameters + frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None) + if frozen_param_shapes is not None: + if debug: + print(f"Found frozen_param_shapes: {frozen_param_shapes}") + param_names += list(frozen_param_shapes.keys()) + + # handle shared params + shared_params = [[k, v] for k, v in state_dict["shared_params"].items()] + + ds_version = state_dict.get(DS_VERSION, None) + + frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None) + + z_model_state = zero_model_state(buffers=buffers, + param_shapes=param_shapes, + shared_params=shared_params, + ds_version=ds_version, + frozen_param_shapes=frozen_param_shapes, + frozen_param_fragments=frozen_param_fragments) + zero_model_states.append(z_model_state) + + return zero_model_states + + +def parse_optim_states(files, ds_checkpoint_dir): + + total_files = len(files) + state_dicts = [] + for f in files: + state_dict = torch.load(f, map_location=device) + # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights + # and also handle the case where it was already removed by another helper script + state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None) + state_dicts.append(state_dict) + + if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]: + raise ValueError(f"{files[0]} is not a zero checkpoint") + zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE] + world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT] + + # For ZeRO-2 each param group can have different partition_count as data parallelism for expert + # parameters can be different from data parallelism for non-expert parameters. So we can just + # use the max of the partition_count to get the dp world_size. + + if type(world_size) is list: + world_size = max(world_size) + + if world_size != total_files: + raise ValueError( + f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. " + "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes." + ) + + # the groups are named differently in each stage + if zero_stage <= 2: + fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS + elif zero_stage == 3: + fp32_groups_key = FP32_FLAT_GROUPS + else: + raise ValueError(f"unknown zero stage {zero_stage}") + + if zero_stage <= 2: + fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))] + elif zero_stage == 3: + # if there is more than one param group, there will be multiple flattened tensors - one + # flattened tensor per group - for simplicity merge them into a single tensor + # + # XXX: could make the script more memory efficient for when there are multiple groups - it + # will require matching the sub-lists of param_shapes for each param group flattened tensor + + fp32_flat_groups = [ + torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts)) + ] + + return zero_stage, world_size, fp32_flat_groups + + +def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters): + """ + Returns fp32 state_dict reconstructed from ds checkpoint + + Args: + - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are) + + """ + print(f"Processing zero checkpoint '{ds_checkpoint_dir}'") + + optim_files = get_optim_files(ds_checkpoint_dir) + zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir) + print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}") + + model_files = get_model_state_files(ds_checkpoint_dir) + + zero_model_states = parse_model_states(model_files) + print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}') + + if zero_stage <= 2: + return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + elif zero_stage == 3: + return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + + +def _zero2_merge_frozen_params(state_dict, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + frozen_param_fragments = zero_model_states[0].frozen_param_fragments + + if debug: + num_elem = sum(s.numel() for s in frozen_param_shapes.values()) + print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in frozen_param_fragments.values()]) + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + state_dict[name] = frozen_param_fragments[name] + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _has_callable(obj, fn): + attr = getattr(obj, fn, None) + return callable(attr) + + +def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + + # Reconstruction protocol: + # + # XXX: document this + + if debug: + for i in range(world_size): + for j in range(len(fp32_flat_groups[0])): + print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}") + + # XXX: memory usage doubles here (zero2) + num_param_groups = len(fp32_flat_groups[0]) + merged_single_partition_of_fp32_groups = [] + for i in range(num_param_groups): + merged_partitions = [sd[i] for sd in fp32_flat_groups] + full_single_fp32_vector = torch.cat(merged_partitions, 0) + merged_single_partition_of_fp32_groups.append(full_single_fp32_vector) + avail_numel = sum( + [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups]) + + if debug: + wanted_params = sum([len(shapes) for shapes in param_shapes]) + wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes]) + # not asserting if there is a mismatch due to possible padding + print(f"Have {avail_numel} numels to process.") + print(f"Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + total_numel = 0 + total_params = 0 + for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups): + offset = 0 + avail_numel = full_single_fp32_vector.numel() + for name, shape in shapes.items(): + + unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape) + total_numel += unpartitioned_numel + total_params += 1 + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape) + offset += unpartitioned_numel + + # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and + # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex + # paddings performed in the code it's almost impossible to predict the exact numbers w/o the + # live optimizer object, so we are checking that the numbers are within the right range + align_to = 2 * world_size + + def zero2_align(x): + return align_to * math.ceil(x / align_to) + + if debug: + print(f"original offset={offset}, avail_numel={avail_numel}") + + offset = zero2_align(offset) + avail_numel = zero2_align(avail_numel) + + if debug: + print(f"aligned offset={offset}, avail_numel={avail_numel}") + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero2_merge_frozen_params(state_dict, zero_model_states) + + _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def zero3_partitioned_param_info(unpartitioned_numel, world_size): + remainder = unpartitioned_numel % world_size + padding_numel = (world_size - remainder) if remainder else 0 + partitioned_numel = math.ceil(unpartitioned_numel / world_size) + return partitioned_numel, padding_numel + + +def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + if debug: + for i in range(world_size): + num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values()) + print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in zero_model_states[0].frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states) + state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape) + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + avail_numel = fp32_flat_groups[0].numel() * world_size + # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each + # param, re-consolidating each param, while dealing with padding if any + + # merge list of dicts, preserving order + param_shapes = {k: v for d in param_shapes for k, v in d.items()} + + if debug: + for i in range(world_size): + print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}") + + wanted_params = len(param_shapes) + wanted_numel = sum(shape.numel() for shape in param_shapes.values()) + # not asserting if there is a mismatch due to possible padding + avail_numel = fp32_flat_groups[0].numel() * world_size + print(f"Trainable params: Have {avail_numel} numels to process.") + print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + offset = 0 + total_numel = 0 + total_params = 0 + for name, shape in param_shapes.items(): + + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + total_params += 1 + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + # XXX: memory usage doubles here + state_dict[name] = torch.cat( + tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)), + 0).narrow(0, 0, unpartitioned_numel).view(shape) + offset += partitioned_numel + + offset *= world_size + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero3_merge_frozen_params(state_dict, world_size, zero_model_states) + + _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with + ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example + via a model hub. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + + Returns: + - pytorch ``state_dict`` + + Note: this approach may not work if your application doesn't have sufficient free CPU memory and + you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with + the checkpoint. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + # do the training and checkpoint saving + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu + model = model.cpu() # move to cpu + model.load_state_dict(state_dict) + # submit to model hub or save the model to share with others + + In this example the ``model`` will no longer be usable in the deepspeed context of the same + application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead. + + """ + if tag is None: + latest_path = os.path.join(checkpoint_dir, 'latest') + if os.path.isfile(latest_path): + with open(latest_path, 'r') as fd: + tag = fd.read().strip() + else: + raise ValueError(f"Unable to find 'latest' file at {latest_path}") + + ds_checkpoint_dir = os.path.join(checkpoint_dir, tag) + + if not os.path.isdir(ds_checkpoint_dir): + raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist") + + return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters) + + +def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be + loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + """ + + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters) + print(f"Saving fp32 state dict to {output_file}") + torch.save(state_dict, output_file) + + +def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None): + """ + 1. Put the provided model to cpu + 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` + 3. Load it into the provided model + + Args: + - ``model``: the model object to update + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + + Returns: + - ``model`: modified model + + Make sure you have plenty of CPU memory available before you call this function. If you don't + have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it + conveniently placed for you in the checkpoint folder. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint + model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir) + # submit to model hub or save the model to share with others + + Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context + of the same application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + """ + logger.info(f"Extracting fp32 weights") + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag) + + logger.info(f"Overwriting model with fp32 weights") + model = model.cpu() + model.load_state_dict(state_dict, strict=False) + + return model + + +if __name__ == "__main__": + + parser = argparse.ArgumentParser() + parser.add_argument("checkpoint_dir", + type=str, + help="path to the desired checkpoint folder, e.g., path/checkpoint-12") + parser.add_argument( + "output_file", + type=str, + help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)") + parser.add_argument("-t", + "--tag", + type=str, + default=None, + help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1") + parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters") + parser.add_argument("-d", "--debug", action='store_true', help="enable debug") + args = parser.parse_args() + + debug = args.debug + + convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, + args.output_file, + tag=args.tag, + exclude_frozen_parameters=args.exclude_frozen_parameters) diff --git a/checkpoint-600/README.md b/checkpoint-600/README.md new file mode 100644 index 0000000000000000000000000000000000000000..16b1eacdd9353dec380a08ee77ce6ed5ab50f12e --- /dev/null +++ b/checkpoint-600/README.md @@ -0,0 +1,202 @@ +--- +library_name: peft +base_model: gotzmann/uni +--- + +# Model Card for Model ID + + + + + +## Model Details + +### Model Description + + + + + +- **Developed by:** [More Information Needed] +- **Funded by [optional]:** [More Information Needed] +- **Shared by [optional]:** [More Information Needed] +- **Model type:** [More Information Needed] +- **Language(s) (NLP):** [More Information Needed] +- **License:** [More Information Needed] +- **Finetuned from model [optional]:** [More Information Needed] + +### Model Sources [optional] + + + +- **Repository:** [More Information Needed] +- **Paper [optional]:** [More Information Needed] +- **Demo [optional]:** [More Information Needed] + +## Uses + + + +### Direct Use + + + +[More Information Needed] + +### Downstream Use [optional] + + + +[More Information Needed] + +### Out-of-Scope Use + + + +[More Information Needed] + +## Bias, Risks, and Limitations + + + +[More Information Needed] + +### Recommendations + + + +Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations. + +## How to Get Started with the Model + +Use the code below to get started with the model. + +[More Information Needed] + +## Training Details + +### Training Data + + + +[More Information Needed] + +### Training Procedure + + + +#### Preprocessing [optional] + +[More Information Needed] + + +#### Training Hyperparameters + +- **Training regime:** [More Information Needed] + +#### Speeds, Sizes, Times [optional] + + + +[More Information Needed] + +## Evaluation + + + +### Testing Data, Factors & Metrics + +#### Testing Data + + + +[More Information Needed] + +#### Factors + + + +[More Information Needed] + +#### Metrics + + + +[More Information Needed] + +### Results + +[More Information Needed] + +#### Summary + + + +## Model Examination [optional] + + + +[More Information Needed] + +## Environmental Impact + + + +Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700). + +- **Hardware Type:** [More Information Needed] +- **Hours used:** [More Information Needed] +- **Cloud Provider:** [More Information Needed] +- **Compute Region:** [More Information Needed] +- **Carbon Emitted:** [More Information Needed] + +## Technical Specifications [optional] + +### Model Architecture and Objective + +[More Information Needed] + +### Compute Infrastructure + +[More Information Needed] + +#### Hardware + +[More Information Needed] + +#### Software + +[More Information Needed] + +## Citation [optional] + + + +**BibTeX:** + +[More Information Needed] + +**APA:** + +[More Information Needed] + +## Glossary [optional] + + + +[More Information Needed] + +## More Information [optional] + +[More Information Needed] + +## Model Card Authors [optional] + +[More Information Needed] + +## Model Card Contact + +[More Information Needed] +### Framework versions + +- PEFT 0.10.0 \ No newline at end of file diff --git a/checkpoint-600/adapter_config.json b/checkpoint-600/adapter_config.json new file mode 100644 index 0000000000000000000000000000000000000000..3cd6dba5d79f7ca21fd4ad465cbbcac1e0960476 --- /dev/null +++ b/checkpoint-600/adapter_config.json @@ -0,0 +1,31 @@ +{ + "alpha_pattern": {}, + "auto_mapping": null, + "base_model_name_or_path": "gotzmann/uni", + "bias": "none", + "fan_in_fan_out": false, + "inference_mode": true, + "init_lora_weights": true, + "layer_replication": null, + "layers_pattern": null, + "layers_to_transform": null, + "loftq_config": {}, + "lora_alpha": 128, + "lora_dropout": 0.0, + "megatron_config": null, + "megatron_core": "megatron.core", + "modules_to_save": null, + "peft_type": "LORA", + "r": 128, + "rank_pattern": {}, + "revision": null, + "target_modules": [ + "v_proj", + "k_proj", + "q_proj", + "o_proj" + ], + "task_type": "CAUSAL_LM", + "use_dora": false, + "use_rslora": true +} \ No newline at end of file diff --git a/checkpoint-600/adapter_model.safetensors b/checkpoint-600/adapter_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..00ef5b34c7f75ce63bd1ab43eba0fb951610c198 --- /dev/null +++ b/checkpoint-600/adapter_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cb297e9519efcd7ab54e95cd64b61847178435f7c7a6642c0652f43ae35aa26c +size 1048664848 diff --git a/checkpoint-600/global_step600/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt b/checkpoint-600/global_step600/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..7023dc633db2a0cfae7a2d0ca0db132574a681c1 --- /dev/null +++ b/checkpoint-600/global_step600/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:daeed6817a52847b6785965455a6a15eea57c4c3f7b5c1d035202bd339ac01aa +size 787270042 diff --git a/checkpoint-600/global_step600/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt b/checkpoint-600/global_step600/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..56397218fc74e5af1abfd22837de5c8130c21785 --- /dev/null +++ b/checkpoint-600/global_step600/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:085252fbe925832c9a5c5baa1d5c2a225a822143477e003587bb8d6d593445da +size 787270042 diff --git a/checkpoint-600/global_step600/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt b/checkpoint-600/global_step600/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..72584de54afdf8e52b6be39c84c9771d8e81cdf9 --- /dev/null +++ b/checkpoint-600/global_step600/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8c1974b5b430b34e0c4964eff3ebf140668ec51f5693994ddc330df75c29ada0 +size 787270042 diff --git a/checkpoint-600/global_step600/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt b/checkpoint-600/global_step600/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..abec2f2b3b33d3b329d16cdb92ff651fd9301363 --- /dev/null +++ b/checkpoint-600/global_step600/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8f4be8a150d37f7afb036dded8b756296292589581e3bcdbb6d80e2e43992cce +size 787270042 diff --git a/checkpoint-600/global_step600/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt b/checkpoint-600/global_step600/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..cb3f8c58437980f0fcd84aa5ec73104c4920acbd --- /dev/null +++ b/checkpoint-600/global_step600/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7849b2d8ab9ef87c7b1c3b693e9aba355550030bbbb6fb2b2b9a36bcfe23073f +size 787270042 diff --git a/checkpoint-600/global_step600/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt b/checkpoint-600/global_step600/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..1a76ccf8db17da6de0e2365e6ee5d1052bb52b8b --- /dev/null +++ b/checkpoint-600/global_step600/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:ec1659e11ad3c6a1cfc6acdad2e71d862ede7a846bf76295aef5cb7ea7722af7 +size 787270042 diff --git a/checkpoint-600/global_step600/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt b/checkpoint-600/global_step600/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..119bc348311672b4625013ca7b81510c8b482101 --- /dev/null +++ b/checkpoint-600/global_step600/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:10ff0aeba9ef61ecf988f2448c881a742db7dde7b78faa460dfd424a5549fb3a +size 787270042 diff --git a/checkpoint-600/global_step600/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt b/checkpoint-600/global_step600/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..291253731f84c4c7cf1b34a9297441f87fd417d8 --- /dev/null +++ b/checkpoint-600/global_step600/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9fe648c37b5e0dab4a6387a25b61ce0fbf1ce696a3f8fdbbf6004fe1dedd8b15 +size 787270042 diff --git a/checkpoint-600/global_step600/zero_pp_rank_0_mp_rank_00_model_states.pt b/checkpoint-600/global_step600/zero_pp_rank_0_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..b0700bada730a121c64947397b7fd736206a55bd --- /dev/null +++ b/checkpoint-600/global_step600/zero_pp_rank_0_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ade19297ad74c4436df14f6ff0bbbdf6b3b6e8f902768d5a6ec8fdbb9ca7ddf6 +size 653742 diff --git a/checkpoint-600/global_step600/zero_pp_rank_1_mp_rank_00_model_states.pt b/checkpoint-600/global_step600/zero_pp_rank_1_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..8becc70b7945368127b629c1de0adb0a65a0cc12 --- /dev/null +++ b/checkpoint-600/global_step600/zero_pp_rank_1_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:94f74c568df86e5e544de53354ad2f324e1a6b8137e4be21d31c40fb1ed3e37c +size 653742 diff --git a/checkpoint-600/global_step600/zero_pp_rank_2_mp_rank_00_model_states.pt b/checkpoint-600/global_step600/zero_pp_rank_2_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..65e978d97fffa7f7d694ffd688f0717ba5f76ecb --- /dev/null +++ b/checkpoint-600/global_step600/zero_pp_rank_2_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7cc460d9a4ee1728057abe7d0b24cc1916fc623a84e2bd17492f81de07e2a604 +size 653742 diff --git a/checkpoint-600/global_step600/zero_pp_rank_3_mp_rank_00_model_states.pt b/checkpoint-600/global_step600/zero_pp_rank_3_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..efda9fd64665fe84134fe01aaf555b078df6cff0 --- /dev/null +++ b/checkpoint-600/global_step600/zero_pp_rank_3_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7112e5a6d3e6832046289f5a5e7a568fe717e312eda184c2d837559711f8a0b3 +size 653742 diff --git a/checkpoint-600/global_step600/zero_pp_rank_4_mp_rank_00_model_states.pt b/checkpoint-600/global_step600/zero_pp_rank_4_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..967831a68e7b2b25beb65d0dc585f50ed5534b83 --- /dev/null +++ b/checkpoint-600/global_step600/zero_pp_rank_4_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:85a6aeeb1ee3bdb8e84c808249460b9dcbd4cb755be163aa3e1864825cf67ee2 +size 653742 diff --git a/checkpoint-600/global_step600/zero_pp_rank_5_mp_rank_00_model_states.pt b/checkpoint-600/global_step600/zero_pp_rank_5_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..7e098d9cd120716f3ddd0e1f993a0370e3227ad3 --- /dev/null +++ b/checkpoint-600/global_step600/zero_pp_rank_5_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:72c5926bc2ef060a8067a89a75d566967f25eb34a4f1d2c603701b9bdb27b7a8 +size 653742 diff --git a/checkpoint-600/global_step600/zero_pp_rank_6_mp_rank_00_model_states.pt b/checkpoint-600/global_step600/zero_pp_rank_6_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..41a1ede5829046a80cf400c7a3e546bb60700045 --- /dev/null +++ b/checkpoint-600/global_step600/zero_pp_rank_6_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2aad8cf76a774f0a37ab26f6a2c0513015b27658e3cef347da37dbcc6d56ad6f +size 653742 diff --git a/checkpoint-600/global_step600/zero_pp_rank_7_mp_rank_00_model_states.pt b/checkpoint-600/global_step600/zero_pp_rank_7_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..4140915f67888daf6d80b922b6ee5d8f4ace4f44 --- /dev/null +++ b/checkpoint-600/global_step600/zero_pp_rank_7_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a2efd459e97a86e45a83990e4707bc026470c35737b755d2ea70551b0cbdea6c +size 653742 diff --git a/checkpoint-600/latest b/checkpoint-600/latest new file mode 100644 index 0000000000000000000000000000000000000000..12cae1adf3af8546b4141c6f62261c8e99839a54 --- /dev/null +++ b/checkpoint-600/latest @@ -0,0 +1 @@ +global_step600 \ No newline at end of file diff --git a/checkpoint-600/rng_state_0.pth b/checkpoint-600/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..4e5b7e2ec90fdb824c8932464c1d9068330655a7 --- /dev/null +++ b/checkpoint-600/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:36d2a2034ebb05cb71c510897f2795b31164e50f17b270bc25d2be3ad9a17b22 +size 15984 diff --git a/checkpoint-600/rng_state_1.pth b/checkpoint-600/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..7d8d7722fc72cab6d492b76cb99c8177dcc47544 --- /dev/null +++ b/checkpoint-600/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:060dfdb1c49102cbdc8868a6031e68787601b4ccd782f3fb9b137e20c1fd2c7a +size 15984 diff --git a/checkpoint-600/rng_state_2.pth b/checkpoint-600/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..3c9f84eff30cfa9ea1feedaf262d61fb12e4cba7 --- /dev/null +++ b/checkpoint-600/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:af01895cb66e616591f2e4baa8dcd8151530eab133c73571ccb31c74f35422ce +size 15984 diff --git a/checkpoint-600/rng_state_3.pth b/checkpoint-600/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..6eebfb928f8e91eff0ea1645a20b5aa4465c705b --- /dev/null +++ b/checkpoint-600/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:677921992b1e0cef3aee776f245975003d22f51d9bd6ed20f248ded1deb72fa9 +size 15984 diff --git a/checkpoint-600/rng_state_4.pth b/checkpoint-600/rng_state_4.pth new file mode 100644 index 0000000000000000000000000000000000000000..0866030a266c6d003cc378a9418a723f69e8ab99 --- /dev/null +++ b/checkpoint-600/rng_state_4.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d69353c629541c690c5471f8ec05fdab2bfecf3d37afaa436bc45939da6db68f +size 15984 diff --git a/checkpoint-600/rng_state_5.pth b/checkpoint-600/rng_state_5.pth new file mode 100644 index 0000000000000000000000000000000000000000..554638d77107f832d7aa51c61645ee2d6c48a36d --- /dev/null +++ b/checkpoint-600/rng_state_5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8e40ba6668cc03c9162c68a933d164bf38ae2d196a9a6fec03ae615491201185 +size 15984 diff --git a/checkpoint-600/rng_state_6.pth b/checkpoint-600/rng_state_6.pth new file mode 100644 index 0000000000000000000000000000000000000000..964331b65172a1bcac03e4673415fa787f724268 --- /dev/null +++ b/checkpoint-600/rng_state_6.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:870968fea834e24b2e099cf3e4fe1e3fb8caf38d8f8e5b790d7d47386d4d05f5 +size 15984 diff --git a/checkpoint-600/rng_state_7.pth b/checkpoint-600/rng_state_7.pth new file mode 100644 index 0000000000000000000000000000000000000000..cd4754d65217d0f9d1f2d3334397df7a8a079652 --- /dev/null +++ b/checkpoint-600/rng_state_7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e9e19618bee7c6ef43256fea25abe19bca88535eb1e7dc213cde8929ae4e8180 +size 15984 diff --git a/checkpoint-600/scheduler.pt b/checkpoint-600/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..5e77672ea908f9d30363be50ef230174a2d8afc1 --- /dev/null +++ b/checkpoint-600/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:89fa8d447eeca80a7f7254134cf50626f0ba8da3de27a47f21151e30f33f4960 +size 1064 diff --git a/checkpoint-600/special_tokens_map.json b/checkpoint-600/special_tokens_map.json new file mode 100644 index 0000000000000000000000000000000000000000..72ecfeeb7e14d244c936169d2ed139eeae235ef1 --- /dev/null +++ b/checkpoint-600/special_tokens_map.json @@ -0,0 +1,24 @@ +{ + "bos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "eos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "pad_token": "", + "unk_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + } +} diff --git a/checkpoint-600/tokenizer.model b/checkpoint-600/tokenizer.model new file mode 100644 index 0000000000000000000000000000000000000000..6c00c742ce03c627d6cd5b795984876fa49fa899 --- /dev/null +++ b/checkpoint-600/tokenizer.model @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347 +size 499723 diff --git a/checkpoint-600/tokenizer_config.json b/checkpoint-600/tokenizer_config.json new file mode 100644 index 0000000000000000000000000000000000000000..bb5a9f09d8c0f3c32c66fc6118fe5c76c5c6fd90 --- /dev/null +++ b/checkpoint-600/tokenizer_config.json @@ -0,0 +1,45 @@ +{ + "add_bos_token": true, + "add_eos_token": false, + "add_prefix_space": false, + "added_tokens_decoder": { + "0": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "1": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "2": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + } + }, + "bos_token": "", + "chat_template": "{% if messages[0]['role'] == 'system' %}{% set system_message = messages[0]['content'] %}{% endif %}{% if system_message is defined %}{{ '' + '### System:\\n\\n' + system_message }}{% endif %}{% for message in messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ '\\n\\n### Human:\\n\\n' + content }}{% elif message['role'] == 'assistant' %}{{ '\\n\\n### Assistant:\\n\\n' + content + '' }}{% endif %}{% endfor %}", + "clean_up_tokenization_spaces": false, + "eos_token": "", + "legacy": false, + "model_max_length": 1000000000000000019884624838656, + "pad_token": "", + "padding_side": "right", + "sp_model_kwargs": {}, + "spaces_between_special_tokens": false, + "split_special_tokens": false, + "tokenizer_class": "LlamaTokenizer", + "unk_token": "", + "use_default_system_prompt": false +} diff --git a/checkpoint-600/trainer_state.json b/checkpoint-600/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..fc81282be29d70d8028b799bd2261064cee77c34 --- /dev/null +++ b/checkpoint-600/trainer_state.json @@ -0,0 +1,4221 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 1.0973936899862826, + "eval_steps": 500, + "global_step": 600, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "grad_norm": 0.849355824164473, + "learning_rate": 4.878048780487805e-07, + "loss": 1.3655, + "step": 1 + }, + { + "epoch": 0.0, + "grad_norm": 10.01567518957158, + "learning_rate": 9.75609756097561e-07, + "loss": 1.5767, + "step": 2 + }, + { + "epoch": 0.01, + "grad_norm": 0.6466000875559635, + "learning_rate": 1.4634146341463414e-06, + "loss": 1.3913, + "step": 3 + }, + { + "epoch": 0.01, + "grad_norm": 0.6644565932010504, + "learning_rate": 1.951219512195122e-06, + "loss": 1.3218, + "step": 4 + }, + { + "epoch": 0.01, + "grad_norm": 0.571354207588475, + "learning_rate": 2.4390243902439027e-06, + "loss": 1.3597, + "step": 5 + }, + { + "epoch": 0.01, + "grad_norm": 0.31036262839244955, + "learning_rate": 2.926829268292683e-06, + "loss": 1.2832, + "step": 6 + }, + { + "epoch": 0.01, + "grad_norm": 0.2622135027188184, + "learning_rate": 3.414634146341464e-06, + "loss": 1.2161, + "step": 7 + }, + { + "epoch": 0.01, + "grad_norm": 0.296824630261661, + "learning_rate": 3.902439024390244e-06, + "loss": 1.2985, + "step": 8 + }, + { + "epoch": 0.02, + "grad_norm": 0.2557267467361569, + "learning_rate": 4.390243902439025e-06, + "loss": 1.3175, + "step": 9 + }, + { + "epoch": 0.02, + "grad_norm": 0.23418939513890769, + "learning_rate": 4.8780487804878055e-06, + "loss": 1.2617, + "step": 10 + }, + { + "epoch": 0.02, + "grad_norm": 0.2364760983285843, + "learning_rate": 5.365853658536586e-06, + "loss": 1.3103, + "step": 11 + }, + { + "epoch": 0.02, + "grad_norm": 0.23893034721889, + "learning_rate": 5.853658536585366e-06, + "loss": 1.2405, + "step": 12 + }, + { + "epoch": 0.02, + "grad_norm": 0.25563593295485887, + "learning_rate": 6.341463414634147e-06, + "loss": 1.2831, + "step": 13 + }, + { + "epoch": 0.03, + "grad_norm": 0.23239975352661665, + "learning_rate": 6.829268292682928e-06, + "loss": 1.3125, + "step": 14 + }, + { + "epoch": 0.03, + "grad_norm": 0.3092813858209507, + "learning_rate": 7.317073170731707e-06, + "loss": 1.2422, + "step": 15 + }, + { + "epoch": 0.03, + "grad_norm": 0.282563380367434, + "learning_rate": 7.804878048780489e-06, + "loss": 1.2453, + "step": 16 + }, + { + "epoch": 0.03, + "grad_norm": 0.22065680088315018, + "learning_rate": 8.292682926829268e-06, + "loss": 1.2491, + "step": 17 + }, + { + "epoch": 0.03, + "grad_norm": 0.22777800877980184, + "learning_rate": 8.78048780487805e-06, + "loss": 1.2655, + "step": 18 + }, + { + "epoch": 0.03, + "grad_norm": 0.22145212540177928, + "learning_rate": 9.268292682926831e-06, + "loss": 1.2413, + "step": 19 + }, + { + "epoch": 0.04, + "grad_norm": 0.22482351883112714, + "learning_rate": 9.756097560975611e-06, + "loss": 1.2653, + "step": 20 + }, + { + "epoch": 0.04, + "grad_norm": 0.20823080508385733, + "learning_rate": 1.024390243902439e-05, + "loss": 1.2374, + "step": 21 + }, + { + "epoch": 0.04, + "grad_norm": 0.26025492562935737, + "learning_rate": 1.0731707317073172e-05, + "loss": 1.2065, + "step": 22 + }, + { + "epoch": 0.04, + "grad_norm": 0.2150252124176173, + "learning_rate": 1.1219512195121953e-05, + "loss": 1.2782, + "step": 23 + }, + { + "epoch": 0.04, + "grad_norm": 0.2505915177425618, + "learning_rate": 1.1707317073170731e-05, + "loss": 1.2742, + "step": 24 + }, + { + "epoch": 0.05, + "grad_norm": 0.20129223044786942, + "learning_rate": 1.2195121951219513e-05, + "loss": 1.3366, + "step": 25 + }, + { + "epoch": 0.05, + "grad_norm": 0.1973508510397107, + "learning_rate": 1.2682926829268294e-05, + "loss": 1.2476, + "step": 26 + }, + { + "epoch": 0.05, + "grad_norm": 0.27103325392437194, + "learning_rate": 1.3170731707317076e-05, + "loss": 1.2325, + "step": 27 + }, + { + "epoch": 0.05, + "grad_norm": 0.17954976411006285, + "learning_rate": 1.3658536585365855e-05, + "loss": 1.2523, + "step": 28 + }, + { + "epoch": 0.05, + "grad_norm": 0.22216997851088888, + "learning_rate": 1.4146341463414635e-05, + "loss": 1.3297, + "step": 29 + }, + { + "epoch": 0.05, + "grad_norm": 0.2071458864548587, + "learning_rate": 1.4634146341463415e-05, + "loss": 1.2127, + "step": 30 + }, + { + "epoch": 0.06, + "grad_norm": 0.18039422081622164, + "learning_rate": 1.5121951219512196e-05, + "loss": 1.2509, + "step": 31 + }, + { + "epoch": 0.06, + "grad_norm": 0.18631254372974412, + "learning_rate": 1.5609756097560978e-05, + "loss": 1.2247, + "step": 32 + }, + { + "epoch": 0.06, + "grad_norm": 0.18843872523649827, + "learning_rate": 1.6097560975609757e-05, + "loss": 1.195, + "step": 33 + }, + { + "epoch": 0.06, + "grad_norm": 0.2163847267778325, + "learning_rate": 1.6585365853658537e-05, + "loss": 1.2179, + "step": 34 + }, + { + "epoch": 0.06, + "grad_norm": 0.19687688475496104, + "learning_rate": 1.7073170731707317e-05, + "loss": 1.2763, + "step": 35 + }, + { + "epoch": 0.07, + "grad_norm": 0.20409643064887947, + "learning_rate": 1.75609756097561e-05, + "loss": 1.253, + "step": 36 + }, + { + "epoch": 0.07, + "grad_norm": 0.1879182661759335, + "learning_rate": 1.804878048780488e-05, + "loss": 1.2586, + "step": 37 + }, + { + "epoch": 0.07, + "grad_norm": 0.19400648948514373, + "learning_rate": 1.8536585365853663e-05, + "loss": 1.2154, + "step": 38 + }, + { + "epoch": 0.07, + "grad_norm": 0.1878879343148452, + "learning_rate": 1.902439024390244e-05, + "loss": 1.2304, + "step": 39 + }, + { + "epoch": 0.07, + "grad_norm": 0.17687475469924052, + "learning_rate": 1.9512195121951222e-05, + "loss": 1.2351, + "step": 40 + }, + { + "epoch": 0.07, + "grad_norm": 0.18223935625384885, + "learning_rate": 2e-05, + "loss": 1.2222, + "step": 41 + }, + { + "epoch": 0.08, + "grad_norm": 0.1943061629408338, + "learning_rate": 2.048780487804878e-05, + "loss": 1.2044, + "step": 42 + }, + { + "epoch": 0.08, + "grad_norm": 0.17027514338700078, + "learning_rate": 2.0975609756097564e-05, + "loss": 1.1548, + "step": 43 + }, + { + "epoch": 0.08, + "grad_norm": 0.18553769630586192, + "learning_rate": 2.1463414634146344e-05, + "loss": 1.2721, + "step": 44 + }, + { + "epoch": 0.08, + "grad_norm": 0.19732826914228765, + "learning_rate": 2.1951219512195124e-05, + "loss": 1.3097, + "step": 45 + }, + { + "epoch": 0.08, + "grad_norm": 0.18714230986631472, + "learning_rate": 2.2439024390243907e-05, + "loss": 1.2662, + "step": 46 + }, + { + "epoch": 0.09, + "grad_norm": 0.19988987568002223, + "learning_rate": 2.2926829268292683e-05, + "loss": 1.2904, + "step": 47 + }, + { + "epoch": 0.09, + "grad_norm": 0.17744650133390918, + "learning_rate": 2.3414634146341463e-05, + "loss": 1.1825, + "step": 48 + }, + { + "epoch": 0.09, + "grad_norm": 0.16576734763834533, + "learning_rate": 2.3902439024390246e-05, + "loss": 1.1858, + "step": 49 + }, + { + "epoch": 0.09, + "grad_norm": 0.179591794065527, + "learning_rate": 2.4390243902439026e-05, + "loss": 1.2711, + "step": 50 + }, + { + "epoch": 0.09, + "grad_norm": 0.17923464471176911, + "learning_rate": 2.4878048780487805e-05, + "loss": 1.2289, + "step": 51 + }, + { + "epoch": 0.1, + "grad_norm": 0.18991742907836837, + "learning_rate": 2.536585365853659e-05, + "loss": 1.3097, + "step": 52 + }, + { + "epoch": 0.1, + "grad_norm": 0.19849796137254636, + "learning_rate": 2.5853658536585368e-05, + "loss": 1.2489, + "step": 53 + }, + { + "epoch": 0.1, + "grad_norm": 0.17452371110976383, + "learning_rate": 2.634146341463415e-05, + "loss": 1.2461, + "step": 54 + }, + { + "epoch": 0.1, + "grad_norm": 0.17671022353085036, + "learning_rate": 2.682926829268293e-05, + "loss": 1.153, + "step": 55 + }, + { + "epoch": 0.1, + "grad_norm": 0.36820559192096686, + "learning_rate": 2.731707317073171e-05, + "loss": 1.2431, + "step": 56 + }, + { + "epoch": 0.1, + "grad_norm": 0.20331468526494198, + "learning_rate": 2.7804878048780487e-05, + "loss": 1.2575, + "step": 57 + }, + { + "epoch": 0.11, + "grad_norm": 0.2402486598118377, + "learning_rate": 2.829268292682927e-05, + "loss": 1.2538, + "step": 58 + }, + { + "epoch": 0.11, + "grad_norm": 0.2549409484173144, + "learning_rate": 2.878048780487805e-05, + "loss": 1.2065, + "step": 59 + }, + { + "epoch": 0.11, + "grad_norm": 0.2053105349872685, + "learning_rate": 2.926829268292683e-05, + "loss": 1.2094, + "step": 60 + }, + { + "epoch": 0.11, + "grad_norm": 0.17971910872957886, + "learning_rate": 2.9756097560975613e-05, + "loss": 1.228, + "step": 61 + }, + { + "epoch": 0.11, + "grad_norm": 0.1885853654992973, + "learning_rate": 3.0243902439024392e-05, + "loss": 1.2286, + "step": 62 + }, + { + "epoch": 0.12, + "grad_norm": 0.1848524571968613, + "learning_rate": 3.073170731707317e-05, + "loss": 1.2718, + "step": 63 + }, + { + "epoch": 0.12, + "grad_norm": 0.18734105883548513, + "learning_rate": 3.1219512195121955e-05, + "loss": 1.2357, + "step": 64 + }, + { + "epoch": 0.12, + "grad_norm": 0.17774668052121825, + "learning_rate": 3.170731707317074e-05, + "loss": 1.1509, + "step": 65 + }, + { + "epoch": 0.12, + "grad_norm": 0.17890968008080646, + "learning_rate": 3.2195121951219514e-05, + "loss": 1.1924, + "step": 66 + }, + { + "epoch": 0.12, + "grad_norm": 0.18249273371332375, + "learning_rate": 3.268292682926829e-05, + "loss": 1.2545, + "step": 67 + }, + { + "epoch": 0.12, + "grad_norm": 0.21064122671902577, + "learning_rate": 3.3170731707317074e-05, + "loss": 1.2832, + "step": 68 + }, + { + "epoch": 0.13, + "grad_norm": 0.1820064171955093, + "learning_rate": 3.365853658536586e-05, + "loss": 1.2071, + "step": 69 + }, + { + "epoch": 0.13, + "grad_norm": 0.16996662800553433, + "learning_rate": 3.414634146341463e-05, + "loss": 1.2073, + "step": 70 + }, + { + "epoch": 0.13, + "grad_norm": 0.1618669302922445, + "learning_rate": 3.4634146341463416e-05, + "loss": 1.1289, + "step": 71 + }, + { + "epoch": 0.13, + "grad_norm": 0.18948744950985544, + "learning_rate": 3.51219512195122e-05, + "loss": 1.2915, + "step": 72 + }, + { + "epoch": 0.13, + "grad_norm": 0.18326143691603383, + "learning_rate": 3.5609756097560976e-05, + "loss": 1.2238, + "step": 73 + }, + { + "epoch": 0.14, + "grad_norm": 0.17410704510700503, + "learning_rate": 3.609756097560976e-05, + "loss": 1.1784, + "step": 74 + }, + { + "epoch": 0.14, + "grad_norm": 0.1983667344995625, + "learning_rate": 3.658536585365854e-05, + "loss": 1.2452, + "step": 75 + }, + { + "epoch": 0.14, + "grad_norm": 0.3416310763369357, + "learning_rate": 3.7073170731707325e-05, + "loss": 1.1972, + "step": 76 + }, + { + "epoch": 0.14, + "grad_norm": 0.2776466983511955, + "learning_rate": 3.75609756097561e-05, + "loss": 1.3121, + "step": 77 + }, + { + "epoch": 0.14, + "grad_norm": 0.20026129636576834, + "learning_rate": 3.804878048780488e-05, + "loss": 1.2436, + "step": 78 + }, + { + "epoch": 0.14, + "grad_norm": 0.21064549243917835, + "learning_rate": 3.853658536585366e-05, + "loss": 1.2064, + "step": 79 + }, + { + "epoch": 0.15, + "grad_norm": 0.22119482175714267, + "learning_rate": 3.9024390243902444e-05, + "loss": 1.2715, + "step": 80 + }, + { + "epoch": 0.15, + "grad_norm": 0.23047133748844142, + "learning_rate": 3.951219512195122e-05, + "loss": 1.2888, + "step": 81 + }, + { + "epoch": 0.15, + "grad_norm": 0.18741863156973176, + "learning_rate": 4e-05, + "loss": 1.248, + "step": 82 + }, + { + "epoch": 0.15, + "grad_norm": 0.1747859810629604, + "learning_rate": 4.0487804878048786e-05, + "loss": 1.1683, + "step": 83 + }, + { + "epoch": 0.15, + "grad_norm": 0.1896944798413341, + "learning_rate": 4.097560975609756e-05, + "loss": 1.2155, + "step": 84 + }, + { + "epoch": 0.16, + "grad_norm": 0.18724128114363303, + "learning_rate": 4.1463414634146346e-05, + "loss": 1.2273, + "step": 85 + }, + { + "epoch": 0.16, + "grad_norm": 0.17368125504855478, + "learning_rate": 4.195121951219513e-05, + "loss": 1.224, + "step": 86 + }, + { + "epoch": 0.16, + "grad_norm": 0.18371141013625703, + "learning_rate": 4.2439024390243905e-05, + "loss": 1.2294, + "step": 87 + }, + { + "epoch": 0.16, + "grad_norm": 0.1791029365673714, + "learning_rate": 4.292682926829269e-05, + "loss": 1.2895, + "step": 88 + }, + { + "epoch": 0.16, + "grad_norm": 0.20259974283859655, + "learning_rate": 4.341463414634147e-05, + "loss": 1.1841, + "step": 89 + }, + { + "epoch": 0.16, + "grad_norm": 0.17457456183272174, + "learning_rate": 4.390243902439025e-05, + "loss": 1.2357, + "step": 90 + }, + { + "epoch": 0.17, + "grad_norm": 0.1815824380789748, + "learning_rate": 4.439024390243903e-05, + "loss": 1.2304, + "step": 91 + }, + { + "epoch": 0.17, + "grad_norm": 0.17566480599583392, + "learning_rate": 4.4878048780487814e-05, + "loss": 1.242, + "step": 92 + }, + { + "epoch": 0.17, + "grad_norm": 0.18422975005984474, + "learning_rate": 4.536585365853658e-05, + "loss": 1.2177, + "step": 93 + }, + { + "epoch": 0.17, + "grad_norm": 0.16796781877940678, + "learning_rate": 4.5853658536585366e-05, + "loss": 1.1482, + "step": 94 + }, + { + "epoch": 0.17, + "grad_norm": 0.18636131653783305, + "learning_rate": 4.634146341463415e-05, + "loss": 1.1758, + "step": 95 + }, + { + "epoch": 0.18, + "grad_norm": 0.1823665700289814, + "learning_rate": 4.6829268292682926e-05, + "loss": 1.289, + "step": 96 + }, + { + "epoch": 0.18, + "grad_norm": 0.1719900691262439, + "learning_rate": 4.731707317073171e-05, + "loss": 1.1626, + "step": 97 + }, + { + "epoch": 0.18, + "grad_norm": 0.17937994168039778, + "learning_rate": 4.780487804878049e-05, + "loss": 1.175, + "step": 98 + }, + { + "epoch": 0.18, + "grad_norm": 0.16631851422106986, + "learning_rate": 4.829268292682927e-05, + "loss": 1.2177, + "step": 99 + }, + { + "epoch": 0.18, + "grad_norm": 0.19143696232800309, + "learning_rate": 4.878048780487805e-05, + "loss": 1.3071, + "step": 100 + }, + { + "epoch": 0.18, + "grad_norm": 0.17859506638780318, + "learning_rate": 4.9268292682926835e-05, + "loss": 1.2351, + "step": 101 + }, + { + "epoch": 0.19, + "grad_norm": 0.18381520321248196, + "learning_rate": 4.975609756097561e-05, + "loss": 1.2342, + "step": 102 + }, + { + "epoch": 0.19, + "grad_norm": 0.17968218683773912, + "learning_rate": 5.0243902439024394e-05, + "loss": 1.2074, + "step": 103 + }, + { + "epoch": 0.19, + "grad_norm": 0.18139489969339018, + "learning_rate": 5.073170731707318e-05, + "loss": 1.1558, + "step": 104 + }, + { + "epoch": 0.19, + "grad_norm": 0.17366624842514394, + "learning_rate": 5.121951219512195e-05, + "loss": 1.1897, + "step": 105 + }, + { + "epoch": 0.19, + "grad_norm": 0.16034845455223745, + "learning_rate": 5.1707317073170736e-05, + "loss": 1.179, + "step": 106 + }, + { + "epoch": 0.2, + "grad_norm": 0.17583069577827776, + "learning_rate": 5.219512195121952e-05, + "loss": 1.1856, + "step": 107 + }, + { + "epoch": 0.2, + "grad_norm": 0.1853758076989552, + "learning_rate": 5.26829268292683e-05, + "loss": 1.2072, + "step": 108 + }, + { + "epoch": 0.2, + "grad_norm": 0.19597443965936462, + "learning_rate": 5.317073170731708e-05, + "loss": 1.2271, + "step": 109 + }, + { + "epoch": 0.2, + "grad_norm": 0.1899206334098331, + "learning_rate": 5.365853658536586e-05, + "loss": 1.1961, + "step": 110 + }, + { + "epoch": 0.2, + "grad_norm": 0.17463763837757018, + "learning_rate": 5.4146341463414645e-05, + "loss": 1.2049, + "step": 111 + }, + { + "epoch": 0.2, + "grad_norm": 0.20431371701229986, + "learning_rate": 5.463414634146342e-05, + "loss": 1.2891, + "step": 112 + }, + { + "epoch": 0.21, + "grad_norm": 0.1814475107638498, + "learning_rate": 5.51219512195122e-05, + "loss": 1.2346, + "step": 113 + }, + { + "epoch": 0.21, + "grad_norm": 0.1883849423207823, + "learning_rate": 5.5609756097560974e-05, + "loss": 1.244, + "step": 114 + }, + { + "epoch": 0.21, + "grad_norm": 0.1857258128640568, + "learning_rate": 5.609756097560976e-05, + "loss": 1.2669, + "step": 115 + }, + { + "epoch": 0.21, + "grad_norm": 0.1740768514118401, + "learning_rate": 5.658536585365854e-05, + "loss": 1.2414, + "step": 116 + }, + { + "epoch": 0.21, + "grad_norm": 0.1919320335584178, + "learning_rate": 5.7073170731707317e-05, + "loss": 1.2886, + "step": 117 + }, + { + "epoch": 0.22, + "grad_norm": 0.18288775167828136, + "learning_rate": 5.75609756097561e-05, + "loss": 1.1875, + "step": 118 + }, + { + "epoch": 0.22, + "grad_norm": 0.18208588867750863, + "learning_rate": 5.804878048780488e-05, + "loss": 1.2388, + "step": 119 + }, + { + "epoch": 0.22, + "grad_norm": 0.1743260015658331, + "learning_rate": 5.853658536585366e-05, + "loss": 1.1762, + "step": 120 + }, + { + "epoch": 0.22, + "grad_norm": 0.17856046291517946, + "learning_rate": 5.902439024390244e-05, + "loss": 1.2888, + "step": 121 + }, + { + "epoch": 0.22, + "grad_norm": 0.17493794870966536, + "learning_rate": 5.9512195121951225e-05, + "loss": 1.2222, + "step": 122 + }, + { + "epoch": 0.22, + "grad_norm": 0.1909202655203384, + "learning_rate": 6.000000000000001e-05, + "loss": 1.2414, + "step": 123 + }, + { + "epoch": 0.23, + "grad_norm": 0.18345819482834988, + "learning_rate": 6.0487804878048785e-05, + "loss": 1.2756, + "step": 124 + }, + { + "epoch": 0.23, + "grad_norm": 0.2057069352956621, + "learning_rate": 6.097560975609757e-05, + "loss": 1.261, + "step": 125 + }, + { + "epoch": 0.23, + "grad_norm": 0.299775882469108, + "learning_rate": 6.146341463414634e-05, + "loss": 1.2566, + "step": 126 + }, + { + "epoch": 0.23, + "grad_norm": 0.1869687633018095, + "learning_rate": 6.195121951219513e-05, + "loss": 1.3039, + "step": 127 + }, + { + "epoch": 0.23, + "grad_norm": 0.17747149926197442, + "learning_rate": 6.243902439024391e-05, + "loss": 1.2524, + "step": 128 + }, + { + "epoch": 0.24, + "grad_norm": 0.17885157788044242, + "learning_rate": 6.29268292682927e-05, + "loss": 1.2455, + "step": 129 + }, + { + "epoch": 0.24, + "grad_norm": 0.17617298187845123, + "learning_rate": 6.341463414634148e-05, + "loss": 1.2009, + "step": 130 + }, + { + "epoch": 0.24, + "grad_norm": 0.20164176323497066, + "learning_rate": 6.390243902439025e-05, + "loss": 1.2634, + "step": 131 + }, + { + "epoch": 0.24, + "grad_norm": 0.20459903417307612, + "learning_rate": 6.439024390243903e-05, + "loss": 1.1963, + "step": 132 + }, + { + "epoch": 0.24, + "grad_norm": 0.1863755486334296, + "learning_rate": 6.487804878048781e-05, + "loss": 1.2387, + "step": 133 + }, + { + "epoch": 0.25, + "grad_norm": 0.19265866140295207, + "learning_rate": 6.536585365853658e-05, + "loss": 1.2688, + "step": 134 + }, + { + "epoch": 0.25, + "grad_norm": 0.1823425868969493, + "learning_rate": 6.585365853658536e-05, + "loss": 1.2041, + "step": 135 + }, + { + "epoch": 0.25, + "grad_norm": 0.2016853266472781, + "learning_rate": 6.634146341463415e-05, + "loss": 1.1223, + "step": 136 + }, + { + "epoch": 0.25, + "grad_norm": 0.17282675192463448, + "learning_rate": 6.682926829268293e-05, + "loss": 1.1879, + "step": 137 + }, + { + "epoch": 0.25, + "grad_norm": 0.17398811693399288, + "learning_rate": 6.731707317073171e-05, + "loss": 1.2682, + "step": 138 + }, + { + "epoch": 0.25, + "grad_norm": 0.18516916965434696, + "learning_rate": 6.78048780487805e-05, + "loss": 1.1666, + "step": 139 + }, + { + "epoch": 0.26, + "grad_norm": 0.1852213129647933, + "learning_rate": 6.829268292682927e-05, + "loss": 1.2501, + "step": 140 + }, + { + "epoch": 0.26, + "grad_norm": 0.17915948766591883, + "learning_rate": 6.878048780487805e-05, + "loss": 1.2264, + "step": 141 + }, + { + "epoch": 0.26, + "grad_norm": 0.21599939417233183, + "learning_rate": 6.926829268292683e-05, + "loss": 1.2376, + "step": 142 + }, + { + "epoch": 0.26, + "grad_norm": 0.17839304459521851, + "learning_rate": 6.975609756097562e-05, + "loss": 1.2353, + "step": 143 + }, + { + "epoch": 0.26, + "grad_norm": 0.20826913231380875, + "learning_rate": 7.02439024390244e-05, + "loss": 1.1901, + "step": 144 + }, + { + "epoch": 0.27, + "grad_norm": 0.20788894913361589, + "learning_rate": 7.073170731707318e-05, + "loss": 1.2577, + "step": 145 + }, + { + "epoch": 0.27, + "grad_norm": 0.18420055842301297, + "learning_rate": 7.121951219512195e-05, + "loss": 1.1393, + "step": 146 + }, + { + "epoch": 0.27, + "grad_norm": 0.19903048468685589, + "learning_rate": 7.170731707317073e-05, + "loss": 1.2321, + "step": 147 + }, + { + "epoch": 0.27, + "grad_norm": 0.19074116314985748, + "learning_rate": 7.219512195121952e-05, + "loss": 1.1912, + "step": 148 + }, + { + "epoch": 0.27, + "grad_norm": 0.2353816469403903, + "learning_rate": 7.26829268292683e-05, + "loss": 1.28, + "step": 149 + }, + { + "epoch": 0.27, + "grad_norm": 0.21634875684769345, + "learning_rate": 7.317073170731708e-05, + "loss": 1.3312, + "step": 150 + }, + { + "epoch": 0.28, + "grad_norm": 0.18290969006743918, + "learning_rate": 7.365853658536587e-05, + "loss": 1.2214, + "step": 151 + }, + { + "epoch": 0.28, + "grad_norm": 0.18484243897545208, + "learning_rate": 7.414634146341465e-05, + "loss": 1.1895, + "step": 152 + }, + { + "epoch": 0.28, + "grad_norm": 0.21882343112978872, + "learning_rate": 7.463414634146342e-05, + "loss": 1.2219, + "step": 153 + }, + { + "epoch": 0.28, + "grad_norm": 0.19868284379241205, + "learning_rate": 7.51219512195122e-05, + "loss": 1.2176, + "step": 154 + }, + { + "epoch": 0.28, + "grad_norm": 0.20912516312950613, + "learning_rate": 7.560975609756097e-05, + "loss": 1.242, + "step": 155 + }, + { + "epoch": 0.29, + "grad_norm": 0.23811880045549916, + "learning_rate": 7.609756097560976e-05, + "loss": 1.2838, + "step": 156 + }, + { + "epoch": 0.29, + "grad_norm": 0.19511077122033713, + "learning_rate": 7.658536585365854e-05, + "loss": 1.1594, + "step": 157 + }, + { + "epoch": 0.29, + "grad_norm": 0.20094129399534238, + "learning_rate": 7.707317073170732e-05, + "loss": 1.2966, + "step": 158 + }, + { + "epoch": 0.29, + "grad_norm": 0.19366245038292418, + "learning_rate": 7.75609756097561e-05, + "loss": 1.2246, + "step": 159 + }, + { + "epoch": 0.29, + "grad_norm": 0.19409570223867306, + "learning_rate": 7.804878048780489e-05, + "loss": 1.2312, + "step": 160 + }, + { + "epoch": 0.29, + "grad_norm": 0.2087258457033805, + "learning_rate": 7.853658536585366e-05, + "loss": 1.2169, + "step": 161 + }, + { + "epoch": 0.3, + "grad_norm": 0.18765223996270428, + "learning_rate": 7.902439024390244e-05, + "loss": 1.2383, + "step": 162 + }, + { + "epoch": 0.3, + "grad_norm": 0.20734180224147242, + "learning_rate": 7.951219512195122e-05, + "loss": 1.2587, + "step": 163 + }, + { + "epoch": 0.3, + "grad_norm": 0.24690929540287834, + "learning_rate": 8e-05, + "loss": 1.1951, + "step": 164 + }, + { + "epoch": 0.3, + "grad_norm": 0.2003538797619543, + "learning_rate": 7.999990914797545e-05, + "loss": 1.1982, + "step": 165 + }, + { + "epoch": 0.3, + "grad_norm": 0.22469075613510484, + "learning_rate": 7.99996365923145e-05, + "loss": 1.2355, + "step": 166 + }, + { + "epoch": 0.31, + "grad_norm": 0.21870100788336058, + "learning_rate": 7.999918233425526e-05, + "loss": 1.1103, + "step": 167 + }, + { + "epoch": 0.31, + "grad_norm": 0.20939989594131886, + "learning_rate": 7.999854637586122e-05, + "loss": 1.1966, + "step": 168 + }, + { + "epoch": 0.31, + "grad_norm": 0.43108211416237796, + "learning_rate": 7.999772872002132e-05, + "loss": 1.2882, + "step": 169 + }, + { + "epoch": 0.31, + "grad_norm": 0.27045413432174487, + "learning_rate": 7.999672937044984e-05, + "loss": 1.2399, + "step": 170 + }, + { + "epoch": 0.31, + "grad_norm": 0.19700483036740515, + "learning_rate": 7.999554833168642e-05, + "loss": 1.202, + "step": 171 + }, + { + "epoch": 0.31, + "grad_norm": 0.3335979493370708, + "learning_rate": 7.999418560909604e-05, + "loss": 1.1995, + "step": 172 + }, + { + "epoch": 0.32, + "grad_norm": 0.3165803974474567, + "learning_rate": 7.999264120886902e-05, + "loss": 1.1569, + "step": 173 + }, + { + "epoch": 0.32, + "grad_norm": 0.1951699080346223, + "learning_rate": 7.999091513802093e-05, + "loss": 1.1778, + "step": 174 + }, + { + "epoch": 0.32, + "grad_norm": 0.2087559121749787, + "learning_rate": 7.998900740439265e-05, + "loss": 1.1736, + "step": 175 + }, + { + "epoch": 0.32, + "grad_norm": 0.20345180977460478, + "learning_rate": 7.998691801665024e-05, + "loss": 1.2281, + "step": 176 + }, + { + "epoch": 0.32, + "grad_norm": 0.24617644827252333, + "learning_rate": 7.998464698428495e-05, + "loss": 1.2072, + "step": 177 + }, + { + "epoch": 0.33, + "grad_norm": 0.2469050959356265, + "learning_rate": 7.998219431761318e-05, + "loss": 1.2242, + "step": 178 + }, + { + "epoch": 0.33, + "grad_norm": 0.19529317748460623, + "learning_rate": 7.997956002777642e-05, + "loss": 1.2567, + "step": 179 + }, + { + "epoch": 0.33, + "grad_norm": 0.19048389491381376, + "learning_rate": 7.99767441267412e-05, + "loss": 1.2982, + "step": 180 + }, + { + "epoch": 0.33, + "grad_norm": 0.2085799116493225, + "learning_rate": 7.997374662729904e-05, + "loss": 1.1254, + "step": 181 + }, + { + "epoch": 0.33, + "grad_norm": 0.20636853256378995, + "learning_rate": 7.997056754306636e-05, + "loss": 1.2435, + "step": 182 + }, + { + "epoch": 0.33, + "grad_norm": 0.20590016382290252, + "learning_rate": 7.99672068884845e-05, + "loss": 1.2658, + "step": 183 + }, + { + "epoch": 0.34, + "grad_norm": 0.1931166169764433, + "learning_rate": 7.996366467881955e-05, + "loss": 1.1637, + "step": 184 + }, + { + "epoch": 0.34, + "grad_norm": 0.18873318157988098, + "learning_rate": 7.995994093016237e-05, + "loss": 1.1335, + "step": 185 + }, + { + "epoch": 0.34, + "grad_norm": 0.19210254625199108, + "learning_rate": 7.995603565942846e-05, + "loss": 1.1928, + "step": 186 + }, + { + "epoch": 0.34, + "grad_norm": 0.2130986479765664, + "learning_rate": 7.995194888435792e-05, + "loss": 1.2158, + "step": 187 + }, + { + "epoch": 0.34, + "grad_norm": 0.22003854501814088, + "learning_rate": 7.994768062351532e-05, + "loss": 1.2288, + "step": 188 + }, + { + "epoch": 0.35, + "grad_norm": 0.20330803191993058, + "learning_rate": 7.994323089628968e-05, + "loss": 1.2426, + "step": 189 + }, + { + "epoch": 0.35, + "grad_norm": 0.20567314642208634, + "learning_rate": 7.993859972289434e-05, + "loss": 1.2649, + "step": 190 + }, + { + "epoch": 0.35, + "grad_norm": 0.21556663727342962, + "learning_rate": 7.993378712436686e-05, + "loss": 1.2545, + "step": 191 + }, + { + "epoch": 0.35, + "grad_norm": 0.20309165469109888, + "learning_rate": 7.992879312256897e-05, + "loss": 1.3338, + "step": 192 + }, + { + "epoch": 0.35, + "grad_norm": 0.19574356669421325, + "learning_rate": 7.992361774018641e-05, + "loss": 1.278, + "step": 193 + }, + { + "epoch": 0.35, + "grad_norm": 0.2763613746722313, + "learning_rate": 7.991826100072891e-05, + "loss": 1.2571, + "step": 194 + }, + { + "epoch": 0.36, + "grad_norm": 0.19346552479915102, + "learning_rate": 7.991272292852996e-05, + "loss": 1.2027, + "step": 195 + }, + { + "epoch": 0.36, + "grad_norm": 0.2281167812123908, + "learning_rate": 7.990700354874683e-05, + "loss": 1.2586, + "step": 196 + }, + { + "epoch": 0.36, + "grad_norm": 0.19699013712137542, + "learning_rate": 7.990110288736042e-05, + "loss": 1.1371, + "step": 197 + }, + { + "epoch": 0.36, + "grad_norm": 0.21768209981475933, + "learning_rate": 7.989502097117503e-05, + "loss": 1.2522, + "step": 198 + }, + { + "epoch": 0.36, + "grad_norm": 0.21335427847754582, + "learning_rate": 7.988875782781838e-05, + "loss": 1.2437, + "step": 199 + }, + { + "epoch": 0.37, + "grad_norm": 0.21856710629066897, + "learning_rate": 7.988231348574147e-05, + "loss": 1.2135, + "step": 200 + }, + { + "epoch": 0.37, + "grad_norm": 0.20482062658774797, + "learning_rate": 7.987568797421836e-05, + "loss": 1.1755, + "step": 201 + }, + { + "epoch": 0.37, + "grad_norm": 0.2017756813960897, + "learning_rate": 7.986888132334608e-05, + "loss": 1.1699, + "step": 202 + }, + { + "epoch": 0.37, + "grad_norm": 0.20496443848153809, + "learning_rate": 7.986189356404458e-05, + "loss": 1.2125, + "step": 203 + }, + { + "epoch": 0.37, + "grad_norm": 0.2134603800558358, + "learning_rate": 7.985472472805643e-05, + "loss": 1.2391, + "step": 204 + }, + { + "epoch": 0.37, + "grad_norm": 0.2364175573420861, + "learning_rate": 7.98473748479468e-05, + "loss": 1.2384, + "step": 205 + }, + { + "epoch": 0.38, + "grad_norm": 0.1872419861598724, + "learning_rate": 7.983984395710326e-05, + "loss": 1.1457, + "step": 206 + }, + { + "epoch": 0.38, + "grad_norm": 0.28222194007095774, + "learning_rate": 7.983213208973566e-05, + "loss": 1.2952, + "step": 207 + }, + { + "epoch": 0.38, + "grad_norm": 0.1916094851162064, + "learning_rate": 7.982423928087593e-05, + "loss": 1.1763, + "step": 208 + }, + { + "epoch": 0.38, + "grad_norm": 0.18446245256166657, + "learning_rate": 7.981616556637795e-05, + "loss": 1.1863, + "step": 209 + }, + { + "epoch": 0.38, + "grad_norm": 0.195191961022491, + "learning_rate": 7.980791098291737e-05, + "loss": 1.2036, + "step": 210 + }, + { + "epoch": 0.39, + "grad_norm": 0.2652439657825496, + "learning_rate": 7.979947556799151e-05, + "loss": 1.2834, + "step": 211 + }, + { + "epoch": 0.39, + "grad_norm": 0.24308438957843412, + "learning_rate": 7.979085935991906e-05, + "loss": 1.234, + "step": 212 + }, + { + "epoch": 0.39, + "grad_norm": 0.21294701043622016, + "learning_rate": 7.978206239784004e-05, + "loss": 1.3006, + "step": 213 + }, + { + "epoch": 0.39, + "grad_norm": 0.25809277041859524, + "learning_rate": 7.977308472171553e-05, + "loss": 1.2272, + "step": 214 + }, + { + "epoch": 0.39, + "grad_norm": 0.193463860107294, + "learning_rate": 7.976392637232754e-05, + "loss": 1.2295, + "step": 215 + }, + { + "epoch": 0.4, + "grad_norm": 0.2150023760609626, + "learning_rate": 7.975458739127877e-05, + "loss": 1.2135, + "step": 216 + }, + { + "epoch": 0.4, + "grad_norm": 0.22590495955605894, + "learning_rate": 7.974506782099253e-05, + "loss": 1.2532, + "step": 217 + }, + { + "epoch": 0.4, + "grad_norm": 0.21023744668403702, + "learning_rate": 7.973536770471242e-05, + "loss": 1.2472, + "step": 218 + }, + { + "epoch": 0.4, + "grad_norm": 0.2345749799511543, + "learning_rate": 7.972548708650218e-05, + "loss": 1.1791, + "step": 219 + }, + { + "epoch": 0.4, + "grad_norm": 0.2158876734005217, + "learning_rate": 7.971542601124553e-05, + "loss": 1.2483, + "step": 220 + }, + { + "epoch": 0.4, + "grad_norm": 0.29455339949432446, + "learning_rate": 7.970518452464593e-05, + "loss": 1.2894, + "step": 221 + }, + { + "epoch": 0.41, + "grad_norm": 0.23983708730626851, + "learning_rate": 7.969476267322636e-05, + "loss": 1.271, + "step": 222 + }, + { + "epoch": 0.41, + "grad_norm": 0.1922400905426158, + "learning_rate": 7.968416050432912e-05, + "loss": 1.2139, + "step": 223 + }, + { + "epoch": 0.41, + "grad_norm": 0.2238136844422931, + "learning_rate": 7.967337806611568e-05, + "loss": 1.2655, + "step": 224 + }, + { + "epoch": 0.41, + "grad_norm": 0.21230292828267672, + "learning_rate": 7.966241540756631e-05, + "loss": 1.2406, + "step": 225 + }, + { + "epoch": 0.41, + "grad_norm": 0.26656119419070456, + "learning_rate": 7.965127257848004e-05, + "loss": 1.2595, + "step": 226 + }, + { + "epoch": 0.42, + "grad_norm": 0.22381385502992684, + "learning_rate": 7.963994962947426e-05, + "loss": 1.1737, + "step": 227 + }, + { + "epoch": 0.42, + "grad_norm": 0.20056702203994298, + "learning_rate": 7.962844661198462e-05, + "loss": 1.1969, + "step": 228 + }, + { + "epoch": 0.42, + "grad_norm": 0.20148701321526885, + "learning_rate": 7.961676357826478e-05, + "loss": 1.2151, + "step": 229 + }, + { + "epoch": 0.42, + "grad_norm": 0.20034834807028637, + "learning_rate": 7.960490058138604e-05, + "loss": 1.1455, + "step": 230 + }, + { + "epoch": 0.42, + "grad_norm": 0.21050838521846033, + "learning_rate": 7.959285767523732e-05, + "loss": 1.2223, + "step": 231 + }, + { + "epoch": 0.42, + "grad_norm": 0.20904772138969777, + "learning_rate": 7.95806349145247e-05, + "loss": 1.2534, + "step": 232 + }, + { + "epoch": 0.43, + "grad_norm": 0.20307877304792957, + "learning_rate": 7.956823235477134e-05, + "loss": 1.1352, + "step": 233 + }, + { + "epoch": 0.43, + "grad_norm": 0.20501105270897094, + "learning_rate": 7.95556500523171e-05, + "loss": 1.2031, + "step": 234 + }, + { + "epoch": 0.43, + "grad_norm": 0.19800586972038586, + "learning_rate": 7.954288806431838e-05, + "loss": 1.2567, + "step": 235 + }, + { + "epoch": 0.43, + "grad_norm": 0.2175102450594135, + "learning_rate": 7.952994644874777e-05, + "loss": 1.2538, + "step": 236 + }, + { + "epoch": 0.43, + "grad_norm": 0.22698189300067595, + "learning_rate": 7.951682526439391e-05, + "loss": 1.3088, + "step": 237 + }, + { + "epoch": 0.44, + "grad_norm": 0.19208392014975315, + "learning_rate": 7.950352457086109e-05, + "loss": 1.2336, + "step": 238 + }, + { + "epoch": 0.44, + "grad_norm": 0.27004086334319655, + "learning_rate": 7.949004442856905e-05, + "loss": 1.2012, + "step": 239 + }, + { + "epoch": 0.44, + "grad_norm": 0.23420974954538043, + "learning_rate": 7.947638489875272e-05, + "loss": 1.2244, + "step": 240 + }, + { + "epoch": 0.44, + "grad_norm": 0.20514399124802024, + "learning_rate": 7.946254604346186e-05, + "loss": 1.2548, + "step": 241 + }, + { + "epoch": 0.44, + "grad_norm": 0.19334973602372896, + "learning_rate": 7.944852792556092e-05, + "loss": 1.2104, + "step": 242 + }, + { + "epoch": 0.44, + "grad_norm": 0.1992640714537956, + "learning_rate": 7.943433060872858e-05, + "loss": 1.2628, + "step": 243 + }, + { + "epoch": 0.45, + "grad_norm": 0.203284617090413, + "learning_rate": 7.941995415745761e-05, + "loss": 1.2002, + "step": 244 + }, + { + "epoch": 0.45, + "grad_norm": 0.22795306969682058, + "learning_rate": 7.94053986370545e-05, + "loss": 1.2215, + "step": 245 + }, + { + "epoch": 0.45, + "grad_norm": 0.20789041346838505, + "learning_rate": 7.939066411363915e-05, + "loss": 1.0998, + "step": 246 + }, + { + "epoch": 0.45, + "grad_norm": 0.22354868884742066, + "learning_rate": 7.937575065414464e-05, + "loss": 1.2564, + "step": 247 + }, + { + "epoch": 0.45, + "grad_norm": 0.21176392726647736, + "learning_rate": 7.936065832631687e-05, + "loss": 1.2816, + "step": 248 + }, + { + "epoch": 0.46, + "grad_norm": 0.19967179557235587, + "learning_rate": 7.934538719871427e-05, + "loss": 1.1961, + "step": 249 + }, + { + "epoch": 0.46, + "grad_norm": 0.210819577350627, + "learning_rate": 7.932993734070747e-05, + "loss": 1.2167, + "step": 250 + }, + { + "epoch": 0.46, + "grad_norm": 0.21537794551756187, + "learning_rate": 7.931430882247903e-05, + "loss": 1.2341, + "step": 251 + }, + { + "epoch": 0.46, + "grad_norm": 0.22850872387256574, + "learning_rate": 7.929850171502304e-05, + "loss": 1.1686, + "step": 252 + }, + { + "epoch": 0.46, + "grad_norm": 0.22380366415076383, + "learning_rate": 7.928251609014493e-05, + "loss": 1.1462, + "step": 253 + }, + { + "epoch": 0.46, + "grad_norm": 0.22426923149036065, + "learning_rate": 7.926635202046102e-05, + "loss": 1.1792, + "step": 254 + }, + { + "epoch": 0.47, + "grad_norm": 0.42082703321103965, + "learning_rate": 7.925000957939822e-05, + "loss": 1.2718, + "step": 255 + }, + { + "epoch": 0.47, + "grad_norm": 0.2235432774854074, + "learning_rate": 7.92334888411937e-05, + "loss": 1.2598, + "step": 256 + }, + { + "epoch": 0.47, + "grad_norm": 0.281644028934108, + "learning_rate": 7.92167898808946e-05, + "loss": 1.2205, + "step": 257 + }, + { + "epoch": 0.47, + "grad_norm": 0.2037705143888748, + "learning_rate": 7.919991277435763e-05, + "loss": 1.1737, + "step": 258 + }, + { + "epoch": 0.47, + "grad_norm": 0.20917419230028977, + "learning_rate": 7.918285759824879e-05, + "loss": 1.2035, + "step": 259 + }, + { + "epoch": 0.48, + "grad_norm": 0.20510847570635518, + "learning_rate": 7.916562443004292e-05, + "loss": 1.2135, + "step": 260 + }, + { + "epoch": 0.48, + "grad_norm": 0.25172483071092466, + "learning_rate": 7.914821334802342e-05, + "loss": 1.2218, + "step": 261 + }, + { + "epoch": 0.48, + "grad_norm": 0.21102706700634313, + "learning_rate": 7.91306244312819e-05, + "loss": 1.1738, + "step": 262 + }, + { + "epoch": 0.48, + "grad_norm": 0.22626060872645815, + "learning_rate": 7.911285775971781e-05, + "loss": 1.238, + "step": 263 + }, + { + "epoch": 0.48, + "grad_norm": 0.22448567539778486, + "learning_rate": 7.909491341403805e-05, + "loss": 1.2404, + "step": 264 + }, + { + "epoch": 0.48, + "grad_norm": 0.2019099786139193, + "learning_rate": 7.907679147575661e-05, + "loss": 1.213, + "step": 265 + }, + { + "epoch": 0.49, + "grad_norm": 0.24307234839096267, + "learning_rate": 7.905849202719422e-05, + "loss": 1.2322, + "step": 266 + }, + { + "epoch": 0.49, + "grad_norm": 0.19801890521743487, + "learning_rate": 7.904001515147802e-05, + "loss": 1.2448, + "step": 267 + }, + { + "epoch": 0.49, + "grad_norm": 0.2102742273575385, + "learning_rate": 7.902136093254106e-05, + "loss": 1.1657, + "step": 268 + }, + { + "epoch": 0.49, + "grad_norm": 0.2173464476815016, + "learning_rate": 7.900252945512201e-05, + "loss": 1.2549, + "step": 269 + }, + { + "epoch": 0.49, + "grad_norm": 0.20957275458699595, + "learning_rate": 7.898352080476479e-05, + "loss": 1.2536, + "step": 270 + }, + { + "epoch": 0.5, + "grad_norm": 0.20691966388952363, + "learning_rate": 7.896433506781811e-05, + "loss": 1.2661, + "step": 271 + }, + { + "epoch": 0.5, + "grad_norm": 0.2276662275112648, + "learning_rate": 7.894497233143509e-05, + "loss": 1.2409, + "step": 272 + }, + { + "epoch": 0.5, + "grad_norm": 0.23854109569301263, + "learning_rate": 7.892543268357297e-05, + "loss": 1.2681, + "step": 273 + }, + { + "epoch": 0.5, + "grad_norm": 0.2233864156677627, + "learning_rate": 7.890571621299252e-05, + "loss": 1.1687, + "step": 274 + }, + { + "epoch": 0.5, + "grad_norm": 0.20114129147925475, + "learning_rate": 7.888582300925787e-05, + "loss": 1.2184, + "step": 275 + }, + { + "epoch": 0.5, + "grad_norm": 0.2154654670569462, + "learning_rate": 7.886575316273586e-05, + "loss": 1.1982, + "step": 276 + }, + { + "epoch": 0.51, + "grad_norm": 0.2292982209343639, + "learning_rate": 7.884550676459583e-05, + "loss": 1.2129, + "step": 277 + }, + { + "epoch": 0.51, + "grad_norm": 0.21302713135229548, + "learning_rate": 7.882508390680908e-05, + "loss": 1.1605, + "step": 278 + }, + { + "epoch": 0.51, + "grad_norm": 0.2123661020671048, + "learning_rate": 7.88044846821485e-05, + "loss": 1.2308, + "step": 279 + }, + { + "epoch": 0.51, + "grad_norm": 0.2080577410800404, + "learning_rate": 7.878370918418818e-05, + "loss": 1.2195, + "step": 280 + }, + { + "epoch": 0.51, + "grad_norm": 0.19663901881127385, + "learning_rate": 7.876275750730289e-05, + "loss": 1.1591, + "step": 281 + }, + { + "epoch": 0.52, + "grad_norm": 0.20534502031312163, + "learning_rate": 7.874162974666776e-05, + "loss": 1.2664, + "step": 282 + }, + { + "epoch": 0.52, + "grad_norm": 0.23240445399513837, + "learning_rate": 7.872032599825779e-05, + "loss": 1.2151, + "step": 283 + }, + { + "epoch": 0.52, + "grad_norm": 0.2672527316717507, + "learning_rate": 7.86988463588474e-05, + "loss": 1.2406, + "step": 284 + }, + { + "epoch": 0.52, + "grad_norm": 0.19893903058743695, + "learning_rate": 7.867719092601003e-05, + "loss": 1.1291, + "step": 285 + }, + { + "epoch": 0.52, + "grad_norm": 0.33275268109930917, + "learning_rate": 7.865535979811768e-05, + "loss": 1.1406, + "step": 286 + }, + { + "epoch": 0.52, + "grad_norm": 0.2373619455690358, + "learning_rate": 7.863335307434045e-05, + "loss": 1.2799, + "step": 287 + }, + { + "epoch": 0.53, + "grad_norm": 0.263235735390858, + "learning_rate": 7.861117085464612e-05, + "loss": 1.2415, + "step": 288 + }, + { + "epoch": 0.53, + "grad_norm": 0.25884281780784324, + "learning_rate": 7.858881323979965e-05, + "loss": 1.3919, + "step": 289 + }, + { + "epoch": 0.53, + "grad_norm": 0.25426288332255736, + "learning_rate": 7.85662803313628e-05, + "loss": 1.174, + "step": 290 + }, + { + "epoch": 0.53, + "grad_norm": 0.26655405527881243, + "learning_rate": 7.854357223169356e-05, + "loss": 1.2806, + "step": 291 + }, + { + "epoch": 0.53, + "grad_norm": 0.20909844432349833, + "learning_rate": 7.852068904394579e-05, + "loss": 1.2627, + "step": 292 + }, + { + "epoch": 0.54, + "grad_norm": 0.21307115068935759, + "learning_rate": 7.849763087206866e-05, + "loss": 1.1879, + "step": 293 + }, + { + "epoch": 0.54, + "grad_norm": 0.25009949471398946, + "learning_rate": 7.847439782080628e-05, + "loss": 1.2881, + "step": 294 + }, + { + "epoch": 0.54, + "grad_norm": 0.20960783418679174, + "learning_rate": 7.845098999569712e-05, + "loss": 1.2723, + "step": 295 + }, + { + "epoch": 0.54, + "grad_norm": 0.24968832437925104, + "learning_rate": 7.842740750307362e-05, + "loss": 1.2029, + "step": 296 + }, + { + "epoch": 0.54, + "grad_norm": 0.22981196585125677, + "learning_rate": 7.84036504500616e-05, + "loss": 1.1695, + "step": 297 + }, + { + "epoch": 0.55, + "grad_norm": 0.2320606844751365, + "learning_rate": 7.837971894457991e-05, + "loss": 1.2317, + "step": 298 + }, + { + "epoch": 0.55, + "grad_norm": 0.23051459673906124, + "learning_rate": 7.835561309533981e-05, + "loss": 1.2046, + "step": 299 + }, + { + "epoch": 0.55, + "grad_norm": 0.2510027231060586, + "learning_rate": 7.833133301184457e-05, + "loss": 1.199, + "step": 300 + }, + { + "epoch": 0.55, + "grad_norm": 0.23601180466018787, + "learning_rate": 7.830687880438895e-05, + "loss": 1.1755, + "step": 301 + }, + { + "epoch": 0.55, + "grad_norm": 0.24740820934385369, + "learning_rate": 7.828225058405864e-05, + "loss": 1.2054, + "step": 302 + }, + { + "epoch": 0.55, + "grad_norm": 0.23065372979111173, + "learning_rate": 7.825744846272984e-05, + "loss": 1.2066, + "step": 303 + }, + { + "epoch": 0.56, + "grad_norm": 0.22385077334838213, + "learning_rate": 7.823247255306866e-05, + "loss": 1.2147, + "step": 304 + }, + { + "epoch": 0.56, + "grad_norm": 0.42981213948386104, + "learning_rate": 7.820732296853074e-05, + "loss": 1.2314, + "step": 305 + }, + { + "epoch": 0.56, + "grad_norm": 0.21122844902751076, + "learning_rate": 7.818199982336058e-05, + "loss": 1.1462, + "step": 306 + }, + { + "epoch": 0.56, + "grad_norm": 0.23374869692118933, + "learning_rate": 7.815650323259117e-05, + "loss": 1.2051, + "step": 307 + }, + { + "epoch": 0.56, + "grad_norm": 0.21662363795962128, + "learning_rate": 7.813083331204332e-05, + "loss": 1.1575, + "step": 308 + }, + { + "epoch": 0.57, + "grad_norm": 0.2088315773384112, + "learning_rate": 7.810499017832526e-05, + "loss": 1.1316, + "step": 309 + }, + { + "epoch": 0.57, + "grad_norm": 0.2095238410730976, + "learning_rate": 7.807897394883203e-05, + "loss": 1.2087, + "step": 310 + }, + { + "epoch": 0.57, + "grad_norm": 0.22672932127256515, + "learning_rate": 7.805278474174499e-05, + "loss": 1.2512, + "step": 311 + }, + { + "epoch": 0.57, + "grad_norm": 0.21873052340922736, + "learning_rate": 7.802642267603126e-05, + "loss": 1.1909, + "step": 312 + }, + { + "epoch": 0.57, + "grad_norm": 0.219814521916342, + "learning_rate": 7.79998878714432e-05, + "loss": 1.1669, + "step": 313 + }, + { + "epoch": 0.57, + "grad_norm": 0.3049426027257317, + "learning_rate": 7.797318044851786e-05, + "loss": 1.1797, + "step": 314 + }, + { + "epoch": 0.58, + "grad_norm": 0.22309435690065985, + "learning_rate": 7.794630052857638e-05, + "loss": 1.1417, + "step": 315 + }, + { + "epoch": 0.58, + "grad_norm": 0.3891885169154885, + "learning_rate": 7.791924823372354e-05, + "loss": 1.2369, + "step": 316 + }, + { + "epoch": 0.58, + "grad_norm": 0.24780269452456372, + "learning_rate": 7.789202368684711e-05, + "loss": 1.2521, + "step": 317 + }, + { + "epoch": 0.58, + "grad_norm": 0.21660460720269362, + "learning_rate": 7.786462701161738e-05, + "loss": 1.2151, + "step": 318 + }, + { + "epoch": 0.58, + "grad_norm": 0.23635409466561857, + "learning_rate": 7.783705833248649e-05, + "loss": 1.2363, + "step": 319 + }, + { + "epoch": 0.59, + "grad_norm": 0.2616135839903218, + "learning_rate": 7.780931777468797e-05, + "loss": 1.2428, + "step": 320 + }, + { + "epoch": 0.59, + "grad_norm": 0.21461059159245083, + "learning_rate": 7.77814054642361e-05, + "loss": 1.1434, + "step": 321 + }, + { + "epoch": 0.59, + "grad_norm": 0.25348824286656163, + "learning_rate": 7.775332152792539e-05, + "loss": 1.2368, + "step": 322 + }, + { + "epoch": 0.59, + "grad_norm": 0.22275034726331247, + "learning_rate": 7.772506609332995e-05, + "loss": 1.1827, + "step": 323 + }, + { + "epoch": 0.59, + "grad_norm": 0.25030821228147526, + "learning_rate": 7.769663928880298e-05, + "loss": 1.2428, + "step": 324 + }, + { + "epoch": 0.59, + "grad_norm": 0.22251804398745534, + "learning_rate": 7.766804124347608e-05, + "loss": 1.1889, + "step": 325 + }, + { + "epoch": 0.6, + "grad_norm": 0.23381455520411995, + "learning_rate": 7.763927208725879e-05, + "loss": 1.2115, + "step": 326 + }, + { + "epoch": 0.6, + "grad_norm": 0.27341902651946226, + "learning_rate": 7.761033195083791e-05, + "loss": 1.2535, + "step": 327 + }, + { + "epoch": 0.6, + "grad_norm": 0.24862471659814522, + "learning_rate": 7.758122096567694e-05, + "loss": 1.2128, + "step": 328 + }, + { + "epoch": 0.6, + "grad_norm": 0.2251357082045494, + "learning_rate": 7.755193926401547e-05, + "loss": 1.2334, + "step": 329 + }, + { + "epoch": 0.6, + "grad_norm": 0.3173274941622932, + "learning_rate": 7.752248697886857e-05, + "loss": 1.226, + "step": 330 + }, + { + "epoch": 0.61, + "grad_norm": 0.23056440717672175, + "learning_rate": 7.74928642440263e-05, + "loss": 1.2339, + "step": 331 + }, + { + "epoch": 0.61, + "grad_norm": 0.2801507500859342, + "learning_rate": 7.746307119405286e-05, + "loss": 1.287, + "step": 332 + }, + { + "epoch": 0.61, + "grad_norm": 0.2267818430426272, + "learning_rate": 7.743310796428622e-05, + "loss": 1.1916, + "step": 333 + }, + { + "epoch": 0.61, + "grad_norm": 0.2777329160365585, + "learning_rate": 7.74029746908374e-05, + "loss": 1.252, + "step": 334 + }, + { + "epoch": 0.61, + "grad_norm": 0.25289169762353, + "learning_rate": 7.737267151058983e-05, + "loss": 1.2153, + "step": 335 + }, + { + "epoch": 0.61, + "grad_norm": 0.2424670686901653, + "learning_rate": 7.734219856119875e-05, + "loss": 1.2227, + "step": 336 + }, + { + "epoch": 0.62, + "grad_norm": 0.22747092217441645, + "learning_rate": 7.731155598109067e-05, + "loss": 1.19, + "step": 337 + }, + { + "epoch": 0.62, + "grad_norm": 0.2307810940100189, + "learning_rate": 7.728074390946257e-05, + "loss": 1.1818, + "step": 338 + }, + { + "epoch": 0.62, + "grad_norm": 0.2583402574655623, + "learning_rate": 7.724976248628142e-05, + "loss": 1.1608, + "step": 339 + }, + { + "epoch": 0.62, + "grad_norm": 0.22140209760890694, + "learning_rate": 7.721861185228347e-05, + "loss": 1.1245, + "step": 340 + }, + { + "epoch": 0.62, + "grad_norm": 0.25859310758244686, + "learning_rate": 7.718729214897362e-05, + "loss": 1.2247, + "step": 341 + }, + { + "epoch": 0.63, + "grad_norm": 0.26371179531372124, + "learning_rate": 7.715580351862482e-05, + "loss": 1.2128, + "step": 342 + }, + { + "epoch": 0.63, + "grad_norm": 0.26575541302851047, + "learning_rate": 7.712414610427733e-05, + "loss": 1.2443, + "step": 343 + }, + { + "epoch": 0.63, + "grad_norm": 0.269978305197599, + "learning_rate": 7.709232004973816e-05, + "loss": 1.2231, + "step": 344 + }, + { + "epoch": 0.63, + "grad_norm": 0.26583998705977047, + "learning_rate": 7.70603254995804e-05, + "loss": 1.2476, + "step": 345 + }, + { + "epoch": 0.63, + "grad_norm": 0.24256062164066097, + "learning_rate": 7.702816259914253e-05, + "loss": 1.2901, + "step": 346 + }, + { + "epoch": 0.63, + "grad_norm": 0.3463123472658915, + "learning_rate": 7.699583149452779e-05, + "loss": 1.3277, + "step": 347 + }, + { + "epoch": 0.64, + "grad_norm": 0.2269096590531878, + "learning_rate": 7.696333233260345e-05, + "loss": 1.2047, + "step": 348 + }, + { + "epoch": 0.64, + "grad_norm": 0.25136883001050025, + "learning_rate": 7.693066526100031e-05, + "loss": 1.1619, + "step": 349 + }, + { + "epoch": 0.64, + "grad_norm": 0.2565112571116145, + "learning_rate": 7.68978304281118e-05, + "loss": 1.2389, + "step": 350 + }, + { + "epoch": 0.64, + "grad_norm": 0.22175779550828703, + "learning_rate": 7.686482798309349e-05, + "loss": 1.2238, + "step": 351 + }, + { + "epoch": 0.64, + "grad_norm": 0.22588304332216555, + "learning_rate": 7.683165807586234e-05, + "loss": 1.174, + "step": 352 + }, + { + "epoch": 0.65, + "grad_norm": 0.24889474296529737, + "learning_rate": 7.6798320857096e-05, + "loss": 1.2366, + "step": 353 + }, + { + "epoch": 0.65, + "grad_norm": 0.27339703806525034, + "learning_rate": 7.676481647823214e-05, + "loss": 1.2356, + "step": 354 + }, + { + "epoch": 0.65, + "grad_norm": 0.23424666722888365, + "learning_rate": 7.673114509146782e-05, + "loss": 1.2089, + "step": 355 + }, + { + "epoch": 0.65, + "grad_norm": 0.27978285392461766, + "learning_rate": 7.66973068497587e-05, + "loss": 1.2609, + "step": 356 + }, + { + "epoch": 0.65, + "grad_norm": 0.2509423350138824, + "learning_rate": 7.666330190681844e-05, + "loss": 1.1777, + "step": 357 + }, + { + "epoch": 0.65, + "grad_norm": 0.23007730927468031, + "learning_rate": 7.662913041711793e-05, + "loss": 1.154, + "step": 358 + }, + { + "epoch": 0.66, + "grad_norm": 0.2438648674953112, + "learning_rate": 7.659479253588462e-05, + "loss": 1.2257, + "step": 359 + }, + { + "epoch": 0.66, + "grad_norm": 0.28816093242092233, + "learning_rate": 7.65602884191018e-05, + "loss": 1.2558, + "step": 360 + }, + { + "epoch": 0.66, + "grad_norm": 0.24972815300596035, + "learning_rate": 7.652561822350793e-05, + "loss": 1.2837, + "step": 361 + }, + { + "epoch": 0.66, + "grad_norm": 0.2543189139697063, + "learning_rate": 7.649078210659587e-05, + "loss": 1.2193, + "step": 362 + }, + { + "epoch": 0.66, + "grad_norm": 0.2237937956718952, + "learning_rate": 7.645578022661224e-05, + "loss": 1.2237, + "step": 363 + }, + { + "epoch": 0.67, + "grad_norm": 0.29742029408787396, + "learning_rate": 7.642061274255657e-05, + "loss": 1.2116, + "step": 364 + }, + { + "epoch": 0.67, + "grad_norm": 0.2462883147335493, + "learning_rate": 7.638527981418075e-05, + "loss": 1.1827, + "step": 365 + }, + { + "epoch": 0.67, + "grad_norm": 0.2647802498907096, + "learning_rate": 7.634978160198817e-05, + "loss": 1.2739, + "step": 366 + }, + { + "epoch": 0.67, + "grad_norm": 0.22360398779217264, + "learning_rate": 7.631411826723306e-05, + "loss": 1.2185, + "step": 367 + }, + { + "epoch": 0.67, + "grad_norm": 0.2635048004593543, + "learning_rate": 7.627828997191973e-05, + "loss": 1.2317, + "step": 368 + }, + { + "epoch": 0.67, + "grad_norm": 0.2764803449917684, + "learning_rate": 7.624229687880184e-05, + "loss": 1.1923, + "step": 369 + }, + { + "epoch": 0.68, + "grad_norm": 0.25724943233414527, + "learning_rate": 7.620613915138166e-05, + "loss": 1.2218, + "step": 370 + }, + { + "epoch": 0.68, + "grad_norm": 0.2858318045794755, + "learning_rate": 7.61698169539093e-05, + "loss": 1.1496, + "step": 371 + }, + { + "epoch": 0.68, + "grad_norm": 0.23547216647460364, + "learning_rate": 7.613333045138206e-05, + "loss": 1.1905, + "step": 372 + }, + { + "epoch": 0.68, + "grad_norm": 0.22984814903684375, + "learning_rate": 7.609667980954355e-05, + "loss": 1.2009, + "step": 373 + }, + { + "epoch": 0.68, + "grad_norm": 0.2551903754079084, + "learning_rate": 7.605986519488301e-05, + "loss": 1.2042, + "step": 374 + }, + { + "epoch": 0.69, + "grad_norm": 0.2508257410125616, + "learning_rate": 7.602288677463457e-05, + "loss": 1.2468, + "step": 375 + }, + { + "epoch": 0.69, + "grad_norm": 0.25324577774935964, + "learning_rate": 7.598574471677644e-05, + "loss": 1.2603, + "step": 376 + }, + { + "epoch": 0.69, + "grad_norm": 0.35888776531769967, + "learning_rate": 7.59484391900302e-05, + "loss": 1.1929, + "step": 377 + }, + { + "epoch": 0.69, + "grad_norm": 0.22048517191014724, + "learning_rate": 7.591097036385994e-05, + "loss": 1.1783, + "step": 378 + }, + { + "epoch": 0.69, + "grad_norm": 0.2781160412746083, + "learning_rate": 7.587333840847162e-05, + "loss": 1.3397, + "step": 379 + }, + { + "epoch": 0.7, + "grad_norm": 0.24033046830332258, + "learning_rate": 7.583554349481222e-05, + "loss": 1.2436, + "step": 380 + }, + { + "epoch": 0.7, + "grad_norm": 0.26413762380260003, + "learning_rate": 7.579758579456893e-05, + "loss": 1.1917, + "step": 381 + }, + { + "epoch": 0.7, + "grad_norm": 0.2390937887338632, + "learning_rate": 7.575946548016847e-05, + "loss": 1.2186, + "step": 382 + }, + { + "epoch": 0.7, + "grad_norm": 0.25131263043429275, + "learning_rate": 7.572118272477622e-05, + "loss": 1.2538, + "step": 383 + }, + { + "epoch": 0.7, + "grad_norm": 0.223974104870702, + "learning_rate": 7.568273770229546e-05, + "loss": 1.2165, + "step": 384 + }, + { + "epoch": 0.7, + "grad_norm": 0.25840356830252875, + "learning_rate": 7.564413058736663e-05, + "loss": 1.1848, + "step": 385 + }, + { + "epoch": 0.71, + "grad_norm": 0.2723156683076603, + "learning_rate": 7.560536155536641e-05, + "loss": 1.1982, + "step": 386 + }, + { + "epoch": 0.71, + "grad_norm": 0.265687427976889, + "learning_rate": 7.556643078240708e-05, + "loss": 1.231, + "step": 387 + }, + { + "epoch": 0.71, + "grad_norm": 0.25152762080976077, + "learning_rate": 7.552733844533562e-05, + "loss": 1.1974, + "step": 388 + }, + { + "epoch": 0.71, + "grad_norm": 0.2366049485053541, + "learning_rate": 7.548808472173292e-05, + "loss": 1.3119, + "step": 389 + }, + { + "epoch": 0.71, + "grad_norm": 0.22092196577077122, + "learning_rate": 7.5448669789913e-05, + "loss": 1.195, + "step": 390 + }, + { + "epoch": 0.72, + "grad_norm": 0.22667521540462374, + "learning_rate": 7.540909382892217e-05, + "loss": 1.1431, + "step": 391 + }, + { + "epoch": 0.72, + "grad_norm": 0.25432207282646513, + "learning_rate": 7.536935701853823e-05, + "loss": 1.2173, + "step": 392 + }, + { + "epoch": 0.72, + "grad_norm": 0.29950506457923864, + "learning_rate": 7.53294595392697e-05, + "loss": 1.1962, + "step": 393 + }, + { + "epoch": 0.72, + "grad_norm": 0.24735689607229913, + "learning_rate": 7.528940157235487e-05, + "loss": 1.2053, + "step": 394 + }, + { + "epoch": 0.72, + "grad_norm": 0.24394198607459663, + "learning_rate": 7.524918329976114e-05, + "loss": 1.1979, + "step": 395 + }, + { + "epoch": 0.72, + "grad_norm": 0.2630369372689188, + "learning_rate": 7.520880490418409e-05, + "loss": 1.2111, + "step": 396 + }, + { + "epoch": 0.73, + "grad_norm": 0.26275028416291457, + "learning_rate": 7.516826656904664e-05, + "loss": 1.2133, + "step": 397 + }, + { + "epoch": 0.73, + "grad_norm": 0.23938074620956928, + "learning_rate": 7.512756847849831e-05, + "loss": 1.1355, + "step": 398 + }, + { + "epoch": 0.73, + "grad_norm": 0.3724960610098138, + "learning_rate": 7.508671081741428e-05, + "loss": 1.2572, + "step": 399 + }, + { + "epoch": 0.73, + "grad_norm": 0.24161685847894723, + "learning_rate": 7.504569377139462e-05, + "loss": 1.1706, + "step": 400 + }, + { + "epoch": 0.73, + "grad_norm": 0.26121591322670523, + "learning_rate": 7.50045175267634e-05, + "loss": 1.2135, + "step": 401 + }, + { + "epoch": 0.74, + "grad_norm": 0.2465579498164775, + "learning_rate": 7.496318227056788e-05, + "loss": 1.1641, + "step": 402 + }, + { + "epoch": 0.74, + "grad_norm": 0.2556288696122787, + "learning_rate": 7.492168819057767e-05, + "loss": 1.2939, + "step": 403 + }, + { + "epoch": 0.74, + "grad_norm": 0.261481216336303, + "learning_rate": 7.488003547528382e-05, + "loss": 1.2026, + "step": 404 + }, + { + "epoch": 0.74, + "grad_norm": 0.2389415135676362, + "learning_rate": 7.483822431389799e-05, + "loss": 1.2131, + "step": 405 + }, + { + "epoch": 0.74, + "grad_norm": 0.2559201956627192, + "learning_rate": 7.479625489635162e-05, + "loss": 1.1246, + "step": 406 + }, + { + "epoch": 0.74, + "grad_norm": 0.27127932491822604, + "learning_rate": 7.475412741329504e-05, + "loss": 1.2429, + "step": 407 + }, + { + "epoch": 0.75, + "grad_norm": 0.27006004008695594, + "learning_rate": 7.47118420560966e-05, + "loss": 1.2388, + "step": 408 + }, + { + "epoch": 0.75, + "grad_norm": 0.23716823297200537, + "learning_rate": 7.466939901684182e-05, + "loss": 1.1264, + "step": 409 + }, + { + "epoch": 0.75, + "grad_norm": 0.2885373898669248, + "learning_rate": 7.462679848833252e-05, + "loss": 1.2786, + "step": 410 + }, + { + "epoch": 0.75, + "grad_norm": 0.49215227598639927, + "learning_rate": 7.458404066408588e-05, + "loss": 1.2386, + "step": 411 + }, + { + "epoch": 0.75, + "grad_norm": 0.24235735604947403, + "learning_rate": 7.454112573833368e-05, + "loss": 1.1423, + "step": 412 + }, + { + "epoch": 0.76, + "grad_norm": 0.2584614748054343, + "learning_rate": 7.449805390602127e-05, + "loss": 1.2669, + "step": 413 + }, + { + "epoch": 0.76, + "grad_norm": 0.23806123085998873, + "learning_rate": 7.445482536280684e-05, + "loss": 1.1763, + "step": 414 + }, + { + "epoch": 0.76, + "grad_norm": 0.24459517607786851, + "learning_rate": 7.441144030506043e-05, + "loss": 1.198, + "step": 415 + }, + { + "epoch": 0.76, + "grad_norm": 0.25801616402700395, + "learning_rate": 7.436789892986304e-05, + "loss": 1.2136, + "step": 416 + }, + { + "epoch": 0.76, + "grad_norm": 0.2814819942392514, + "learning_rate": 7.432420143500578e-05, + "loss": 1.2398, + "step": 417 + }, + { + "epoch": 0.76, + "grad_norm": 0.22134709322606153, + "learning_rate": 7.428034801898893e-05, + "loss": 1.1592, + "step": 418 + }, + { + "epoch": 0.77, + "grad_norm": 0.2899677536995633, + "learning_rate": 7.42363388810211e-05, + "loss": 1.2296, + "step": 419 + }, + { + "epoch": 0.77, + "grad_norm": 0.24005943230262294, + "learning_rate": 7.419217422101822e-05, + "loss": 1.2223, + "step": 420 + }, + { + "epoch": 0.77, + "grad_norm": 0.26417562369496167, + "learning_rate": 7.414785423960275e-05, + "loss": 1.2261, + "step": 421 + }, + { + "epoch": 0.77, + "grad_norm": 0.2580815883535521, + "learning_rate": 7.410337913810271e-05, + "loss": 1.2021, + "step": 422 + }, + { + "epoch": 0.77, + "grad_norm": 0.25242217589496435, + "learning_rate": 7.405874911855071e-05, + "loss": 1.239, + "step": 423 + }, + { + "epoch": 0.78, + "grad_norm": 0.21991733999839932, + "learning_rate": 7.401396438368315e-05, + "loss": 1.1716, + "step": 424 + }, + { + "epoch": 0.78, + "grad_norm": 0.40116538322720213, + "learning_rate": 7.396902513693924e-05, + "loss": 1.2773, + "step": 425 + }, + { + "epoch": 0.78, + "grad_norm": 0.277333939455099, + "learning_rate": 7.392393158246002e-05, + "loss": 1.2574, + "step": 426 + }, + { + "epoch": 0.78, + "grad_norm": 0.27146087746385755, + "learning_rate": 7.387868392508756e-05, + "loss": 1.2243, + "step": 427 + }, + { + "epoch": 0.78, + "grad_norm": 0.255881055620786, + "learning_rate": 7.38332823703639e-05, + "loss": 1.223, + "step": 428 + }, + { + "epoch": 0.78, + "grad_norm": 0.24807364856677255, + "learning_rate": 7.378772712453021e-05, + "loss": 1.1985, + "step": 429 + }, + { + "epoch": 0.79, + "grad_norm": 0.25746257617764423, + "learning_rate": 7.37420183945258e-05, + "loss": 1.2502, + "step": 430 + }, + { + "epoch": 0.79, + "grad_norm": 0.28851991049982234, + "learning_rate": 7.369615638798722e-05, + "loss": 1.2535, + "step": 431 + }, + { + "epoch": 0.79, + "grad_norm": 0.24113389811604363, + "learning_rate": 7.365014131324725e-05, + "loss": 1.2227, + "step": 432 + }, + { + "epoch": 0.79, + "grad_norm": 0.2414465151257969, + "learning_rate": 7.360397337933405e-05, + "loss": 1.1884, + "step": 433 + }, + { + "epoch": 0.79, + "grad_norm": 0.2735463134699831, + "learning_rate": 7.355765279597011e-05, + "loss": 1.2756, + "step": 434 + }, + { + "epoch": 0.8, + "grad_norm": 0.2588437452987293, + "learning_rate": 7.351117977357139e-05, + "loss": 1.2108, + "step": 435 + }, + { + "epoch": 0.8, + "grad_norm": 0.26573294117796553, + "learning_rate": 7.346455452324629e-05, + "loss": 1.1821, + "step": 436 + }, + { + "epoch": 0.8, + "grad_norm": 0.2555476577827304, + "learning_rate": 7.341777725679473e-05, + "loss": 1.1937, + "step": 437 + }, + { + "epoch": 0.8, + "grad_norm": 0.2867704132108098, + "learning_rate": 7.337084818670716e-05, + "loss": 1.2272, + "step": 438 + }, + { + "epoch": 0.8, + "grad_norm": 0.27726678115981157, + "learning_rate": 7.332376752616367e-05, + "loss": 1.2331, + "step": 439 + }, + { + "epoch": 0.8, + "grad_norm": 0.26955338021079955, + "learning_rate": 7.32765354890329e-05, + "loss": 1.1731, + "step": 440 + }, + { + "epoch": 0.81, + "grad_norm": 0.25250321202536524, + "learning_rate": 7.322915228987116e-05, + "loss": 1.2653, + "step": 441 + }, + { + "epoch": 0.81, + "grad_norm": 0.24748844179765395, + "learning_rate": 7.318161814392143e-05, + "loss": 1.24, + "step": 442 + }, + { + "epoch": 0.81, + "grad_norm": 0.28177805247356325, + "learning_rate": 7.313393326711239e-05, + "loss": 1.185, + "step": 443 + }, + { + "epoch": 0.81, + "grad_norm": 0.24093242000396312, + "learning_rate": 7.30860978760574e-05, + "loss": 1.1994, + "step": 444 + }, + { + "epoch": 0.81, + "grad_norm": 0.26277803901457075, + "learning_rate": 7.30381121880536e-05, + "loss": 1.212, + "step": 445 + }, + { + "epoch": 0.82, + "grad_norm": 0.2506524258682433, + "learning_rate": 7.298997642108079e-05, + "loss": 1.2421, + "step": 446 + }, + { + "epoch": 0.82, + "grad_norm": 0.2840599700015824, + "learning_rate": 7.294169079380061e-05, + "loss": 1.1818, + "step": 447 + }, + { + "epoch": 0.82, + "grad_norm": 0.24892184038117549, + "learning_rate": 7.289325552555538e-05, + "loss": 1.1916, + "step": 448 + }, + { + "epoch": 0.82, + "grad_norm": 0.2700898428541357, + "learning_rate": 7.284467083636722e-05, + "loss": 1.2517, + "step": 449 + }, + { + "epoch": 0.82, + "grad_norm": 0.2617848546539419, + "learning_rate": 7.279593694693698e-05, + "loss": 1.2063, + "step": 450 + }, + { + "epoch": 0.82, + "grad_norm": 0.2698278585334131, + "learning_rate": 7.274705407864332e-05, + "loss": 1.194, + "step": 451 + }, + { + "epoch": 0.83, + "grad_norm": 0.23678313024953834, + "learning_rate": 7.26980224535416e-05, + "loss": 1.2349, + "step": 452 + }, + { + "epoch": 0.83, + "grad_norm": 0.24851875792002978, + "learning_rate": 7.264884229436293e-05, + "loss": 1.1758, + "step": 453 + }, + { + "epoch": 0.83, + "grad_norm": 0.24122080121681125, + "learning_rate": 7.259951382451318e-05, + "loss": 1.1962, + "step": 454 + }, + { + "epoch": 0.83, + "grad_norm": 0.22741322959884405, + "learning_rate": 7.25500372680719e-05, + "loss": 1.1702, + "step": 455 + }, + { + "epoch": 0.83, + "grad_norm": 0.2297475610861458, + "learning_rate": 7.250041284979137e-05, + "loss": 1.1466, + "step": 456 + }, + { + "epoch": 0.84, + "grad_norm": 0.3057605989721467, + "learning_rate": 7.245064079509553e-05, + "loss": 1.246, + "step": 457 + }, + { + "epoch": 0.84, + "grad_norm": 0.2719638501597136, + "learning_rate": 7.240072133007899e-05, + "loss": 1.2184, + "step": 458 + }, + { + "epoch": 0.84, + "grad_norm": 0.2436807816414479, + "learning_rate": 7.235065468150593e-05, + "loss": 1.2324, + "step": 459 + }, + { + "epoch": 0.84, + "grad_norm": 0.23436349430255515, + "learning_rate": 7.23004410768092e-05, + "loss": 1.1813, + "step": 460 + }, + { + "epoch": 0.84, + "grad_norm": 0.2398940990211377, + "learning_rate": 7.22500807440892e-05, + "loss": 1.1924, + "step": 461 + }, + { + "epoch": 0.84, + "grad_norm": 0.2605716625062531, + "learning_rate": 7.219957391211281e-05, + "loss": 1.182, + "step": 462 + }, + { + "epoch": 0.85, + "grad_norm": 0.260462524570941, + "learning_rate": 7.214892081031244e-05, + "loss": 1.2136, + "step": 463 + }, + { + "epoch": 0.85, + "grad_norm": 0.21979766512306334, + "learning_rate": 7.209812166878491e-05, + "loss": 1.2066, + "step": 464 + }, + { + "epoch": 0.85, + "grad_norm": 0.23324453647530663, + "learning_rate": 7.204717671829051e-05, + "loss": 1.1657, + "step": 465 + }, + { + "epoch": 0.85, + "grad_norm": 0.2529434935507481, + "learning_rate": 7.199608619025177e-05, + "loss": 1.2093, + "step": 466 + }, + { + "epoch": 0.85, + "grad_norm": 0.25371701891720116, + "learning_rate": 7.194485031675265e-05, + "loss": 1.2225, + "step": 467 + }, + { + "epoch": 0.86, + "grad_norm": 0.23272423066292103, + "learning_rate": 7.189346933053725e-05, + "loss": 1.1721, + "step": 468 + }, + { + "epoch": 0.86, + "grad_norm": 0.25122928735587546, + "learning_rate": 7.184194346500892e-05, + "loss": 1.2537, + "step": 469 + }, + { + "epoch": 0.86, + "grad_norm": 0.2159270875490409, + "learning_rate": 7.179027295422913e-05, + "loss": 1.197, + "step": 470 + }, + { + "epoch": 0.86, + "grad_norm": 0.2633111059076544, + "learning_rate": 7.173845803291636e-05, + "loss": 1.1721, + "step": 471 + }, + { + "epoch": 0.86, + "grad_norm": 0.30555936322098703, + "learning_rate": 7.168649893644517e-05, + "loss": 1.3011, + "step": 472 + }, + { + "epoch": 0.87, + "grad_norm": 0.23492670111453726, + "learning_rate": 7.163439590084502e-05, + "loss": 1.1601, + "step": 473 + }, + { + "epoch": 0.87, + "grad_norm": 0.26602734263721806, + "learning_rate": 7.158214916279923e-05, + "loss": 1.2808, + "step": 474 + }, + { + "epoch": 0.87, + "grad_norm": 0.3182695007856262, + "learning_rate": 7.152975895964386e-05, + "loss": 1.2967, + "step": 475 + }, + { + "epoch": 0.87, + "grad_norm": 0.2785021674736721, + "learning_rate": 7.147722552936673e-05, + "loss": 1.1789, + "step": 476 + }, + { + "epoch": 0.87, + "grad_norm": 0.279474303138652, + "learning_rate": 7.142454911060627e-05, + "loss": 1.2596, + "step": 477 + }, + { + "epoch": 0.87, + "grad_norm": 0.2556980144910755, + "learning_rate": 7.137172994265044e-05, + "loss": 1.2426, + "step": 478 + }, + { + "epoch": 0.88, + "grad_norm": 0.3311256331993533, + "learning_rate": 7.131876826543565e-05, + "loss": 1.2059, + "step": 479 + }, + { + "epoch": 0.88, + "grad_norm": 0.26467296197775253, + "learning_rate": 7.12656643195457e-05, + "loss": 1.2482, + "step": 480 + }, + { + "epoch": 0.88, + "grad_norm": 0.27444885274652553, + "learning_rate": 7.121241834621064e-05, + "loss": 1.2528, + "step": 481 + }, + { + "epoch": 0.88, + "grad_norm": 0.2572283861115396, + "learning_rate": 7.115903058730567e-05, + "loss": 1.1849, + "step": 482 + }, + { + "epoch": 0.88, + "grad_norm": 0.2677065778235683, + "learning_rate": 7.11055012853501e-05, + "loss": 1.2011, + "step": 483 + }, + { + "epoch": 0.89, + "grad_norm": 0.29470622036742816, + "learning_rate": 7.105183068350619e-05, + "loss": 1.2398, + "step": 484 + }, + { + "epoch": 0.89, + "grad_norm": 0.27609230248969197, + "learning_rate": 7.099801902557811e-05, + "loss": 1.2259, + "step": 485 + }, + { + "epoch": 0.89, + "grad_norm": 0.24248634168099284, + "learning_rate": 7.094406655601073e-05, + "loss": 1.2282, + "step": 486 + }, + { + "epoch": 0.89, + "grad_norm": 0.2765941767688746, + "learning_rate": 7.088997351988865e-05, + "loss": 1.2319, + "step": 487 + }, + { + "epoch": 0.89, + "grad_norm": 0.29347776909858947, + "learning_rate": 7.083574016293493e-05, + "loss": 1.1765, + "step": 488 + }, + { + "epoch": 0.89, + "grad_norm": 0.285370295424537, + "learning_rate": 7.078136673151008e-05, + "loss": 1.26, + "step": 489 + }, + { + "epoch": 0.9, + "grad_norm": 0.29408734903836536, + "learning_rate": 7.072685347261093e-05, + "loss": 1.226, + "step": 490 + }, + { + "epoch": 0.9, + "grad_norm": 0.27437470239205813, + "learning_rate": 7.067220063386947e-05, + "loss": 1.1976, + "step": 491 + }, + { + "epoch": 0.9, + "grad_norm": 0.2680770258777871, + "learning_rate": 7.061740846355176e-05, + "loss": 1.1915, + "step": 492 + }, + { + "epoch": 0.9, + "grad_norm": 0.27200362879502954, + "learning_rate": 7.056247721055678e-05, + "loss": 1.2002, + "step": 493 + }, + { + "epoch": 0.9, + "grad_norm": 0.2637811092577037, + "learning_rate": 7.050740712441528e-05, + "loss": 1.287, + "step": 494 + }, + { + "epoch": 0.91, + "grad_norm": 0.24657959209271266, + "learning_rate": 7.045219845528875e-05, + "loss": 1.2284, + "step": 495 + }, + { + "epoch": 0.91, + "grad_norm": 0.25311992110358666, + "learning_rate": 7.039685145396812e-05, + "loss": 1.1616, + "step": 496 + }, + { + "epoch": 0.91, + "grad_norm": 0.2564633694193358, + "learning_rate": 7.034136637187275e-05, + "loss": 1.2067, + "step": 497 + }, + { + "epoch": 0.91, + "grad_norm": 0.2446797651174144, + "learning_rate": 7.028574346104926e-05, + "loss": 1.2284, + "step": 498 + }, + { + "epoch": 0.91, + "grad_norm": 0.2592751463399255, + "learning_rate": 7.022998297417034e-05, + "loss": 1.2371, + "step": 499 + }, + { + "epoch": 0.91, + "grad_norm": 0.2500713943206808, + "learning_rate": 7.017408516453365e-05, + "loss": 1.1061, + "step": 500 + }, + { + "epoch": 0.92, + "grad_norm": 0.2812266276040743, + "learning_rate": 7.011805028606064e-05, + "loss": 1.1949, + "step": 501 + }, + { + "epoch": 0.92, + "grad_norm": 0.298829667668083, + "learning_rate": 7.006187859329544e-05, + "loss": 1.2313, + "step": 502 + }, + { + "epoch": 0.92, + "grad_norm": 0.26518768159745104, + "learning_rate": 7.000557034140361e-05, + "loss": 1.2246, + "step": 503 + }, + { + "epoch": 0.92, + "grad_norm": 0.3037280360760458, + "learning_rate": 6.994912578617113e-05, + "loss": 1.1617, + "step": 504 + }, + { + "epoch": 0.92, + "grad_norm": 0.2726903109255714, + "learning_rate": 6.989254518400309e-05, + "loss": 1.2415, + "step": 505 + }, + { + "epoch": 0.93, + "grad_norm": 0.25568082003046966, + "learning_rate": 6.98358287919226e-05, + "loss": 1.1817, + "step": 506 + }, + { + "epoch": 0.93, + "grad_norm": 0.25633294893705044, + "learning_rate": 6.97789768675696e-05, + "loss": 1.2149, + "step": 507 + }, + { + "epoch": 0.93, + "grad_norm": 0.28291439435087123, + "learning_rate": 6.972198966919972e-05, + "loss": 1.1578, + "step": 508 + }, + { + "epoch": 0.93, + "grad_norm": 0.27195184756655516, + "learning_rate": 6.966486745568308e-05, + "loss": 1.2355, + "step": 509 + }, + { + "epoch": 0.93, + "grad_norm": 0.239159568376005, + "learning_rate": 6.960761048650312e-05, + "loss": 1.1688, + "step": 510 + }, + { + "epoch": 0.93, + "grad_norm": 0.22961475425949177, + "learning_rate": 6.955021902175543e-05, + "loss": 1.2094, + "step": 511 + }, + { + "epoch": 0.94, + "grad_norm": 0.27443773600741117, + "learning_rate": 6.949269332214651e-05, + "loss": 1.2559, + "step": 512 + }, + { + "epoch": 0.94, + "grad_norm": 0.26230551832002097, + "learning_rate": 6.94350336489927e-05, + "loss": 1.2121, + "step": 513 + }, + { + "epoch": 0.94, + "grad_norm": 0.2716742985303849, + "learning_rate": 6.937724026421892e-05, + "loss": 1.2444, + "step": 514 + }, + { + "epoch": 0.94, + "grad_norm": 0.2537850139439542, + "learning_rate": 6.931931343035742e-05, + "loss": 1.1327, + "step": 515 + }, + { + "epoch": 0.94, + "grad_norm": 0.28599587967496826, + "learning_rate": 6.926125341054676e-05, + "loss": 1.2236, + "step": 516 + }, + { + "epoch": 0.95, + "grad_norm": 0.26780654378470103, + "learning_rate": 6.920306046853043e-05, + "loss": 1.2295, + "step": 517 + }, + { + "epoch": 0.95, + "grad_norm": 0.23606296888412015, + "learning_rate": 6.914473486865577e-05, + "loss": 1.1543, + "step": 518 + }, + { + "epoch": 0.95, + "grad_norm": 0.34976881174240837, + "learning_rate": 6.90862768758727e-05, + "loss": 1.2067, + "step": 519 + }, + { + "epoch": 0.95, + "grad_norm": 0.2481257873494882, + "learning_rate": 6.902768675573258e-05, + "loss": 1.2188, + "step": 520 + }, + { + "epoch": 0.95, + "grad_norm": 0.2996395778117021, + "learning_rate": 6.896896477438699e-05, + "loss": 1.2326, + "step": 521 + }, + { + "epoch": 0.95, + "grad_norm": 0.8839768816333193, + "learning_rate": 6.891011119858643e-05, + "loss": 1.2435, + "step": 522 + }, + { + "epoch": 0.96, + "grad_norm": 0.2851882482058998, + "learning_rate": 6.885112629567927e-05, + "loss": 1.2644, + "step": 523 + }, + { + "epoch": 0.96, + "grad_norm": 0.2813663482913699, + "learning_rate": 6.879201033361035e-05, + "loss": 1.2309, + "step": 524 + }, + { + "epoch": 0.96, + "grad_norm": 0.3257551560135454, + "learning_rate": 6.873276358091996e-05, + "loss": 1.2755, + "step": 525 + }, + { + "epoch": 0.96, + "grad_norm": 0.28930479952494365, + "learning_rate": 6.867338630674247e-05, + "loss": 1.1962, + "step": 526 + }, + { + "epoch": 0.96, + "grad_norm": 0.3077462996938649, + "learning_rate": 6.861387878080511e-05, + "loss": 1.2402, + "step": 527 + }, + { + "epoch": 0.97, + "grad_norm": 0.2848900193452761, + "learning_rate": 6.855424127342688e-05, + "loss": 1.2748, + "step": 528 + }, + { + "epoch": 0.97, + "grad_norm": 0.4765938812802202, + "learning_rate": 6.849447405551718e-05, + "loss": 1.2226, + "step": 529 + }, + { + "epoch": 0.97, + "grad_norm": 0.53184473292579, + "learning_rate": 6.843457739857467e-05, + "loss": 1.2347, + "step": 530 + }, + { + "epoch": 0.97, + "grad_norm": 0.6416239346492343, + "learning_rate": 6.837455157468596e-05, + "loss": 1.2429, + "step": 531 + }, + { + "epoch": 0.97, + "grad_norm": 0.3188092712502773, + "learning_rate": 6.831439685652442e-05, + "loss": 1.216, + "step": 532 + }, + { + "epoch": 0.97, + "grad_norm": 0.3527495731006385, + "learning_rate": 6.825411351734895e-05, + "loss": 1.1682, + "step": 533 + }, + { + "epoch": 0.98, + "grad_norm": 0.29603753744741856, + "learning_rate": 6.819370183100274e-05, + "loss": 1.1434, + "step": 534 + }, + { + "epoch": 0.98, + "grad_norm": 0.5252450389976622, + "learning_rate": 6.813316207191198e-05, + "loss": 1.1943, + "step": 535 + }, + { + "epoch": 0.98, + "grad_norm": 0.32999419558659937, + "learning_rate": 6.807249451508466e-05, + "loss": 1.192, + "step": 536 + }, + { + "epoch": 0.98, + "grad_norm": 0.3650175469778724, + "learning_rate": 6.801169943610929e-05, + "loss": 1.2141, + "step": 537 + }, + { + "epoch": 0.98, + "grad_norm": 1.0643532150783557, + "learning_rate": 6.795077711115368e-05, + "loss": 1.2253, + "step": 538 + }, + { + "epoch": 0.99, + "grad_norm": 0.5041310609130145, + "learning_rate": 6.788972781696363e-05, + "loss": 1.278, + "step": 539 + }, + { + "epoch": 0.99, + "grad_norm": 0.5123058164360991, + "learning_rate": 6.782855183086177e-05, + "loss": 1.2231, + "step": 540 + }, + { + "epoch": 0.99, + "grad_norm": 0.3533015702394419, + "learning_rate": 6.776724943074619e-05, + "loss": 1.2072, + "step": 541 + }, + { + "epoch": 0.99, + "grad_norm": 0.30253964625417207, + "learning_rate": 6.770582089508927e-05, + "loss": 1.1382, + "step": 542 + }, + { + "epoch": 0.99, + "grad_norm": 0.348991618828202, + "learning_rate": 6.764426650293633e-05, + "loss": 1.2079, + "step": 543 + }, + { + "epoch": 0.99, + "grad_norm": 0.46017440578788743, + "learning_rate": 6.758258653390444e-05, + "loss": 1.1813, + "step": 544 + }, + { + "epoch": 1.0, + "grad_norm": 0.31962101755594885, + "learning_rate": 6.75207812681811e-05, + "loss": 1.1339, + "step": 545 + }, + { + "epoch": 1.0, + "grad_norm": 0.37092024548285923, + "learning_rate": 6.745885098652298e-05, + "loss": 1.2591, + "step": 546 + }, + { + "epoch": 1.0, + "grad_norm": 0.32347106450715835, + "learning_rate": 6.739679597025466e-05, + "loss": 1.2017, + "step": 547 + }, + { + "epoch": 1.0, + "grad_norm": 0.39250187112342494, + "learning_rate": 6.733461650126733e-05, + "loss": 1.0933, + "step": 548 + }, + { + "epoch": 1.0, + "grad_norm": 0.473522452217324, + "learning_rate": 6.727231286201752e-05, + "loss": 1.1124, + "step": 549 + }, + { + "epoch": 1.01, + "grad_norm": 0.4809062179622052, + "learning_rate": 6.720988533552582e-05, + "loss": 1.1585, + "step": 550 + }, + { + "epoch": 1.01, + "grad_norm": 0.3529662801059162, + "learning_rate": 6.714733420537559e-05, + "loss": 1.0501, + "step": 551 + }, + { + "epoch": 1.01, + "grad_norm": 0.5958247214391118, + "learning_rate": 6.708465975571168e-05, + "loss": 1.1086, + "step": 552 + }, + { + "epoch": 1.01, + "grad_norm": 0.5341364205022454, + "learning_rate": 6.70218622712391e-05, + "loss": 1.0518, + "step": 553 + }, + { + "epoch": 1.01, + "grad_norm": 0.3601805724462006, + "learning_rate": 6.695894203722181e-05, + "loss": 1.1779, + "step": 554 + }, + { + "epoch": 1.02, + "grad_norm": 0.43410190338280613, + "learning_rate": 6.68958993394813e-05, + "loss": 1.093, + "step": 555 + }, + { + "epoch": 1.02, + "grad_norm": 0.46217742572873594, + "learning_rate": 6.683273446439546e-05, + "loss": 1.0117, + "step": 556 + }, + { + "epoch": 1.02, + "grad_norm": 0.8591682373623357, + "learning_rate": 6.676944769889708e-05, + "loss": 1.1002, + "step": 557 + }, + { + "epoch": 1.02, + "grad_norm": 0.7383229487622726, + "learning_rate": 6.670603933047272e-05, + "loss": 1.0779, + "step": 558 + }, + { + "epoch": 1.02, + "grad_norm": 0.5965305891207813, + "learning_rate": 6.664250964716131e-05, + "loss": 1.0889, + "step": 559 + }, + { + "epoch": 1.02, + "grad_norm": 0.6030858606684543, + "learning_rate": 6.657885893755288e-05, + "loss": 1.0982, + "step": 560 + }, + { + "epoch": 1.03, + "grad_norm": 0.4644510682398409, + "learning_rate": 6.65150874907872e-05, + "loss": 1.1004, + "step": 561 + }, + { + "epoch": 1.03, + "grad_norm": 0.43943285132452564, + "learning_rate": 6.645119559655254e-05, + "loss": 1.0536, + "step": 562 + }, + { + "epoch": 1.03, + "grad_norm": 0.4456395978600012, + "learning_rate": 6.638718354508427e-05, + "loss": 1.0733, + "step": 563 + }, + { + "epoch": 1.03, + "grad_norm": 0.3303824433217466, + "learning_rate": 6.632305162716365e-05, + "loss": 1.0552, + "step": 564 + }, + { + "epoch": 1.03, + "grad_norm": 0.3617704823170143, + "learning_rate": 6.62588001341164e-05, + "loss": 1.1092, + "step": 565 + }, + { + "epoch": 1.04, + "grad_norm": 0.4465013349903427, + "learning_rate": 6.619442935781141e-05, + "loss": 1.0781, + "step": 566 + }, + { + "epoch": 1.04, + "grad_norm": 0.48516780613791277, + "learning_rate": 6.612993959065947e-05, + "loss": 1.0686, + "step": 567 + }, + { + "epoch": 1.04, + "grad_norm": 0.38867820318536633, + "learning_rate": 6.606533112561186e-05, + "loss": 1.1215, + "step": 568 + }, + { + "epoch": 1.04, + "grad_norm": 0.38566119820378336, + "learning_rate": 6.600060425615907e-05, + "loss": 1.1213, + "step": 569 + }, + { + "epoch": 1.04, + "grad_norm": 0.35534855445058544, + "learning_rate": 6.593575927632947e-05, + "loss": 1.0955, + "step": 570 + }, + { + "epoch": 1.04, + "grad_norm": 0.38124406233349717, + "learning_rate": 6.587079648068795e-05, + "loss": 1.0659, + "step": 571 + }, + { + "epoch": 1.05, + "grad_norm": 0.454750160923548, + "learning_rate": 6.580571616433457e-05, + "loss": 1.1149, + "step": 572 + }, + { + "epoch": 1.05, + "grad_norm": 0.35353190088025255, + "learning_rate": 6.574051862290325e-05, + "loss": 1.0388, + "step": 573 + }, + { + "epoch": 1.05, + "grad_norm": 0.3249395594793626, + "learning_rate": 6.567520415256045e-05, + "loss": 1.0784, + "step": 574 + }, + { + "epoch": 1.05, + "grad_norm": 0.40078898818247227, + "learning_rate": 6.560977305000375e-05, + "loss": 1.0859, + "step": 575 + }, + { + "epoch": 1.05, + "grad_norm": 0.4115264795060035, + "learning_rate": 6.554422561246054e-05, + "loss": 1.1828, + "step": 576 + }, + { + "epoch": 1.06, + "grad_norm": 0.30090229228069215, + "learning_rate": 6.54785621376867e-05, + "loss": 1.0901, + "step": 577 + }, + { + "epoch": 1.06, + "grad_norm": 0.28827860350299206, + "learning_rate": 6.541278292396523e-05, + "loss": 1.0277, + "step": 578 + }, + { + "epoch": 1.06, + "grad_norm": 0.34690404488996757, + "learning_rate": 6.534688827010484e-05, + "loss": 1.048, + "step": 579 + }, + { + "epoch": 1.06, + "grad_norm": 0.29943113556644785, + "learning_rate": 6.528087847543867e-05, + "loss": 1.0646, + "step": 580 + }, + { + "epoch": 1.06, + "grad_norm": 0.37318202575874415, + "learning_rate": 6.521475383982291e-05, + "loss": 1.1091, + "step": 581 + }, + { + "epoch": 1.06, + "grad_norm": 0.3049663659203959, + "learning_rate": 6.51485146636354e-05, + "loss": 1.0552, + "step": 582 + }, + { + "epoch": 1.07, + "grad_norm": 0.3342407867509692, + "learning_rate": 6.508216124777431e-05, + "loss": 1.2227, + "step": 583 + }, + { + "epoch": 1.07, + "grad_norm": 0.3348396047855952, + "learning_rate": 6.501569389365674e-05, + "loss": 1.0861, + "step": 584 + }, + { + "epoch": 1.07, + "grad_norm": 0.30951429367513383, + "learning_rate": 6.494911290321737e-05, + "loss": 1.0461, + "step": 585 + }, + { + "epoch": 1.07, + "grad_norm": 0.33898401361064606, + "learning_rate": 6.488241857890711e-05, + "loss": 1.0854, + "step": 586 + }, + { + "epoch": 1.07, + "grad_norm": 0.4901462068263497, + "learning_rate": 6.481561122369164e-05, + "loss": 1.1012, + "step": 587 + }, + { + "epoch": 1.08, + "grad_norm": 0.3179574879809652, + "learning_rate": 6.474869114105018e-05, + "loss": 1.0451, + "step": 588 + }, + { + "epoch": 1.08, + "grad_norm": 0.32159328915060714, + "learning_rate": 6.468165863497395e-05, + "loss": 1.0458, + "step": 589 + }, + { + "epoch": 1.08, + "grad_norm": 0.36462235008537297, + "learning_rate": 6.461451400996491e-05, + "loss": 1.1247, + "step": 590 + }, + { + "epoch": 1.08, + "grad_norm": 0.5373862753611778, + "learning_rate": 6.454725757103432e-05, + "loss": 1.0542, + "step": 591 + }, + { + "epoch": 1.08, + "grad_norm": 0.3160409270291303, + "learning_rate": 6.447988962370133e-05, + "loss": 1.0829, + "step": 592 + }, + { + "epoch": 1.08, + "grad_norm": 0.390452102978435, + "learning_rate": 6.441241047399169e-05, + "loss": 1.192, + "step": 593 + }, + { + "epoch": 1.09, + "grad_norm": 0.3802122712014928, + "learning_rate": 6.434482042843627e-05, + "loss": 1.1153, + "step": 594 + }, + { + "epoch": 1.09, + "grad_norm": 0.4081584328242501, + "learning_rate": 6.427711979406966e-05, + "loss": 1.1635, + "step": 595 + }, + { + "epoch": 1.09, + "grad_norm": 0.3791962989638633, + "learning_rate": 6.420930887842889e-05, + "loss": 1.1581, + "step": 596 + }, + { + "epoch": 1.09, + "grad_norm": 0.33239440056484193, + "learning_rate": 6.414138798955189e-05, + "loss": 1.0926, + "step": 597 + }, + { + "epoch": 1.09, + "grad_norm": 0.3279881540815014, + "learning_rate": 6.407335743597616e-05, + "loss": 1.1386, + "step": 598 + }, + { + "epoch": 1.1, + "grad_norm": 0.30309644763750837, + "learning_rate": 6.40052175267374e-05, + "loss": 1.0523, + "step": 599 + }, + { + "epoch": 1.1, + "grad_norm": 0.3349097308403333, + "learning_rate": 6.393696857136801e-05, + "loss": 1.0815, + "step": 600 + } + ], + "logging_steps": 1.0, + "max_steps": 1638, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 50, + "total_flos": 622081452539904.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-600/training_args.bin b/checkpoint-600/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..c5d2416a3b70bb5260978ec9996f00154a724ba7 --- /dev/null +++ b/checkpoint-600/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b22e8f9d51a16d03a2c506fa3d1eafa8f4b1ae992992c2086a4d435ffd97387e +size 6712 diff --git a/checkpoint-600/zero_to_fp32.py b/checkpoint-600/zero_to_fp32.py new file mode 100755 index 0000000000000000000000000000000000000000..24cc342e78d1a006c782b3a4cd68d9ce786d8fd8 --- /dev/null +++ b/checkpoint-600/zero_to_fp32.py @@ -0,0 +1,604 @@ +#!/usr/bin/env python + +# Copyright (c) Microsoft Corporation. +# SPDX-License-Identifier: Apache-2.0 + +# DeepSpeed Team + +# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets +# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in +# the future. Once extracted, the weights don't require DeepSpeed and can be used in any +# application. +# +# example: python zero_to_fp32.py . pytorch_model.bin + +import argparse +import torch +import glob +import math +import os +import re +from collections import OrderedDict +from dataclasses import dataclass + +# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with +# DeepSpeed data structures it has to be available in the current python environment. +from deepspeed.utils import logger +from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS, + FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES, + FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS) + + +@dataclass +class zero_model_state: + buffers: dict() + param_shapes: dict() + shared_params: list + ds_version: int + frozen_param_shapes: dict() + frozen_param_fragments: dict() + + +debug = 0 + +# load to cpu +device = torch.device('cpu') + + +def atoi(text): + return int(text) if text.isdigit() else text + + +def natural_keys(text): + ''' + alist.sort(key=natural_keys) sorts in human order + http://nedbatchelder.com/blog/200712/human_sorting.html + (See Toothy's implementation in the comments) + ''' + return [atoi(c) for c in re.split(r'(\d+)', text)] + + +def get_model_state_file(checkpoint_dir, zero_stage): + if not os.path.isdir(checkpoint_dir): + raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist") + + # there should be only one file + if zero_stage <= 2: + file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt") + elif zero_stage == 3: + file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt") + + if not os.path.exists(file): + raise FileNotFoundError(f"can't find model states file at '{file}'") + + return file + + +def get_checkpoint_files(checkpoint_dir, glob_pattern): + # XXX: need to test that this simple glob rule works for multi-node setup too + ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys) + + if len(ckpt_files) == 0: + raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'") + + return ckpt_files + + +def get_optim_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt") + + +def get_model_state_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_model_states.pt") + + +def parse_model_states(files): + zero_model_states = [] + for file in files: + state_dict = torch.load(file, map_location=device) + + if BUFFER_NAMES not in state_dict: + raise ValueError(f"{file} is not a model state checkpoint") + buffer_names = state_dict[BUFFER_NAMES] + if debug: + print("Found buffers:", buffer_names) + + # recover just the buffers while restoring them to fp32 if they were saved in fp16 + buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names} + param_shapes = state_dict[PARAM_SHAPES] + + # collect parameters that are included in param_shapes + param_names = [] + for s in param_shapes: + for name in s.keys(): + param_names.append(name) + + # update with frozen parameters + frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None) + if frozen_param_shapes is not None: + if debug: + print(f"Found frozen_param_shapes: {frozen_param_shapes}") + param_names += list(frozen_param_shapes.keys()) + + # handle shared params + shared_params = [[k, v] for k, v in state_dict["shared_params"].items()] + + ds_version = state_dict.get(DS_VERSION, None) + + frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None) + + z_model_state = zero_model_state(buffers=buffers, + param_shapes=param_shapes, + shared_params=shared_params, + ds_version=ds_version, + frozen_param_shapes=frozen_param_shapes, + frozen_param_fragments=frozen_param_fragments) + zero_model_states.append(z_model_state) + + return zero_model_states + + +def parse_optim_states(files, ds_checkpoint_dir): + + total_files = len(files) + state_dicts = [] + for f in files: + state_dict = torch.load(f, map_location=device) + # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights + # and also handle the case where it was already removed by another helper script + state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None) + state_dicts.append(state_dict) + + if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]: + raise ValueError(f"{files[0]} is not a zero checkpoint") + zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE] + world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT] + + # For ZeRO-2 each param group can have different partition_count as data parallelism for expert + # parameters can be different from data parallelism for non-expert parameters. So we can just + # use the max of the partition_count to get the dp world_size. + + if type(world_size) is list: + world_size = max(world_size) + + if world_size != total_files: + raise ValueError( + f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. " + "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes." + ) + + # the groups are named differently in each stage + if zero_stage <= 2: + fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS + elif zero_stage == 3: + fp32_groups_key = FP32_FLAT_GROUPS + else: + raise ValueError(f"unknown zero stage {zero_stage}") + + if zero_stage <= 2: + fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))] + elif zero_stage == 3: + # if there is more than one param group, there will be multiple flattened tensors - one + # flattened tensor per group - for simplicity merge them into a single tensor + # + # XXX: could make the script more memory efficient for when there are multiple groups - it + # will require matching the sub-lists of param_shapes for each param group flattened tensor + + fp32_flat_groups = [ + torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts)) + ] + + return zero_stage, world_size, fp32_flat_groups + + +def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters): + """ + Returns fp32 state_dict reconstructed from ds checkpoint + + Args: + - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are) + + """ + print(f"Processing zero checkpoint '{ds_checkpoint_dir}'") + + optim_files = get_optim_files(ds_checkpoint_dir) + zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir) + print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}") + + model_files = get_model_state_files(ds_checkpoint_dir) + + zero_model_states = parse_model_states(model_files) + print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}') + + if zero_stage <= 2: + return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + elif zero_stage == 3: + return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + + +def _zero2_merge_frozen_params(state_dict, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + frozen_param_fragments = zero_model_states[0].frozen_param_fragments + + if debug: + num_elem = sum(s.numel() for s in frozen_param_shapes.values()) + print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in frozen_param_fragments.values()]) + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + state_dict[name] = frozen_param_fragments[name] + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _has_callable(obj, fn): + attr = getattr(obj, fn, None) + return callable(attr) + + +def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + + # Reconstruction protocol: + # + # XXX: document this + + if debug: + for i in range(world_size): + for j in range(len(fp32_flat_groups[0])): + print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}") + + # XXX: memory usage doubles here (zero2) + num_param_groups = len(fp32_flat_groups[0]) + merged_single_partition_of_fp32_groups = [] + for i in range(num_param_groups): + merged_partitions = [sd[i] for sd in fp32_flat_groups] + full_single_fp32_vector = torch.cat(merged_partitions, 0) + merged_single_partition_of_fp32_groups.append(full_single_fp32_vector) + avail_numel = sum( + [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups]) + + if debug: + wanted_params = sum([len(shapes) for shapes in param_shapes]) + wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes]) + # not asserting if there is a mismatch due to possible padding + print(f"Have {avail_numel} numels to process.") + print(f"Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + total_numel = 0 + total_params = 0 + for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups): + offset = 0 + avail_numel = full_single_fp32_vector.numel() + for name, shape in shapes.items(): + + unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape) + total_numel += unpartitioned_numel + total_params += 1 + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape) + offset += unpartitioned_numel + + # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and + # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex + # paddings performed in the code it's almost impossible to predict the exact numbers w/o the + # live optimizer object, so we are checking that the numbers are within the right range + align_to = 2 * world_size + + def zero2_align(x): + return align_to * math.ceil(x / align_to) + + if debug: + print(f"original offset={offset}, avail_numel={avail_numel}") + + offset = zero2_align(offset) + avail_numel = zero2_align(avail_numel) + + if debug: + print(f"aligned offset={offset}, avail_numel={avail_numel}") + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero2_merge_frozen_params(state_dict, zero_model_states) + + _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def zero3_partitioned_param_info(unpartitioned_numel, world_size): + remainder = unpartitioned_numel % world_size + padding_numel = (world_size - remainder) if remainder else 0 + partitioned_numel = math.ceil(unpartitioned_numel / world_size) + return partitioned_numel, padding_numel + + +def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + if debug: + for i in range(world_size): + num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values()) + print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in zero_model_states[0].frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states) + state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape) + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + avail_numel = fp32_flat_groups[0].numel() * world_size + # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each + # param, re-consolidating each param, while dealing with padding if any + + # merge list of dicts, preserving order + param_shapes = {k: v for d in param_shapes for k, v in d.items()} + + if debug: + for i in range(world_size): + print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}") + + wanted_params = len(param_shapes) + wanted_numel = sum(shape.numel() for shape in param_shapes.values()) + # not asserting if there is a mismatch due to possible padding + avail_numel = fp32_flat_groups[0].numel() * world_size + print(f"Trainable params: Have {avail_numel} numels to process.") + print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + offset = 0 + total_numel = 0 + total_params = 0 + for name, shape in param_shapes.items(): + + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + total_params += 1 + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + # XXX: memory usage doubles here + state_dict[name] = torch.cat( + tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)), + 0).narrow(0, 0, unpartitioned_numel).view(shape) + offset += partitioned_numel + + offset *= world_size + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero3_merge_frozen_params(state_dict, world_size, zero_model_states) + + _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with + ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example + via a model hub. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + + Returns: + - pytorch ``state_dict`` + + Note: this approach may not work if your application doesn't have sufficient free CPU memory and + you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with + the checkpoint. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + # do the training and checkpoint saving + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu + model = model.cpu() # move to cpu + model.load_state_dict(state_dict) + # submit to model hub or save the model to share with others + + In this example the ``model`` will no longer be usable in the deepspeed context of the same + application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead. + + """ + if tag is None: + latest_path = os.path.join(checkpoint_dir, 'latest') + if os.path.isfile(latest_path): + with open(latest_path, 'r') as fd: + tag = fd.read().strip() + else: + raise ValueError(f"Unable to find 'latest' file at {latest_path}") + + ds_checkpoint_dir = os.path.join(checkpoint_dir, tag) + + if not os.path.isdir(ds_checkpoint_dir): + raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist") + + return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters) + + +def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be + loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + """ + + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters) + print(f"Saving fp32 state dict to {output_file}") + torch.save(state_dict, output_file) + + +def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None): + """ + 1. Put the provided model to cpu + 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` + 3. Load it into the provided model + + Args: + - ``model``: the model object to update + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + + Returns: + - ``model`: modified model + + Make sure you have plenty of CPU memory available before you call this function. If you don't + have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it + conveniently placed for you in the checkpoint folder. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint + model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir) + # submit to model hub or save the model to share with others + + Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context + of the same application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + """ + logger.info(f"Extracting fp32 weights") + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag) + + logger.info(f"Overwriting model with fp32 weights") + model = model.cpu() + model.load_state_dict(state_dict, strict=False) + + return model + + +if __name__ == "__main__": + + parser = argparse.ArgumentParser() + parser.add_argument("checkpoint_dir", + type=str, + help="path to the desired checkpoint folder, e.g., path/checkpoint-12") + parser.add_argument( + "output_file", + type=str, + help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)") + parser.add_argument("-t", + "--tag", + type=str, + default=None, + help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1") + parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters") + parser.add_argument("-d", "--debug", action='store_true', help="enable debug") + args = parser.parse_args() + + debug = args.debug + + convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, + args.output_file, + tag=args.tag, + exclude_frozen_parameters=args.exclude_frozen_parameters) diff --git a/checkpoint-750/README.md b/checkpoint-750/README.md new file mode 100644 index 0000000000000000000000000000000000000000..16b1eacdd9353dec380a08ee77ce6ed5ab50f12e --- /dev/null +++ b/checkpoint-750/README.md @@ -0,0 +1,202 @@ +--- +library_name: peft +base_model: gotzmann/uni +--- + +# Model Card for Model ID + + + + + +## Model Details + +### Model Description + + + + + +- **Developed by:** [More Information Needed] +- **Funded by [optional]:** [More Information Needed] +- **Shared by [optional]:** [More Information Needed] +- **Model type:** [More Information Needed] +- **Language(s) (NLP):** [More Information Needed] +- **License:** [More Information Needed] +- **Finetuned from model [optional]:** [More Information Needed] + +### Model Sources [optional] + + + +- **Repository:** [More Information Needed] +- **Paper [optional]:** [More Information Needed] +- **Demo [optional]:** [More Information Needed] + +## Uses + + + +### Direct Use + + + +[More Information Needed] + +### Downstream Use [optional] + + + +[More Information Needed] + +### Out-of-Scope Use + + + +[More Information Needed] + +## Bias, Risks, and Limitations + + + +[More Information Needed] + +### Recommendations + + + +Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations. + +## How to Get Started with the Model + +Use the code below to get started with the model. + +[More Information Needed] + +## Training Details + +### Training Data + + + +[More Information Needed] + +### Training Procedure + + + +#### Preprocessing [optional] + +[More Information Needed] + + +#### Training Hyperparameters + +- **Training regime:** [More Information Needed] + +#### Speeds, Sizes, Times [optional] + + + +[More Information Needed] + +## Evaluation + + + +### Testing Data, Factors & Metrics + +#### Testing Data + + + +[More Information Needed] + +#### Factors + + + +[More Information Needed] + +#### Metrics + + + +[More Information Needed] + +### Results + +[More Information Needed] + +#### Summary + + + +## Model Examination [optional] + + + +[More Information Needed] + +## Environmental Impact + + + +Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700). + +- **Hardware Type:** [More Information Needed] +- **Hours used:** [More Information Needed] +- **Cloud Provider:** [More Information Needed] +- **Compute Region:** [More Information Needed] +- **Carbon Emitted:** [More Information Needed] + +## Technical Specifications [optional] + +### Model Architecture and Objective + +[More Information Needed] + +### Compute Infrastructure + +[More Information Needed] + +#### Hardware + +[More Information Needed] + +#### Software + +[More Information Needed] + +## Citation [optional] + + + +**BibTeX:** + +[More Information Needed] + +**APA:** + +[More Information Needed] + +## Glossary [optional] + + + +[More Information Needed] + +## More Information [optional] + +[More Information Needed] + +## Model Card Authors [optional] + +[More Information Needed] + +## Model Card Contact + +[More Information Needed] +### Framework versions + +- PEFT 0.10.0 \ No newline at end of file diff --git a/checkpoint-750/adapter_config.json b/checkpoint-750/adapter_config.json new file mode 100644 index 0000000000000000000000000000000000000000..3cd6dba5d79f7ca21fd4ad465cbbcac1e0960476 --- /dev/null +++ b/checkpoint-750/adapter_config.json @@ -0,0 +1,31 @@ +{ + "alpha_pattern": {}, + "auto_mapping": null, + "base_model_name_or_path": "gotzmann/uni", + "bias": "none", + "fan_in_fan_out": false, + "inference_mode": true, + "init_lora_weights": true, + "layer_replication": null, + "layers_pattern": null, + "layers_to_transform": null, + "loftq_config": {}, + "lora_alpha": 128, + "lora_dropout": 0.0, + "megatron_config": null, + "megatron_core": "megatron.core", + "modules_to_save": null, + "peft_type": "LORA", + "r": 128, + "rank_pattern": {}, + "revision": null, + "target_modules": [ + "v_proj", + "k_proj", + "q_proj", + "o_proj" + ], + "task_type": "CAUSAL_LM", + "use_dora": false, + "use_rslora": true +} \ No newline at end of file diff --git a/checkpoint-750/adapter_model.safetensors b/checkpoint-750/adapter_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..a13b064da6a9ec698c137a73eba9a668c95aabc9 --- /dev/null +++ b/checkpoint-750/adapter_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d08962f3cc64278013e1154366e9ca7f6847a52987d6f5be68a01f40a5344d8d +size 1048664848 diff --git a/checkpoint-750/global_step750/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt b/checkpoint-750/global_step750/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..156285fe76bb84fc9b9ea5f08ee13307ef2c66cd --- /dev/null +++ b/checkpoint-750/global_step750/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:a711fabb31e0d2159d23671070ef18972c8c1a55fb06f148b6ce0b9b54052fd3 +size 787270042 diff --git a/checkpoint-750/global_step750/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt b/checkpoint-750/global_step750/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..f9276383cba2ca5acfcd136cfe7a3872ff4948e2 --- /dev/null +++ b/checkpoint-750/global_step750/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:02b8656caf74761fc753d0398eb4b8d4b2a1f284d77cdaa713da66b088d8f494 +size 787270042 diff --git a/checkpoint-750/global_step750/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt b/checkpoint-750/global_step750/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..88dfa87da72326795ed42294dd3bee13d0f3d8df --- /dev/null +++ b/checkpoint-750/global_step750/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:fafcd8b6b974b21163df6a4fd91c78e4a79519595f2412c66fe9a38e2c350725 +size 787270042 diff --git a/checkpoint-750/global_step750/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt b/checkpoint-750/global_step750/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..2f86f2cb93a4004efc85a4b24747a0aadb1481c4 --- /dev/null +++ b/checkpoint-750/global_step750/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:dbb11855fbd7a091f1085b2057cb49b96a8726a48bde0fe69c8778480ab8941f +size 787270042 diff --git a/checkpoint-750/global_step750/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt b/checkpoint-750/global_step750/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..c3af1c86414aec1cb392f6c6571273a5fd8c2f19 --- /dev/null +++ b/checkpoint-750/global_step750/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:c2a49eb0fe6cfeb58c9fbfa4393a239434908146331a56f42ef6041bed7d719b +size 787270042 diff --git a/checkpoint-750/global_step750/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt b/checkpoint-750/global_step750/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..9062c6bd1e9f96d605e6b761c22e07a47746d8dc --- /dev/null +++ b/checkpoint-750/global_step750/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:970f21d22af841bcb7d7b1d57d8f2000c8b7cf698551b3b69757fbdc337dd112 +size 787270042 diff --git a/checkpoint-750/global_step750/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt b/checkpoint-750/global_step750/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..41acff84ea4aae21964cd74e14e741cbc4f86bb9 --- /dev/null +++ b/checkpoint-750/global_step750/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8ed9432117d78258c483a7bc8af52ce435685949414278321041bb30cdf9b138 +size 787270042 diff --git a/checkpoint-750/global_step750/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt b/checkpoint-750/global_step750/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..9331c8c8a96f0203f54c20facc1391c2bfd22413 --- /dev/null +++ b/checkpoint-750/global_step750/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:dcf390b7531b58dd58a8c9fa8dbc722d316d46661a99d4cbe21bdb1a913b9183 +size 787270042 diff --git a/checkpoint-750/global_step750/zero_pp_rank_0_mp_rank_00_model_states.pt b/checkpoint-750/global_step750/zero_pp_rank_0_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..fe79240f6b4a706b86c019a7a4009f1657ec5974 --- /dev/null +++ b/checkpoint-750/global_step750/zero_pp_rank_0_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e1c3767dfc19806a88ba86871e990625153fd9b5e05127f3cfcccc700f6c07e6 +size 653742 diff --git a/checkpoint-750/global_step750/zero_pp_rank_1_mp_rank_00_model_states.pt b/checkpoint-750/global_step750/zero_pp_rank_1_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..177f90f14ef5e987a3d5886470afa80c620a80eb --- /dev/null +++ b/checkpoint-750/global_step750/zero_pp_rank_1_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b7dc078f6c51e1ac0798eb3cfcd60e50ad6f34a4faa986ca780e2f9d3f0fd776 +size 653742 diff --git a/checkpoint-750/global_step750/zero_pp_rank_2_mp_rank_00_model_states.pt b/checkpoint-750/global_step750/zero_pp_rank_2_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..5b7a513ff6a7c6051c59d443e8bbab412aef475a --- /dev/null +++ b/checkpoint-750/global_step750/zero_pp_rank_2_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a367ca87901d820a9554c8d7296018a6d28f33d21c8e7d087ac59bba94204cc7 +size 653742 diff --git a/checkpoint-750/global_step750/zero_pp_rank_3_mp_rank_00_model_states.pt b/checkpoint-750/global_step750/zero_pp_rank_3_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..8af29a8256c90fc06047690be799aafd6621a12a --- /dev/null +++ b/checkpoint-750/global_step750/zero_pp_rank_3_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:762d9a703f46f0ce600c1edf3107a2ca634ac1c6664925f5a3190c51698dc801 +size 653742 diff --git a/checkpoint-750/global_step750/zero_pp_rank_4_mp_rank_00_model_states.pt b/checkpoint-750/global_step750/zero_pp_rank_4_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..f4c0e663c450df3de6ae5ad4769eb1940d725be3 --- /dev/null +++ b/checkpoint-750/global_step750/zero_pp_rank_4_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:69ab1708ce4fae1650ea631de1ef6e460926fa6a787cdd7d989e35749a754dca +size 653742 diff --git a/checkpoint-750/global_step750/zero_pp_rank_5_mp_rank_00_model_states.pt b/checkpoint-750/global_step750/zero_pp_rank_5_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..6f613e029841b0f414d01e85f4b066e85314643f --- /dev/null +++ b/checkpoint-750/global_step750/zero_pp_rank_5_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0f81695549d4f5801aa7104af821ca3c18fca257c62a20eb1e9b65a69462930b +size 653742 diff --git a/checkpoint-750/global_step750/zero_pp_rank_6_mp_rank_00_model_states.pt b/checkpoint-750/global_step750/zero_pp_rank_6_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..005adfd9b83cf9ff36d3072c52b7df7c0337d32a --- /dev/null +++ b/checkpoint-750/global_step750/zero_pp_rank_6_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1791542f5e2c71a6c3fdaba8ef916dcbe1edce8292a656307dc9f5298bb3fde4 +size 653742 diff --git a/checkpoint-750/global_step750/zero_pp_rank_7_mp_rank_00_model_states.pt b/checkpoint-750/global_step750/zero_pp_rank_7_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..815e479cc697e2a18a6369b7d0ca607810e9c88d --- /dev/null +++ b/checkpoint-750/global_step750/zero_pp_rank_7_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:da14438c3212dca3b622deb8f8508e816c58644265ee0ff0ca9152efc9b9d8e6 +size 653742 diff --git a/checkpoint-750/latest b/checkpoint-750/latest new file mode 100644 index 0000000000000000000000000000000000000000..f443e084e0e73b2cb9226c3b73c42d443059068f --- /dev/null +++ b/checkpoint-750/latest @@ -0,0 +1 @@ +global_step750 \ No newline at end of file diff --git a/checkpoint-750/rng_state_0.pth b/checkpoint-750/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..4e5b7e2ec90fdb824c8932464c1d9068330655a7 --- /dev/null +++ b/checkpoint-750/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:36d2a2034ebb05cb71c510897f2795b31164e50f17b270bc25d2be3ad9a17b22 +size 15984 diff --git a/checkpoint-750/rng_state_1.pth b/checkpoint-750/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..7d8d7722fc72cab6d492b76cb99c8177dcc47544 --- /dev/null +++ b/checkpoint-750/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:060dfdb1c49102cbdc8868a6031e68787601b4ccd782f3fb9b137e20c1fd2c7a +size 15984 diff --git a/checkpoint-750/rng_state_2.pth b/checkpoint-750/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..3c9f84eff30cfa9ea1feedaf262d61fb12e4cba7 --- /dev/null +++ b/checkpoint-750/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:af01895cb66e616591f2e4baa8dcd8151530eab133c73571ccb31c74f35422ce +size 15984 diff --git a/checkpoint-750/rng_state_3.pth b/checkpoint-750/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..6eebfb928f8e91eff0ea1645a20b5aa4465c705b --- /dev/null +++ b/checkpoint-750/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:677921992b1e0cef3aee776f245975003d22f51d9bd6ed20f248ded1deb72fa9 +size 15984 diff --git a/checkpoint-750/rng_state_4.pth b/checkpoint-750/rng_state_4.pth new file mode 100644 index 0000000000000000000000000000000000000000..0866030a266c6d003cc378a9418a723f69e8ab99 --- /dev/null +++ b/checkpoint-750/rng_state_4.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d69353c629541c690c5471f8ec05fdab2bfecf3d37afaa436bc45939da6db68f +size 15984 diff --git a/checkpoint-750/rng_state_5.pth b/checkpoint-750/rng_state_5.pth new file mode 100644 index 0000000000000000000000000000000000000000..554638d77107f832d7aa51c61645ee2d6c48a36d --- /dev/null +++ b/checkpoint-750/rng_state_5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8e40ba6668cc03c9162c68a933d164bf38ae2d196a9a6fec03ae615491201185 +size 15984 diff --git a/checkpoint-750/rng_state_6.pth b/checkpoint-750/rng_state_6.pth new file mode 100644 index 0000000000000000000000000000000000000000..964331b65172a1bcac03e4673415fa787f724268 --- /dev/null +++ b/checkpoint-750/rng_state_6.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:870968fea834e24b2e099cf3e4fe1e3fb8caf38d8f8e5b790d7d47386d4d05f5 +size 15984 diff --git a/checkpoint-750/rng_state_7.pth b/checkpoint-750/rng_state_7.pth new file mode 100644 index 0000000000000000000000000000000000000000..cd4754d65217d0f9d1f2d3334397df7a8a079652 --- /dev/null +++ b/checkpoint-750/rng_state_7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e9e19618bee7c6ef43256fea25abe19bca88535eb1e7dc213cde8929ae4e8180 +size 15984 diff --git a/checkpoint-750/scheduler.pt b/checkpoint-750/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..41ff0494d20d190c25da992ece6b8b358028d564 --- /dev/null +++ b/checkpoint-750/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:59e12f88a62db407aa272bd77b60965bec97929153b681326f419d8de8cda662 +size 1064 diff --git a/checkpoint-750/special_tokens_map.json b/checkpoint-750/special_tokens_map.json new file mode 100644 index 0000000000000000000000000000000000000000..72ecfeeb7e14d244c936169d2ed139eeae235ef1 --- /dev/null +++ b/checkpoint-750/special_tokens_map.json @@ -0,0 +1,24 @@ +{ + "bos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "eos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "pad_token": "", + "unk_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + } +} diff --git a/checkpoint-750/tokenizer.model b/checkpoint-750/tokenizer.model new file mode 100644 index 0000000000000000000000000000000000000000..6c00c742ce03c627d6cd5b795984876fa49fa899 --- /dev/null +++ b/checkpoint-750/tokenizer.model @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347 +size 499723 diff --git a/checkpoint-750/tokenizer_config.json b/checkpoint-750/tokenizer_config.json new file mode 100644 index 0000000000000000000000000000000000000000..bb5a9f09d8c0f3c32c66fc6118fe5c76c5c6fd90 --- /dev/null +++ b/checkpoint-750/tokenizer_config.json @@ -0,0 +1,45 @@ +{ + "add_bos_token": true, + "add_eos_token": false, + "add_prefix_space": false, + "added_tokens_decoder": { + "0": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "1": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "2": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + } + }, + "bos_token": "", + "chat_template": "{% if messages[0]['role'] == 'system' %}{% set system_message = messages[0]['content'] %}{% endif %}{% if system_message is defined %}{{ '' + '### System:\\n\\n' + system_message }}{% endif %}{% for message in messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ '\\n\\n### Human:\\n\\n' + content }}{% elif message['role'] == 'assistant' %}{{ '\\n\\n### Assistant:\\n\\n' + content + '' }}{% endif %}{% endfor %}", + "clean_up_tokenization_spaces": false, + "eos_token": "", + "legacy": false, + "model_max_length": 1000000000000000019884624838656, + "pad_token": "", + "padding_side": "right", + "sp_model_kwargs": {}, + "spaces_between_special_tokens": false, + "split_special_tokens": false, + "tokenizer_class": "LlamaTokenizer", + "unk_token": "", + "use_default_system_prompt": false +} diff --git a/checkpoint-750/trainer_state.json b/checkpoint-750/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..449168623e2d504d23331bacd9d0ed25aaf509c7 --- /dev/null +++ b/checkpoint-750/trainer_state.json @@ -0,0 +1,5271 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 1.3717421124828533, + "eval_steps": 500, + "global_step": 750, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "grad_norm": 0.849355824164473, + "learning_rate": 4.878048780487805e-07, + "loss": 1.3655, + "step": 1 + }, + { + "epoch": 0.0, + "grad_norm": 10.01567518957158, + "learning_rate": 9.75609756097561e-07, + "loss": 1.5767, + "step": 2 + }, + { + "epoch": 0.01, + "grad_norm": 0.6466000875559635, + "learning_rate": 1.4634146341463414e-06, + "loss": 1.3913, + "step": 3 + }, + { + "epoch": 0.01, + "grad_norm": 0.6644565932010504, + "learning_rate": 1.951219512195122e-06, + "loss": 1.3218, + "step": 4 + }, + { + "epoch": 0.01, + "grad_norm": 0.571354207588475, + "learning_rate": 2.4390243902439027e-06, + "loss": 1.3597, + "step": 5 + }, + { + "epoch": 0.01, + "grad_norm": 0.31036262839244955, + "learning_rate": 2.926829268292683e-06, + "loss": 1.2832, + "step": 6 + }, + { + "epoch": 0.01, + "grad_norm": 0.2622135027188184, + "learning_rate": 3.414634146341464e-06, + "loss": 1.2161, + "step": 7 + }, + { + "epoch": 0.01, + "grad_norm": 0.296824630261661, + "learning_rate": 3.902439024390244e-06, + "loss": 1.2985, + "step": 8 + }, + { + "epoch": 0.02, + "grad_norm": 0.2557267467361569, + "learning_rate": 4.390243902439025e-06, + "loss": 1.3175, + "step": 9 + }, + { + "epoch": 0.02, + "grad_norm": 0.23418939513890769, + "learning_rate": 4.8780487804878055e-06, + "loss": 1.2617, + "step": 10 + }, + { + "epoch": 0.02, + "grad_norm": 0.2364760983285843, + "learning_rate": 5.365853658536586e-06, + "loss": 1.3103, + "step": 11 + }, + { + "epoch": 0.02, + "grad_norm": 0.23893034721889, + "learning_rate": 5.853658536585366e-06, + "loss": 1.2405, + "step": 12 + }, + { + "epoch": 0.02, + "grad_norm": 0.25563593295485887, + "learning_rate": 6.341463414634147e-06, + "loss": 1.2831, + "step": 13 + }, + { + "epoch": 0.03, + "grad_norm": 0.23239975352661665, + "learning_rate": 6.829268292682928e-06, + "loss": 1.3125, + "step": 14 + }, + { + "epoch": 0.03, + "grad_norm": 0.3092813858209507, + "learning_rate": 7.317073170731707e-06, + "loss": 1.2422, + "step": 15 + }, + { + "epoch": 0.03, + "grad_norm": 0.282563380367434, + "learning_rate": 7.804878048780489e-06, + "loss": 1.2453, + "step": 16 + }, + { + "epoch": 0.03, + "grad_norm": 0.22065680088315018, + "learning_rate": 8.292682926829268e-06, + "loss": 1.2491, + "step": 17 + }, + { + "epoch": 0.03, + "grad_norm": 0.22777800877980184, + "learning_rate": 8.78048780487805e-06, + "loss": 1.2655, + "step": 18 + }, + { + "epoch": 0.03, + "grad_norm": 0.22145212540177928, + "learning_rate": 9.268292682926831e-06, + "loss": 1.2413, + "step": 19 + }, + { + "epoch": 0.04, + "grad_norm": 0.22482351883112714, + "learning_rate": 9.756097560975611e-06, + "loss": 1.2653, + "step": 20 + }, + { + "epoch": 0.04, + "grad_norm": 0.20823080508385733, + "learning_rate": 1.024390243902439e-05, + "loss": 1.2374, + "step": 21 + }, + { + "epoch": 0.04, + "grad_norm": 0.26025492562935737, + "learning_rate": 1.0731707317073172e-05, + "loss": 1.2065, + "step": 22 + }, + { + "epoch": 0.04, + "grad_norm": 0.2150252124176173, + "learning_rate": 1.1219512195121953e-05, + "loss": 1.2782, + "step": 23 + }, + { + "epoch": 0.04, + "grad_norm": 0.2505915177425618, + "learning_rate": 1.1707317073170731e-05, + "loss": 1.2742, + "step": 24 + }, + { + "epoch": 0.05, + "grad_norm": 0.20129223044786942, + "learning_rate": 1.2195121951219513e-05, + "loss": 1.3366, + "step": 25 + }, + { + "epoch": 0.05, + "grad_norm": 0.1973508510397107, + "learning_rate": 1.2682926829268294e-05, + "loss": 1.2476, + "step": 26 + }, + { + "epoch": 0.05, + "grad_norm": 0.27103325392437194, + "learning_rate": 1.3170731707317076e-05, + "loss": 1.2325, + "step": 27 + }, + { + "epoch": 0.05, + "grad_norm": 0.17954976411006285, + "learning_rate": 1.3658536585365855e-05, + "loss": 1.2523, + "step": 28 + }, + { + "epoch": 0.05, + "grad_norm": 0.22216997851088888, + "learning_rate": 1.4146341463414635e-05, + "loss": 1.3297, + "step": 29 + }, + { + "epoch": 0.05, + "grad_norm": 0.2071458864548587, + "learning_rate": 1.4634146341463415e-05, + "loss": 1.2127, + "step": 30 + }, + { + "epoch": 0.06, + "grad_norm": 0.18039422081622164, + "learning_rate": 1.5121951219512196e-05, + "loss": 1.2509, + "step": 31 + }, + { + "epoch": 0.06, + "grad_norm": 0.18631254372974412, + "learning_rate": 1.5609756097560978e-05, + "loss": 1.2247, + "step": 32 + }, + { + "epoch": 0.06, + "grad_norm": 0.18843872523649827, + "learning_rate": 1.6097560975609757e-05, + "loss": 1.195, + "step": 33 + }, + { + "epoch": 0.06, + "grad_norm": 0.2163847267778325, + "learning_rate": 1.6585365853658537e-05, + "loss": 1.2179, + "step": 34 + }, + { + "epoch": 0.06, + "grad_norm": 0.19687688475496104, + "learning_rate": 1.7073170731707317e-05, + "loss": 1.2763, + "step": 35 + }, + { + "epoch": 0.07, + "grad_norm": 0.20409643064887947, + "learning_rate": 1.75609756097561e-05, + "loss": 1.253, + "step": 36 + }, + { + "epoch": 0.07, + "grad_norm": 0.1879182661759335, + "learning_rate": 1.804878048780488e-05, + "loss": 1.2586, + "step": 37 + }, + { + "epoch": 0.07, + "grad_norm": 0.19400648948514373, + "learning_rate": 1.8536585365853663e-05, + "loss": 1.2154, + "step": 38 + }, + { + "epoch": 0.07, + "grad_norm": 0.1878879343148452, + "learning_rate": 1.902439024390244e-05, + "loss": 1.2304, + "step": 39 + }, + { + "epoch": 0.07, + "grad_norm": 0.17687475469924052, + "learning_rate": 1.9512195121951222e-05, + "loss": 1.2351, + "step": 40 + }, + { + "epoch": 0.07, + "grad_norm": 0.18223935625384885, + "learning_rate": 2e-05, + "loss": 1.2222, + "step": 41 + }, + { + "epoch": 0.08, + "grad_norm": 0.1943061629408338, + "learning_rate": 2.048780487804878e-05, + "loss": 1.2044, + "step": 42 + }, + { + "epoch": 0.08, + "grad_norm": 0.17027514338700078, + "learning_rate": 2.0975609756097564e-05, + "loss": 1.1548, + "step": 43 + }, + { + "epoch": 0.08, + "grad_norm": 0.18553769630586192, + "learning_rate": 2.1463414634146344e-05, + "loss": 1.2721, + "step": 44 + }, + { + "epoch": 0.08, + "grad_norm": 0.19732826914228765, + "learning_rate": 2.1951219512195124e-05, + "loss": 1.3097, + "step": 45 + }, + { + "epoch": 0.08, + "grad_norm": 0.18714230986631472, + "learning_rate": 2.2439024390243907e-05, + "loss": 1.2662, + "step": 46 + }, + { + "epoch": 0.09, + "grad_norm": 0.19988987568002223, + "learning_rate": 2.2926829268292683e-05, + "loss": 1.2904, + "step": 47 + }, + { + "epoch": 0.09, + "grad_norm": 0.17744650133390918, + "learning_rate": 2.3414634146341463e-05, + "loss": 1.1825, + "step": 48 + }, + { + "epoch": 0.09, + "grad_norm": 0.16576734763834533, + "learning_rate": 2.3902439024390246e-05, + "loss": 1.1858, + "step": 49 + }, + { + "epoch": 0.09, + "grad_norm": 0.179591794065527, + "learning_rate": 2.4390243902439026e-05, + "loss": 1.2711, + "step": 50 + }, + { + "epoch": 0.09, + "grad_norm": 0.17923464471176911, + "learning_rate": 2.4878048780487805e-05, + "loss": 1.2289, + "step": 51 + }, + { + "epoch": 0.1, + "grad_norm": 0.18991742907836837, + "learning_rate": 2.536585365853659e-05, + "loss": 1.3097, + "step": 52 + }, + { + "epoch": 0.1, + "grad_norm": 0.19849796137254636, + "learning_rate": 2.5853658536585368e-05, + "loss": 1.2489, + "step": 53 + }, + { + "epoch": 0.1, + "grad_norm": 0.17452371110976383, + "learning_rate": 2.634146341463415e-05, + "loss": 1.2461, + "step": 54 + }, + { + "epoch": 0.1, + "grad_norm": 0.17671022353085036, + "learning_rate": 2.682926829268293e-05, + "loss": 1.153, + "step": 55 + }, + { + "epoch": 0.1, + "grad_norm": 0.36820559192096686, + "learning_rate": 2.731707317073171e-05, + "loss": 1.2431, + "step": 56 + }, + { + "epoch": 0.1, + "grad_norm": 0.20331468526494198, + "learning_rate": 2.7804878048780487e-05, + "loss": 1.2575, + "step": 57 + }, + { + "epoch": 0.11, + "grad_norm": 0.2402486598118377, + "learning_rate": 2.829268292682927e-05, + "loss": 1.2538, + "step": 58 + }, + { + "epoch": 0.11, + "grad_norm": 0.2549409484173144, + "learning_rate": 2.878048780487805e-05, + "loss": 1.2065, + "step": 59 + }, + { + "epoch": 0.11, + "grad_norm": 0.2053105349872685, + "learning_rate": 2.926829268292683e-05, + "loss": 1.2094, + "step": 60 + }, + { + "epoch": 0.11, + "grad_norm": 0.17971910872957886, + "learning_rate": 2.9756097560975613e-05, + "loss": 1.228, + "step": 61 + }, + { + "epoch": 0.11, + "grad_norm": 0.1885853654992973, + "learning_rate": 3.0243902439024392e-05, + "loss": 1.2286, + "step": 62 + }, + { + "epoch": 0.12, + "grad_norm": 0.1848524571968613, + "learning_rate": 3.073170731707317e-05, + "loss": 1.2718, + "step": 63 + }, + { + "epoch": 0.12, + "grad_norm": 0.18734105883548513, + "learning_rate": 3.1219512195121955e-05, + "loss": 1.2357, + "step": 64 + }, + { + "epoch": 0.12, + "grad_norm": 0.17774668052121825, + "learning_rate": 3.170731707317074e-05, + "loss": 1.1509, + "step": 65 + }, + { + "epoch": 0.12, + "grad_norm": 0.17890968008080646, + "learning_rate": 3.2195121951219514e-05, + "loss": 1.1924, + "step": 66 + }, + { + "epoch": 0.12, + "grad_norm": 0.18249273371332375, + "learning_rate": 3.268292682926829e-05, + "loss": 1.2545, + "step": 67 + }, + { + "epoch": 0.12, + "grad_norm": 0.21064122671902577, + "learning_rate": 3.3170731707317074e-05, + "loss": 1.2832, + "step": 68 + }, + { + "epoch": 0.13, + "grad_norm": 0.1820064171955093, + "learning_rate": 3.365853658536586e-05, + "loss": 1.2071, + "step": 69 + }, + { + "epoch": 0.13, + "grad_norm": 0.16996662800553433, + "learning_rate": 3.414634146341463e-05, + "loss": 1.2073, + "step": 70 + }, + { + "epoch": 0.13, + "grad_norm": 0.1618669302922445, + "learning_rate": 3.4634146341463416e-05, + "loss": 1.1289, + "step": 71 + }, + { + "epoch": 0.13, + "grad_norm": 0.18948744950985544, + "learning_rate": 3.51219512195122e-05, + "loss": 1.2915, + "step": 72 + }, + { + "epoch": 0.13, + "grad_norm": 0.18326143691603383, + "learning_rate": 3.5609756097560976e-05, + "loss": 1.2238, + "step": 73 + }, + { + "epoch": 0.14, + "grad_norm": 0.17410704510700503, + "learning_rate": 3.609756097560976e-05, + "loss": 1.1784, + "step": 74 + }, + { + "epoch": 0.14, + "grad_norm": 0.1983667344995625, + "learning_rate": 3.658536585365854e-05, + "loss": 1.2452, + "step": 75 + }, + { + "epoch": 0.14, + "grad_norm": 0.3416310763369357, + "learning_rate": 3.7073170731707325e-05, + "loss": 1.1972, + "step": 76 + }, + { + "epoch": 0.14, + "grad_norm": 0.2776466983511955, + "learning_rate": 3.75609756097561e-05, + "loss": 1.3121, + "step": 77 + }, + { + "epoch": 0.14, + "grad_norm": 0.20026129636576834, + "learning_rate": 3.804878048780488e-05, + "loss": 1.2436, + "step": 78 + }, + { + "epoch": 0.14, + "grad_norm": 0.21064549243917835, + "learning_rate": 3.853658536585366e-05, + "loss": 1.2064, + "step": 79 + }, + { + "epoch": 0.15, + "grad_norm": 0.22119482175714267, + "learning_rate": 3.9024390243902444e-05, + "loss": 1.2715, + "step": 80 + }, + { + "epoch": 0.15, + "grad_norm": 0.23047133748844142, + "learning_rate": 3.951219512195122e-05, + "loss": 1.2888, + "step": 81 + }, + { + "epoch": 0.15, + "grad_norm": 0.18741863156973176, + "learning_rate": 4e-05, + "loss": 1.248, + "step": 82 + }, + { + "epoch": 0.15, + "grad_norm": 0.1747859810629604, + "learning_rate": 4.0487804878048786e-05, + "loss": 1.1683, + "step": 83 + }, + { + "epoch": 0.15, + "grad_norm": 0.1896944798413341, + "learning_rate": 4.097560975609756e-05, + "loss": 1.2155, + "step": 84 + }, + { + "epoch": 0.16, + "grad_norm": 0.18724128114363303, + "learning_rate": 4.1463414634146346e-05, + "loss": 1.2273, + "step": 85 + }, + { + "epoch": 0.16, + "grad_norm": 0.17368125504855478, + "learning_rate": 4.195121951219513e-05, + "loss": 1.224, + "step": 86 + }, + { + "epoch": 0.16, + "grad_norm": 0.18371141013625703, + "learning_rate": 4.2439024390243905e-05, + "loss": 1.2294, + "step": 87 + }, + { + "epoch": 0.16, + "grad_norm": 0.1791029365673714, + "learning_rate": 4.292682926829269e-05, + "loss": 1.2895, + "step": 88 + }, + { + "epoch": 0.16, + "grad_norm": 0.20259974283859655, + "learning_rate": 4.341463414634147e-05, + "loss": 1.1841, + "step": 89 + }, + { + "epoch": 0.16, + "grad_norm": 0.17457456183272174, + "learning_rate": 4.390243902439025e-05, + "loss": 1.2357, + "step": 90 + }, + { + "epoch": 0.17, + "grad_norm": 0.1815824380789748, + "learning_rate": 4.439024390243903e-05, + "loss": 1.2304, + "step": 91 + }, + { + "epoch": 0.17, + "grad_norm": 0.17566480599583392, + "learning_rate": 4.4878048780487814e-05, + "loss": 1.242, + "step": 92 + }, + { + "epoch": 0.17, + "grad_norm": 0.18422975005984474, + "learning_rate": 4.536585365853658e-05, + "loss": 1.2177, + "step": 93 + }, + { + "epoch": 0.17, + "grad_norm": 0.16796781877940678, + "learning_rate": 4.5853658536585366e-05, + "loss": 1.1482, + "step": 94 + }, + { + "epoch": 0.17, + "grad_norm": 0.18636131653783305, + "learning_rate": 4.634146341463415e-05, + "loss": 1.1758, + "step": 95 + }, + { + "epoch": 0.18, + "grad_norm": 0.1823665700289814, + "learning_rate": 4.6829268292682926e-05, + "loss": 1.289, + "step": 96 + }, + { + "epoch": 0.18, + "grad_norm": 0.1719900691262439, + "learning_rate": 4.731707317073171e-05, + "loss": 1.1626, + "step": 97 + }, + { + "epoch": 0.18, + "grad_norm": 0.17937994168039778, + "learning_rate": 4.780487804878049e-05, + "loss": 1.175, + "step": 98 + }, + { + "epoch": 0.18, + "grad_norm": 0.16631851422106986, + "learning_rate": 4.829268292682927e-05, + "loss": 1.2177, + "step": 99 + }, + { + "epoch": 0.18, + "grad_norm": 0.19143696232800309, + "learning_rate": 4.878048780487805e-05, + "loss": 1.3071, + "step": 100 + }, + { + "epoch": 0.18, + "grad_norm": 0.17859506638780318, + "learning_rate": 4.9268292682926835e-05, + "loss": 1.2351, + "step": 101 + }, + { + "epoch": 0.19, + "grad_norm": 0.18381520321248196, + "learning_rate": 4.975609756097561e-05, + "loss": 1.2342, + "step": 102 + }, + { + "epoch": 0.19, + "grad_norm": 0.17968218683773912, + "learning_rate": 5.0243902439024394e-05, + "loss": 1.2074, + "step": 103 + }, + { + "epoch": 0.19, + "grad_norm": 0.18139489969339018, + "learning_rate": 5.073170731707318e-05, + "loss": 1.1558, + "step": 104 + }, + { + "epoch": 0.19, + "grad_norm": 0.17366624842514394, + "learning_rate": 5.121951219512195e-05, + "loss": 1.1897, + "step": 105 + }, + { + "epoch": 0.19, + "grad_norm": 0.16034845455223745, + "learning_rate": 5.1707317073170736e-05, + "loss": 1.179, + "step": 106 + }, + { + "epoch": 0.2, + "grad_norm": 0.17583069577827776, + "learning_rate": 5.219512195121952e-05, + "loss": 1.1856, + "step": 107 + }, + { + "epoch": 0.2, + "grad_norm": 0.1853758076989552, + "learning_rate": 5.26829268292683e-05, + "loss": 1.2072, + "step": 108 + }, + { + "epoch": 0.2, + "grad_norm": 0.19597443965936462, + "learning_rate": 5.317073170731708e-05, + "loss": 1.2271, + "step": 109 + }, + { + "epoch": 0.2, + "grad_norm": 0.1899206334098331, + "learning_rate": 5.365853658536586e-05, + "loss": 1.1961, + "step": 110 + }, + { + "epoch": 0.2, + "grad_norm": 0.17463763837757018, + "learning_rate": 5.4146341463414645e-05, + "loss": 1.2049, + "step": 111 + }, + { + "epoch": 0.2, + "grad_norm": 0.20431371701229986, + "learning_rate": 5.463414634146342e-05, + "loss": 1.2891, + "step": 112 + }, + { + "epoch": 0.21, + "grad_norm": 0.1814475107638498, + "learning_rate": 5.51219512195122e-05, + "loss": 1.2346, + "step": 113 + }, + { + "epoch": 0.21, + "grad_norm": 0.1883849423207823, + "learning_rate": 5.5609756097560974e-05, + "loss": 1.244, + "step": 114 + }, + { + "epoch": 0.21, + "grad_norm": 0.1857258128640568, + "learning_rate": 5.609756097560976e-05, + "loss": 1.2669, + "step": 115 + }, + { + "epoch": 0.21, + "grad_norm": 0.1740768514118401, + "learning_rate": 5.658536585365854e-05, + "loss": 1.2414, + "step": 116 + }, + { + "epoch": 0.21, + "grad_norm": 0.1919320335584178, + "learning_rate": 5.7073170731707317e-05, + "loss": 1.2886, + "step": 117 + }, + { + "epoch": 0.22, + "grad_norm": 0.18288775167828136, + "learning_rate": 5.75609756097561e-05, + "loss": 1.1875, + "step": 118 + }, + { + "epoch": 0.22, + "grad_norm": 0.18208588867750863, + "learning_rate": 5.804878048780488e-05, + "loss": 1.2388, + "step": 119 + }, + { + "epoch": 0.22, + "grad_norm": 0.1743260015658331, + "learning_rate": 5.853658536585366e-05, + "loss": 1.1762, + "step": 120 + }, + { + "epoch": 0.22, + "grad_norm": 0.17856046291517946, + "learning_rate": 5.902439024390244e-05, + "loss": 1.2888, + "step": 121 + }, + { + "epoch": 0.22, + "grad_norm": 0.17493794870966536, + "learning_rate": 5.9512195121951225e-05, + "loss": 1.2222, + "step": 122 + }, + { + "epoch": 0.22, + "grad_norm": 0.1909202655203384, + "learning_rate": 6.000000000000001e-05, + "loss": 1.2414, + "step": 123 + }, + { + "epoch": 0.23, + "grad_norm": 0.18345819482834988, + "learning_rate": 6.0487804878048785e-05, + "loss": 1.2756, + "step": 124 + }, + { + "epoch": 0.23, + "grad_norm": 0.2057069352956621, + "learning_rate": 6.097560975609757e-05, + "loss": 1.261, + "step": 125 + }, + { + "epoch": 0.23, + "grad_norm": 0.299775882469108, + "learning_rate": 6.146341463414634e-05, + "loss": 1.2566, + "step": 126 + }, + { + "epoch": 0.23, + "grad_norm": 0.1869687633018095, + "learning_rate": 6.195121951219513e-05, + "loss": 1.3039, + "step": 127 + }, + { + "epoch": 0.23, + "grad_norm": 0.17747149926197442, + "learning_rate": 6.243902439024391e-05, + "loss": 1.2524, + "step": 128 + }, + { + "epoch": 0.24, + "grad_norm": 0.17885157788044242, + "learning_rate": 6.29268292682927e-05, + "loss": 1.2455, + "step": 129 + }, + { + "epoch": 0.24, + "grad_norm": 0.17617298187845123, + "learning_rate": 6.341463414634148e-05, + "loss": 1.2009, + "step": 130 + }, + { + "epoch": 0.24, + "grad_norm": 0.20164176323497066, + "learning_rate": 6.390243902439025e-05, + "loss": 1.2634, + "step": 131 + }, + { + "epoch": 0.24, + "grad_norm": 0.20459903417307612, + "learning_rate": 6.439024390243903e-05, + "loss": 1.1963, + "step": 132 + }, + { + "epoch": 0.24, + "grad_norm": 0.1863755486334296, + "learning_rate": 6.487804878048781e-05, + "loss": 1.2387, + "step": 133 + }, + { + "epoch": 0.25, + "grad_norm": 0.19265866140295207, + "learning_rate": 6.536585365853658e-05, + "loss": 1.2688, + "step": 134 + }, + { + "epoch": 0.25, + "grad_norm": 0.1823425868969493, + "learning_rate": 6.585365853658536e-05, + "loss": 1.2041, + "step": 135 + }, + { + "epoch": 0.25, + "grad_norm": 0.2016853266472781, + "learning_rate": 6.634146341463415e-05, + "loss": 1.1223, + "step": 136 + }, + { + "epoch": 0.25, + "grad_norm": 0.17282675192463448, + "learning_rate": 6.682926829268293e-05, + "loss": 1.1879, + "step": 137 + }, + { + "epoch": 0.25, + "grad_norm": 0.17398811693399288, + "learning_rate": 6.731707317073171e-05, + "loss": 1.2682, + "step": 138 + }, + { + "epoch": 0.25, + "grad_norm": 0.18516916965434696, + "learning_rate": 6.78048780487805e-05, + "loss": 1.1666, + "step": 139 + }, + { + "epoch": 0.26, + "grad_norm": 0.1852213129647933, + "learning_rate": 6.829268292682927e-05, + "loss": 1.2501, + "step": 140 + }, + { + "epoch": 0.26, + "grad_norm": 0.17915948766591883, + "learning_rate": 6.878048780487805e-05, + "loss": 1.2264, + "step": 141 + }, + { + "epoch": 0.26, + "grad_norm": 0.21599939417233183, + "learning_rate": 6.926829268292683e-05, + "loss": 1.2376, + "step": 142 + }, + { + "epoch": 0.26, + "grad_norm": 0.17839304459521851, + "learning_rate": 6.975609756097562e-05, + "loss": 1.2353, + "step": 143 + }, + { + "epoch": 0.26, + "grad_norm": 0.20826913231380875, + "learning_rate": 7.02439024390244e-05, + "loss": 1.1901, + "step": 144 + }, + { + "epoch": 0.27, + "grad_norm": 0.20788894913361589, + "learning_rate": 7.073170731707318e-05, + "loss": 1.2577, + "step": 145 + }, + { + "epoch": 0.27, + "grad_norm": 0.18420055842301297, + "learning_rate": 7.121951219512195e-05, + "loss": 1.1393, + "step": 146 + }, + { + "epoch": 0.27, + "grad_norm": 0.19903048468685589, + "learning_rate": 7.170731707317073e-05, + "loss": 1.2321, + "step": 147 + }, + { + "epoch": 0.27, + "grad_norm": 0.19074116314985748, + "learning_rate": 7.219512195121952e-05, + "loss": 1.1912, + "step": 148 + }, + { + "epoch": 0.27, + "grad_norm": 0.2353816469403903, + "learning_rate": 7.26829268292683e-05, + "loss": 1.28, + "step": 149 + }, + { + "epoch": 0.27, + "grad_norm": 0.21634875684769345, + "learning_rate": 7.317073170731708e-05, + "loss": 1.3312, + "step": 150 + }, + { + "epoch": 0.28, + "grad_norm": 0.18290969006743918, + "learning_rate": 7.365853658536587e-05, + "loss": 1.2214, + "step": 151 + }, + { + "epoch": 0.28, + "grad_norm": 0.18484243897545208, + "learning_rate": 7.414634146341465e-05, + "loss": 1.1895, + "step": 152 + }, + { + "epoch": 0.28, + "grad_norm": 0.21882343112978872, + "learning_rate": 7.463414634146342e-05, + "loss": 1.2219, + "step": 153 + }, + { + "epoch": 0.28, + "grad_norm": 0.19868284379241205, + "learning_rate": 7.51219512195122e-05, + "loss": 1.2176, + "step": 154 + }, + { + "epoch": 0.28, + "grad_norm": 0.20912516312950613, + "learning_rate": 7.560975609756097e-05, + "loss": 1.242, + "step": 155 + }, + { + "epoch": 0.29, + "grad_norm": 0.23811880045549916, + "learning_rate": 7.609756097560976e-05, + "loss": 1.2838, + "step": 156 + }, + { + "epoch": 0.29, + "grad_norm": 0.19511077122033713, + "learning_rate": 7.658536585365854e-05, + "loss": 1.1594, + "step": 157 + }, + { + "epoch": 0.29, + "grad_norm": 0.20094129399534238, + "learning_rate": 7.707317073170732e-05, + "loss": 1.2966, + "step": 158 + }, + { + "epoch": 0.29, + "grad_norm": 0.19366245038292418, + "learning_rate": 7.75609756097561e-05, + "loss": 1.2246, + "step": 159 + }, + { + "epoch": 0.29, + "grad_norm": 0.19409570223867306, + "learning_rate": 7.804878048780489e-05, + "loss": 1.2312, + "step": 160 + }, + { + "epoch": 0.29, + "grad_norm": 0.2087258457033805, + "learning_rate": 7.853658536585366e-05, + "loss": 1.2169, + "step": 161 + }, + { + "epoch": 0.3, + "grad_norm": 0.18765223996270428, + "learning_rate": 7.902439024390244e-05, + "loss": 1.2383, + "step": 162 + }, + { + "epoch": 0.3, + "grad_norm": 0.20734180224147242, + "learning_rate": 7.951219512195122e-05, + "loss": 1.2587, + "step": 163 + }, + { + "epoch": 0.3, + "grad_norm": 0.24690929540287834, + "learning_rate": 8e-05, + "loss": 1.1951, + "step": 164 + }, + { + "epoch": 0.3, + "grad_norm": 0.2003538797619543, + "learning_rate": 7.999990914797545e-05, + "loss": 1.1982, + "step": 165 + }, + { + "epoch": 0.3, + "grad_norm": 0.22469075613510484, + "learning_rate": 7.99996365923145e-05, + "loss": 1.2355, + "step": 166 + }, + { + "epoch": 0.31, + "grad_norm": 0.21870100788336058, + "learning_rate": 7.999918233425526e-05, + "loss": 1.1103, + "step": 167 + }, + { + "epoch": 0.31, + "grad_norm": 0.20939989594131886, + "learning_rate": 7.999854637586122e-05, + "loss": 1.1966, + "step": 168 + }, + { + "epoch": 0.31, + "grad_norm": 0.43108211416237796, + "learning_rate": 7.999772872002132e-05, + "loss": 1.2882, + "step": 169 + }, + { + "epoch": 0.31, + "grad_norm": 0.27045413432174487, + "learning_rate": 7.999672937044984e-05, + "loss": 1.2399, + "step": 170 + }, + { + "epoch": 0.31, + "grad_norm": 0.19700483036740515, + "learning_rate": 7.999554833168642e-05, + "loss": 1.202, + "step": 171 + }, + { + "epoch": 0.31, + "grad_norm": 0.3335979493370708, + "learning_rate": 7.999418560909604e-05, + "loss": 1.1995, + "step": 172 + }, + { + "epoch": 0.32, + "grad_norm": 0.3165803974474567, + "learning_rate": 7.999264120886902e-05, + "loss": 1.1569, + "step": 173 + }, + { + "epoch": 0.32, + "grad_norm": 0.1951699080346223, + "learning_rate": 7.999091513802093e-05, + "loss": 1.1778, + "step": 174 + }, + { + "epoch": 0.32, + "grad_norm": 0.2087559121749787, + "learning_rate": 7.998900740439265e-05, + "loss": 1.1736, + "step": 175 + }, + { + "epoch": 0.32, + "grad_norm": 0.20345180977460478, + "learning_rate": 7.998691801665024e-05, + "loss": 1.2281, + "step": 176 + }, + { + "epoch": 0.32, + "grad_norm": 0.24617644827252333, + "learning_rate": 7.998464698428495e-05, + "loss": 1.2072, + "step": 177 + }, + { + "epoch": 0.33, + "grad_norm": 0.2469050959356265, + "learning_rate": 7.998219431761318e-05, + "loss": 1.2242, + "step": 178 + }, + { + "epoch": 0.33, + "grad_norm": 0.19529317748460623, + "learning_rate": 7.997956002777642e-05, + "loss": 1.2567, + "step": 179 + }, + { + "epoch": 0.33, + "grad_norm": 0.19048389491381376, + "learning_rate": 7.99767441267412e-05, + "loss": 1.2982, + "step": 180 + }, + { + "epoch": 0.33, + "grad_norm": 0.2085799116493225, + "learning_rate": 7.997374662729904e-05, + "loss": 1.1254, + "step": 181 + }, + { + "epoch": 0.33, + "grad_norm": 0.20636853256378995, + "learning_rate": 7.997056754306636e-05, + "loss": 1.2435, + "step": 182 + }, + { + "epoch": 0.33, + "grad_norm": 0.20590016382290252, + "learning_rate": 7.99672068884845e-05, + "loss": 1.2658, + "step": 183 + }, + { + "epoch": 0.34, + "grad_norm": 0.1931166169764433, + "learning_rate": 7.996366467881955e-05, + "loss": 1.1637, + "step": 184 + }, + { + "epoch": 0.34, + "grad_norm": 0.18873318157988098, + "learning_rate": 7.995994093016237e-05, + "loss": 1.1335, + "step": 185 + }, + { + "epoch": 0.34, + "grad_norm": 0.19210254625199108, + "learning_rate": 7.995603565942846e-05, + "loss": 1.1928, + "step": 186 + }, + { + "epoch": 0.34, + "grad_norm": 0.2130986479765664, + "learning_rate": 7.995194888435792e-05, + "loss": 1.2158, + "step": 187 + }, + { + "epoch": 0.34, + "grad_norm": 0.22003854501814088, + "learning_rate": 7.994768062351532e-05, + "loss": 1.2288, + "step": 188 + }, + { + "epoch": 0.35, + "grad_norm": 0.20330803191993058, + "learning_rate": 7.994323089628968e-05, + "loss": 1.2426, + "step": 189 + }, + { + "epoch": 0.35, + "grad_norm": 0.20567314642208634, + "learning_rate": 7.993859972289434e-05, + "loss": 1.2649, + "step": 190 + }, + { + "epoch": 0.35, + "grad_norm": 0.21556663727342962, + "learning_rate": 7.993378712436686e-05, + "loss": 1.2545, + "step": 191 + }, + { + "epoch": 0.35, + "grad_norm": 0.20309165469109888, + "learning_rate": 7.992879312256897e-05, + "loss": 1.3338, + "step": 192 + }, + { + "epoch": 0.35, + "grad_norm": 0.19574356669421325, + "learning_rate": 7.992361774018641e-05, + "loss": 1.278, + "step": 193 + }, + { + "epoch": 0.35, + "grad_norm": 0.2763613746722313, + "learning_rate": 7.991826100072891e-05, + "loss": 1.2571, + "step": 194 + }, + { + "epoch": 0.36, + "grad_norm": 0.19346552479915102, + "learning_rate": 7.991272292852996e-05, + "loss": 1.2027, + "step": 195 + }, + { + "epoch": 0.36, + "grad_norm": 0.2281167812123908, + "learning_rate": 7.990700354874683e-05, + "loss": 1.2586, + "step": 196 + }, + { + "epoch": 0.36, + "grad_norm": 0.19699013712137542, + "learning_rate": 7.990110288736042e-05, + "loss": 1.1371, + "step": 197 + }, + { + "epoch": 0.36, + "grad_norm": 0.21768209981475933, + "learning_rate": 7.989502097117503e-05, + "loss": 1.2522, + "step": 198 + }, + { + "epoch": 0.36, + "grad_norm": 0.21335427847754582, + "learning_rate": 7.988875782781838e-05, + "loss": 1.2437, + "step": 199 + }, + { + "epoch": 0.37, + "grad_norm": 0.21856710629066897, + "learning_rate": 7.988231348574147e-05, + "loss": 1.2135, + "step": 200 + }, + { + "epoch": 0.37, + "grad_norm": 0.20482062658774797, + "learning_rate": 7.987568797421836e-05, + "loss": 1.1755, + "step": 201 + }, + { + "epoch": 0.37, + "grad_norm": 0.2017756813960897, + "learning_rate": 7.986888132334608e-05, + "loss": 1.1699, + "step": 202 + }, + { + "epoch": 0.37, + "grad_norm": 0.20496443848153809, + "learning_rate": 7.986189356404458e-05, + "loss": 1.2125, + "step": 203 + }, + { + "epoch": 0.37, + "grad_norm": 0.2134603800558358, + "learning_rate": 7.985472472805643e-05, + "loss": 1.2391, + "step": 204 + }, + { + "epoch": 0.37, + "grad_norm": 0.2364175573420861, + "learning_rate": 7.98473748479468e-05, + "loss": 1.2384, + "step": 205 + }, + { + "epoch": 0.38, + "grad_norm": 0.1872419861598724, + "learning_rate": 7.983984395710326e-05, + "loss": 1.1457, + "step": 206 + }, + { + "epoch": 0.38, + "grad_norm": 0.28222194007095774, + "learning_rate": 7.983213208973566e-05, + "loss": 1.2952, + "step": 207 + }, + { + "epoch": 0.38, + "grad_norm": 0.1916094851162064, + "learning_rate": 7.982423928087593e-05, + "loss": 1.1763, + "step": 208 + }, + { + "epoch": 0.38, + "grad_norm": 0.18446245256166657, + "learning_rate": 7.981616556637795e-05, + "loss": 1.1863, + "step": 209 + }, + { + "epoch": 0.38, + "grad_norm": 0.195191961022491, + "learning_rate": 7.980791098291737e-05, + "loss": 1.2036, + "step": 210 + }, + { + "epoch": 0.39, + "grad_norm": 0.2652439657825496, + "learning_rate": 7.979947556799151e-05, + "loss": 1.2834, + "step": 211 + }, + { + "epoch": 0.39, + "grad_norm": 0.24308438957843412, + "learning_rate": 7.979085935991906e-05, + "loss": 1.234, + "step": 212 + }, + { + "epoch": 0.39, + "grad_norm": 0.21294701043622016, + "learning_rate": 7.978206239784004e-05, + "loss": 1.3006, + "step": 213 + }, + { + "epoch": 0.39, + "grad_norm": 0.25809277041859524, + "learning_rate": 7.977308472171553e-05, + "loss": 1.2272, + "step": 214 + }, + { + "epoch": 0.39, + "grad_norm": 0.193463860107294, + "learning_rate": 7.976392637232754e-05, + "loss": 1.2295, + "step": 215 + }, + { + "epoch": 0.4, + "grad_norm": 0.2150023760609626, + "learning_rate": 7.975458739127877e-05, + "loss": 1.2135, + "step": 216 + }, + { + "epoch": 0.4, + "grad_norm": 0.22590495955605894, + "learning_rate": 7.974506782099253e-05, + "loss": 1.2532, + "step": 217 + }, + { + "epoch": 0.4, + "grad_norm": 0.21023744668403702, + "learning_rate": 7.973536770471242e-05, + "loss": 1.2472, + "step": 218 + }, + { + "epoch": 0.4, + "grad_norm": 0.2345749799511543, + "learning_rate": 7.972548708650218e-05, + "loss": 1.1791, + "step": 219 + }, + { + "epoch": 0.4, + "grad_norm": 0.2158876734005217, + "learning_rate": 7.971542601124553e-05, + "loss": 1.2483, + "step": 220 + }, + { + "epoch": 0.4, + "grad_norm": 0.29455339949432446, + "learning_rate": 7.970518452464593e-05, + "loss": 1.2894, + "step": 221 + }, + { + "epoch": 0.41, + "grad_norm": 0.23983708730626851, + "learning_rate": 7.969476267322636e-05, + "loss": 1.271, + "step": 222 + }, + { + "epoch": 0.41, + "grad_norm": 0.1922400905426158, + "learning_rate": 7.968416050432912e-05, + "loss": 1.2139, + "step": 223 + }, + { + "epoch": 0.41, + "grad_norm": 0.2238136844422931, + "learning_rate": 7.967337806611568e-05, + "loss": 1.2655, + "step": 224 + }, + { + "epoch": 0.41, + "grad_norm": 0.21230292828267672, + "learning_rate": 7.966241540756631e-05, + "loss": 1.2406, + "step": 225 + }, + { + "epoch": 0.41, + "grad_norm": 0.26656119419070456, + "learning_rate": 7.965127257848004e-05, + "loss": 1.2595, + "step": 226 + }, + { + "epoch": 0.42, + "grad_norm": 0.22381385502992684, + "learning_rate": 7.963994962947426e-05, + "loss": 1.1737, + "step": 227 + }, + { + "epoch": 0.42, + "grad_norm": 0.20056702203994298, + "learning_rate": 7.962844661198462e-05, + "loss": 1.1969, + "step": 228 + }, + { + "epoch": 0.42, + "grad_norm": 0.20148701321526885, + "learning_rate": 7.961676357826478e-05, + "loss": 1.2151, + "step": 229 + }, + { + "epoch": 0.42, + "grad_norm": 0.20034834807028637, + "learning_rate": 7.960490058138604e-05, + "loss": 1.1455, + "step": 230 + }, + { + "epoch": 0.42, + "grad_norm": 0.21050838521846033, + "learning_rate": 7.959285767523732e-05, + "loss": 1.2223, + "step": 231 + }, + { + "epoch": 0.42, + "grad_norm": 0.20904772138969777, + "learning_rate": 7.95806349145247e-05, + "loss": 1.2534, + "step": 232 + }, + { + "epoch": 0.43, + "grad_norm": 0.20307877304792957, + "learning_rate": 7.956823235477134e-05, + "loss": 1.1352, + "step": 233 + }, + { + "epoch": 0.43, + "grad_norm": 0.20501105270897094, + "learning_rate": 7.95556500523171e-05, + "loss": 1.2031, + "step": 234 + }, + { + "epoch": 0.43, + "grad_norm": 0.19800586972038586, + "learning_rate": 7.954288806431838e-05, + "loss": 1.2567, + "step": 235 + }, + { + "epoch": 0.43, + "grad_norm": 0.2175102450594135, + "learning_rate": 7.952994644874777e-05, + "loss": 1.2538, + "step": 236 + }, + { + "epoch": 0.43, + "grad_norm": 0.22698189300067595, + "learning_rate": 7.951682526439391e-05, + "loss": 1.3088, + "step": 237 + }, + { + "epoch": 0.44, + "grad_norm": 0.19208392014975315, + "learning_rate": 7.950352457086109e-05, + "loss": 1.2336, + "step": 238 + }, + { + "epoch": 0.44, + "grad_norm": 0.27004086334319655, + "learning_rate": 7.949004442856905e-05, + "loss": 1.2012, + "step": 239 + }, + { + "epoch": 0.44, + "grad_norm": 0.23420974954538043, + "learning_rate": 7.947638489875272e-05, + "loss": 1.2244, + "step": 240 + }, + { + "epoch": 0.44, + "grad_norm": 0.20514399124802024, + "learning_rate": 7.946254604346186e-05, + "loss": 1.2548, + "step": 241 + }, + { + "epoch": 0.44, + "grad_norm": 0.19334973602372896, + "learning_rate": 7.944852792556092e-05, + "loss": 1.2104, + "step": 242 + }, + { + "epoch": 0.44, + "grad_norm": 0.1992640714537956, + "learning_rate": 7.943433060872858e-05, + "loss": 1.2628, + "step": 243 + }, + { + "epoch": 0.45, + "grad_norm": 0.203284617090413, + "learning_rate": 7.941995415745761e-05, + "loss": 1.2002, + "step": 244 + }, + { + "epoch": 0.45, + "grad_norm": 0.22795306969682058, + "learning_rate": 7.94053986370545e-05, + "loss": 1.2215, + "step": 245 + }, + { + "epoch": 0.45, + "grad_norm": 0.20789041346838505, + "learning_rate": 7.939066411363915e-05, + "loss": 1.0998, + "step": 246 + }, + { + "epoch": 0.45, + "grad_norm": 0.22354868884742066, + "learning_rate": 7.937575065414464e-05, + "loss": 1.2564, + "step": 247 + }, + { + "epoch": 0.45, + "grad_norm": 0.21176392726647736, + "learning_rate": 7.936065832631687e-05, + "loss": 1.2816, + "step": 248 + }, + { + "epoch": 0.46, + "grad_norm": 0.19967179557235587, + "learning_rate": 7.934538719871427e-05, + "loss": 1.1961, + "step": 249 + }, + { + "epoch": 0.46, + "grad_norm": 0.210819577350627, + "learning_rate": 7.932993734070747e-05, + "loss": 1.2167, + "step": 250 + }, + { + "epoch": 0.46, + "grad_norm": 0.21537794551756187, + "learning_rate": 7.931430882247903e-05, + "loss": 1.2341, + "step": 251 + }, + { + "epoch": 0.46, + "grad_norm": 0.22850872387256574, + "learning_rate": 7.929850171502304e-05, + "loss": 1.1686, + "step": 252 + }, + { + "epoch": 0.46, + "grad_norm": 0.22380366415076383, + "learning_rate": 7.928251609014493e-05, + "loss": 1.1462, + "step": 253 + }, + { + "epoch": 0.46, + "grad_norm": 0.22426923149036065, + "learning_rate": 7.926635202046102e-05, + "loss": 1.1792, + "step": 254 + }, + { + "epoch": 0.47, + "grad_norm": 0.42082703321103965, + "learning_rate": 7.925000957939822e-05, + "loss": 1.2718, + "step": 255 + }, + { + "epoch": 0.47, + "grad_norm": 0.2235432774854074, + "learning_rate": 7.92334888411937e-05, + "loss": 1.2598, + "step": 256 + }, + { + "epoch": 0.47, + "grad_norm": 0.281644028934108, + "learning_rate": 7.92167898808946e-05, + "loss": 1.2205, + "step": 257 + }, + { + "epoch": 0.47, + "grad_norm": 0.2037705143888748, + "learning_rate": 7.919991277435763e-05, + "loss": 1.1737, + "step": 258 + }, + { + "epoch": 0.47, + "grad_norm": 0.20917419230028977, + "learning_rate": 7.918285759824879e-05, + "loss": 1.2035, + "step": 259 + }, + { + "epoch": 0.48, + "grad_norm": 0.20510847570635518, + "learning_rate": 7.916562443004292e-05, + "loss": 1.2135, + "step": 260 + }, + { + "epoch": 0.48, + "grad_norm": 0.25172483071092466, + "learning_rate": 7.914821334802342e-05, + "loss": 1.2218, + "step": 261 + }, + { + "epoch": 0.48, + "grad_norm": 0.21102706700634313, + "learning_rate": 7.91306244312819e-05, + "loss": 1.1738, + "step": 262 + }, + { + "epoch": 0.48, + "grad_norm": 0.22626060872645815, + "learning_rate": 7.911285775971781e-05, + "loss": 1.238, + "step": 263 + }, + { + "epoch": 0.48, + "grad_norm": 0.22448567539778486, + "learning_rate": 7.909491341403805e-05, + "loss": 1.2404, + "step": 264 + }, + { + "epoch": 0.48, + "grad_norm": 0.2019099786139193, + "learning_rate": 7.907679147575661e-05, + "loss": 1.213, + "step": 265 + }, + { + "epoch": 0.49, + "grad_norm": 0.24307234839096267, + "learning_rate": 7.905849202719422e-05, + "loss": 1.2322, + "step": 266 + }, + { + "epoch": 0.49, + "grad_norm": 0.19801890521743487, + "learning_rate": 7.904001515147802e-05, + "loss": 1.2448, + "step": 267 + }, + { + "epoch": 0.49, + "grad_norm": 0.2102742273575385, + "learning_rate": 7.902136093254106e-05, + "loss": 1.1657, + "step": 268 + }, + { + "epoch": 0.49, + "grad_norm": 0.2173464476815016, + "learning_rate": 7.900252945512201e-05, + "loss": 1.2549, + "step": 269 + }, + { + "epoch": 0.49, + "grad_norm": 0.20957275458699595, + "learning_rate": 7.898352080476479e-05, + "loss": 1.2536, + "step": 270 + }, + { + "epoch": 0.5, + "grad_norm": 0.20691966388952363, + "learning_rate": 7.896433506781811e-05, + "loss": 1.2661, + "step": 271 + }, + { + "epoch": 0.5, + "grad_norm": 0.2276662275112648, + "learning_rate": 7.894497233143509e-05, + "loss": 1.2409, + "step": 272 + }, + { + "epoch": 0.5, + "grad_norm": 0.23854109569301263, + "learning_rate": 7.892543268357297e-05, + "loss": 1.2681, + "step": 273 + }, + { + "epoch": 0.5, + "grad_norm": 0.2233864156677627, + "learning_rate": 7.890571621299252e-05, + "loss": 1.1687, + "step": 274 + }, + { + "epoch": 0.5, + "grad_norm": 0.20114129147925475, + "learning_rate": 7.888582300925787e-05, + "loss": 1.2184, + "step": 275 + }, + { + "epoch": 0.5, + "grad_norm": 0.2154654670569462, + "learning_rate": 7.886575316273586e-05, + "loss": 1.1982, + "step": 276 + }, + { + "epoch": 0.51, + "grad_norm": 0.2292982209343639, + "learning_rate": 7.884550676459583e-05, + "loss": 1.2129, + "step": 277 + }, + { + "epoch": 0.51, + "grad_norm": 0.21302713135229548, + "learning_rate": 7.882508390680908e-05, + "loss": 1.1605, + "step": 278 + }, + { + "epoch": 0.51, + "grad_norm": 0.2123661020671048, + "learning_rate": 7.88044846821485e-05, + "loss": 1.2308, + "step": 279 + }, + { + "epoch": 0.51, + "grad_norm": 0.2080577410800404, + "learning_rate": 7.878370918418818e-05, + "loss": 1.2195, + "step": 280 + }, + { + "epoch": 0.51, + "grad_norm": 0.19663901881127385, + "learning_rate": 7.876275750730289e-05, + "loss": 1.1591, + "step": 281 + }, + { + "epoch": 0.52, + "grad_norm": 0.20534502031312163, + "learning_rate": 7.874162974666776e-05, + "loss": 1.2664, + "step": 282 + }, + { + "epoch": 0.52, + "grad_norm": 0.23240445399513837, + "learning_rate": 7.872032599825779e-05, + "loss": 1.2151, + "step": 283 + }, + { + "epoch": 0.52, + "grad_norm": 0.2672527316717507, + "learning_rate": 7.86988463588474e-05, + "loss": 1.2406, + "step": 284 + }, + { + "epoch": 0.52, + "grad_norm": 0.19893903058743695, + "learning_rate": 7.867719092601003e-05, + "loss": 1.1291, + "step": 285 + }, + { + "epoch": 0.52, + "grad_norm": 0.33275268109930917, + "learning_rate": 7.865535979811768e-05, + "loss": 1.1406, + "step": 286 + }, + { + "epoch": 0.52, + "grad_norm": 0.2373619455690358, + "learning_rate": 7.863335307434045e-05, + "loss": 1.2799, + "step": 287 + }, + { + "epoch": 0.53, + "grad_norm": 0.263235735390858, + "learning_rate": 7.861117085464612e-05, + "loss": 1.2415, + "step": 288 + }, + { + "epoch": 0.53, + "grad_norm": 0.25884281780784324, + "learning_rate": 7.858881323979965e-05, + "loss": 1.3919, + "step": 289 + }, + { + "epoch": 0.53, + "grad_norm": 0.25426288332255736, + "learning_rate": 7.85662803313628e-05, + "loss": 1.174, + "step": 290 + }, + { + "epoch": 0.53, + "grad_norm": 0.26655405527881243, + "learning_rate": 7.854357223169356e-05, + "loss": 1.2806, + "step": 291 + }, + { + "epoch": 0.53, + "grad_norm": 0.20909844432349833, + "learning_rate": 7.852068904394579e-05, + "loss": 1.2627, + "step": 292 + }, + { + "epoch": 0.54, + "grad_norm": 0.21307115068935759, + "learning_rate": 7.849763087206866e-05, + "loss": 1.1879, + "step": 293 + }, + { + "epoch": 0.54, + "grad_norm": 0.25009949471398946, + "learning_rate": 7.847439782080628e-05, + "loss": 1.2881, + "step": 294 + }, + { + "epoch": 0.54, + "grad_norm": 0.20960783418679174, + "learning_rate": 7.845098999569712e-05, + "loss": 1.2723, + "step": 295 + }, + { + "epoch": 0.54, + "grad_norm": 0.24968832437925104, + "learning_rate": 7.842740750307362e-05, + "loss": 1.2029, + "step": 296 + }, + { + "epoch": 0.54, + "grad_norm": 0.22981196585125677, + "learning_rate": 7.84036504500616e-05, + "loss": 1.1695, + "step": 297 + }, + { + "epoch": 0.55, + "grad_norm": 0.2320606844751365, + "learning_rate": 7.837971894457991e-05, + "loss": 1.2317, + "step": 298 + }, + { + "epoch": 0.55, + "grad_norm": 0.23051459673906124, + "learning_rate": 7.835561309533981e-05, + "loss": 1.2046, + "step": 299 + }, + { + "epoch": 0.55, + "grad_norm": 0.2510027231060586, + "learning_rate": 7.833133301184457e-05, + "loss": 1.199, + "step": 300 + }, + { + "epoch": 0.55, + "grad_norm": 0.23601180466018787, + "learning_rate": 7.830687880438895e-05, + "loss": 1.1755, + "step": 301 + }, + { + "epoch": 0.55, + "grad_norm": 0.24740820934385369, + "learning_rate": 7.828225058405864e-05, + "loss": 1.2054, + "step": 302 + }, + { + "epoch": 0.55, + "grad_norm": 0.23065372979111173, + "learning_rate": 7.825744846272984e-05, + "loss": 1.2066, + "step": 303 + }, + { + "epoch": 0.56, + "grad_norm": 0.22385077334838213, + "learning_rate": 7.823247255306866e-05, + "loss": 1.2147, + "step": 304 + }, + { + "epoch": 0.56, + "grad_norm": 0.42981213948386104, + "learning_rate": 7.820732296853074e-05, + "loss": 1.2314, + "step": 305 + }, + { + "epoch": 0.56, + "grad_norm": 0.21122844902751076, + "learning_rate": 7.818199982336058e-05, + "loss": 1.1462, + "step": 306 + }, + { + "epoch": 0.56, + "grad_norm": 0.23374869692118933, + "learning_rate": 7.815650323259117e-05, + "loss": 1.2051, + "step": 307 + }, + { + "epoch": 0.56, + "grad_norm": 0.21662363795962128, + "learning_rate": 7.813083331204332e-05, + "loss": 1.1575, + "step": 308 + }, + { + "epoch": 0.57, + "grad_norm": 0.2088315773384112, + "learning_rate": 7.810499017832526e-05, + "loss": 1.1316, + "step": 309 + }, + { + "epoch": 0.57, + "grad_norm": 0.2095238410730976, + "learning_rate": 7.807897394883203e-05, + "loss": 1.2087, + "step": 310 + }, + { + "epoch": 0.57, + "grad_norm": 0.22672932127256515, + "learning_rate": 7.805278474174499e-05, + "loss": 1.2512, + "step": 311 + }, + { + "epoch": 0.57, + "grad_norm": 0.21873052340922736, + "learning_rate": 7.802642267603126e-05, + "loss": 1.1909, + "step": 312 + }, + { + "epoch": 0.57, + "grad_norm": 0.219814521916342, + "learning_rate": 7.79998878714432e-05, + "loss": 1.1669, + "step": 313 + }, + { + "epoch": 0.57, + "grad_norm": 0.3049426027257317, + "learning_rate": 7.797318044851786e-05, + "loss": 1.1797, + "step": 314 + }, + { + "epoch": 0.58, + "grad_norm": 0.22309435690065985, + "learning_rate": 7.794630052857638e-05, + "loss": 1.1417, + "step": 315 + }, + { + "epoch": 0.58, + "grad_norm": 0.3891885169154885, + "learning_rate": 7.791924823372354e-05, + "loss": 1.2369, + "step": 316 + }, + { + "epoch": 0.58, + "grad_norm": 0.24780269452456372, + "learning_rate": 7.789202368684711e-05, + "loss": 1.2521, + "step": 317 + }, + { + "epoch": 0.58, + "grad_norm": 0.21660460720269362, + "learning_rate": 7.786462701161738e-05, + "loss": 1.2151, + "step": 318 + }, + { + "epoch": 0.58, + "grad_norm": 0.23635409466561857, + "learning_rate": 7.783705833248649e-05, + "loss": 1.2363, + "step": 319 + }, + { + "epoch": 0.59, + "grad_norm": 0.2616135839903218, + "learning_rate": 7.780931777468797e-05, + "loss": 1.2428, + "step": 320 + }, + { + "epoch": 0.59, + "grad_norm": 0.21461059159245083, + "learning_rate": 7.77814054642361e-05, + "loss": 1.1434, + "step": 321 + }, + { + "epoch": 0.59, + "grad_norm": 0.25348824286656163, + "learning_rate": 7.775332152792539e-05, + "loss": 1.2368, + "step": 322 + }, + { + "epoch": 0.59, + "grad_norm": 0.22275034726331247, + "learning_rate": 7.772506609332995e-05, + "loss": 1.1827, + "step": 323 + }, + { + "epoch": 0.59, + "grad_norm": 0.25030821228147526, + "learning_rate": 7.769663928880298e-05, + "loss": 1.2428, + "step": 324 + }, + { + "epoch": 0.59, + "grad_norm": 0.22251804398745534, + "learning_rate": 7.766804124347608e-05, + "loss": 1.1889, + "step": 325 + }, + { + "epoch": 0.6, + "grad_norm": 0.23381455520411995, + "learning_rate": 7.763927208725879e-05, + "loss": 1.2115, + "step": 326 + }, + { + "epoch": 0.6, + "grad_norm": 0.27341902651946226, + "learning_rate": 7.761033195083791e-05, + "loss": 1.2535, + "step": 327 + }, + { + "epoch": 0.6, + "grad_norm": 0.24862471659814522, + "learning_rate": 7.758122096567694e-05, + "loss": 1.2128, + "step": 328 + }, + { + "epoch": 0.6, + "grad_norm": 0.2251357082045494, + "learning_rate": 7.755193926401547e-05, + "loss": 1.2334, + "step": 329 + }, + { + "epoch": 0.6, + "grad_norm": 0.3173274941622932, + "learning_rate": 7.752248697886857e-05, + "loss": 1.226, + "step": 330 + }, + { + "epoch": 0.61, + "grad_norm": 0.23056440717672175, + "learning_rate": 7.74928642440263e-05, + "loss": 1.2339, + "step": 331 + }, + { + "epoch": 0.61, + "grad_norm": 0.2801507500859342, + "learning_rate": 7.746307119405286e-05, + "loss": 1.287, + "step": 332 + }, + { + "epoch": 0.61, + "grad_norm": 0.2267818430426272, + "learning_rate": 7.743310796428622e-05, + "loss": 1.1916, + "step": 333 + }, + { + "epoch": 0.61, + "grad_norm": 0.2777329160365585, + "learning_rate": 7.74029746908374e-05, + "loss": 1.252, + "step": 334 + }, + { + "epoch": 0.61, + "grad_norm": 0.25289169762353, + "learning_rate": 7.737267151058983e-05, + "loss": 1.2153, + "step": 335 + }, + { + "epoch": 0.61, + "grad_norm": 0.2424670686901653, + "learning_rate": 7.734219856119875e-05, + "loss": 1.2227, + "step": 336 + }, + { + "epoch": 0.62, + "grad_norm": 0.22747092217441645, + "learning_rate": 7.731155598109067e-05, + "loss": 1.19, + "step": 337 + }, + { + "epoch": 0.62, + "grad_norm": 0.2307810940100189, + "learning_rate": 7.728074390946257e-05, + "loss": 1.1818, + "step": 338 + }, + { + "epoch": 0.62, + "grad_norm": 0.2583402574655623, + "learning_rate": 7.724976248628142e-05, + "loss": 1.1608, + "step": 339 + }, + { + "epoch": 0.62, + "grad_norm": 0.22140209760890694, + "learning_rate": 7.721861185228347e-05, + "loss": 1.1245, + "step": 340 + }, + { + "epoch": 0.62, + "grad_norm": 0.25859310758244686, + "learning_rate": 7.718729214897362e-05, + "loss": 1.2247, + "step": 341 + }, + { + "epoch": 0.63, + "grad_norm": 0.26371179531372124, + "learning_rate": 7.715580351862482e-05, + "loss": 1.2128, + "step": 342 + }, + { + "epoch": 0.63, + "grad_norm": 0.26575541302851047, + "learning_rate": 7.712414610427733e-05, + "loss": 1.2443, + "step": 343 + }, + { + "epoch": 0.63, + "grad_norm": 0.269978305197599, + "learning_rate": 7.709232004973816e-05, + "loss": 1.2231, + "step": 344 + }, + { + "epoch": 0.63, + "grad_norm": 0.26583998705977047, + "learning_rate": 7.70603254995804e-05, + "loss": 1.2476, + "step": 345 + }, + { + "epoch": 0.63, + "grad_norm": 0.24256062164066097, + "learning_rate": 7.702816259914253e-05, + "loss": 1.2901, + "step": 346 + }, + { + "epoch": 0.63, + "grad_norm": 0.3463123472658915, + "learning_rate": 7.699583149452779e-05, + "loss": 1.3277, + "step": 347 + }, + { + "epoch": 0.64, + "grad_norm": 0.2269096590531878, + "learning_rate": 7.696333233260345e-05, + "loss": 1.2047, + "step": 348 + }, + { + "epoch": 0.64, + "grad_norm": 0.25136883001050025, + "learning_rate": 7.693066526100031e-05, + "loss": 1.1619, + "step": 349 + }, + { + "epoch": 0.64, + "grad_norm": 0.2565112571116145, + "learning_rate": 7.68978304281118e-05, + "loss": 1.2389, + "step": 350 + }, + { + "epoch": 0.64, + "grad_norm": 0.22175779550828703, + "learning_rate": 7.686482798309349e-05, + "loss": 1.2238, + "step": 351 + }, + { + "epoch": 0.64, + "grad_norm": 0.22588304332216555, + "learning_rate": 7.683165807586234e-05, + "loss": 1.174, + "step": 352 + }, + { + "epoch": 0.65, + "grad_norm": 0.24889474296529737, + "learning_rate": 7.6798320857096e-05, + "loss": 1.2366, + "step": 353 + }, + { + "epoch": 0.65, + "grad_norm": 0.27339703806525034, + "learning_rate": 7.676481647823214e-05, + "loss": 1.2356, + "step": 354 + }, + { + "epoch": 0.65, + "grad_norm": 0.23424666722888365, + "learning_rate": 7.673114509146782e-05, + "loss": 1.2089, + "step": 355 + }, + { + "epoch": 0.65, + "grad_norm": 0.27978285392461766, + "learning_rate": 7.66973068497587e-05, + "loss": 1.2609, + "step": 356 + }, + { + "epoch": 0.65, + "grad_norm": 0.2509423350138824, + "learning_rate": 7.666330190681844e-05, + "loss": 1.1777, + "step": 357 + }, + { + "epoch": 0.65, + "grad_norm": 0.23007730927468031, + "learning_rate": 7.662913041711793e-05, + "loss": 1.154, + "step": 358 + }, + { + "epoch": 0.66, + "grad_norm": 0.2438648674953112, + "learning_rate": 7.659479253588462e-05, + "loss": 1.2257, + "step": 359 + }, + { + "epoch": 0.66, + "grad_norm": 0.28816093242092233, + "learning_rate": 7.65602884191018e-05, + "loss": 1.2558, + "step": 360 + }, + { + "epoch": 0.66, + "grad_norm": 0.24972815300596035, + "learning_rate": 7.652561822350793e-05, + "loss": 1.2837, + "step": 361 + }, + { + "epoch": 0.66, + "grad_norm": 0.2543189139697063, + "learning_rate": 7.649078210659587e-05, + "loss": 1.2193, + "step": 362 + }, + { + "epoch": 0.66, + "grad_norm": 0.2237937956718952, + "learning_rate": 7.645578022661224e-05, + "loss": 1.2237, + "step": 363 + }, + { + "epoch": 0.67, + "grad_norm": 0.29742029408787396, + "learning_rate": 7.642061274255657e-05, + "loss": 1.2116, + "step": 364 + }, + { + "epoch": 0.67, + "grad_norm": 0.2462883147335493, + "learning_rate": 7.638527981418075e-05, + "loss": 1.1827, + "step": 365 + }, + { + "epoch": 0.67, + "grad_norm": 0.2647802498907096, + "learning_rate": 7.634978160198817e-05, + "loss": 1.2739, + "step": 366 + }, + { + "epoch": 0.67, + "grad_norm": 0.22360398779217264, + "learning_rate": 7.631411826723306e-05, + "loss": 1.2185, + "step": 367 + }, + { + "epoch": 0.67, + "grad_norm": 0.2635048004593543, + "learning_rate": 7.627828997191973e-05, + "loss": 1.2317, + "step": 368 + }, + { + "epoch": 0.67, + "grad_norm": 0.2764803449917684, + "learning_rate": 7.624229687880184e-05, + "loss": 1.1923, + "step": 369 + }, + { + "epoch": 0.68, + "grad_norm": 0.25724943233414527, + "learning_rate": 7.620613915138166e-05, + "loss": 1.2218, + "step": 370 + }, + { + "epoch": 0.68, + "grad_norm": 0.2858318045794755, + "learning_rate": 7.61698169539093e-05, + "loss": 1.1496, + "step": 371 + }, + { + "epoch": 0.68, + "grad_norm": 0.23547216647460364, + "learning_rate": 7.613333045138206e-05, + "loss": 1.1905, + "step": 372 + }, + { + "epoch": 0.68, + "grad_norm": 0.22984814903684375, + "learning_rate": 7.609667980954355e-05, + "loss": 1.2009, + "step": 373 + }, + { + "epoch": 0.68, + "grad_norm": 0.2551903754079084, + "learning_rate": 7.605986519488301e-05, + "loss": 1.2042, + "step": 374 + }, + { + "epoch": 0.69, + "grad_norm": 0.2508257410125616, + "learning_rate": 7.602288677463457e-05, + "loss": 1.2468, + "step": 375 + }, + { + "epoch": 0.69, + "grad_norm": 0.25324577774935964, + "learning_rate": 7.598574471677644e-05, + "loss": 1.2603, + "step": 376 + }, + { + "epoch": 0.69, + "grad_norm": 0.35888776531769967, + "learning_rate": 7.59484391900302e-05, + "loss": 1.1929, + "step": 377 + }, + { + "epoch": 0.69, + "grad_norm": 0.22048517191014724, + "learning_rate": 7.591097036385994e-05, + "loss": 1.1783, + "step": 378 + }, + { + "epoch": 0.69, + "grad_norm": 0.2781160412746083, + "learning_rate": 7.587333840847162e-05, + "loss": 1.3397, + "step": 379 + }, + { + "epoch": 0.7, + "grad_norm": 0.24033046830332258, + "learning_rate": 7.583554349481222e-05, + "loss": 1.2436, + "step": 380 + }, + { + "epoch": 0.7, + "grad_norm": 0.26413762380260003, + "learning_rate": 7.579758579456893e-05, + "loss": 1.1917, + "step": 381 + }, + { + "epoch": 0.7, + "grad_norm": 0.2390937887338632, + "learning_rate": 7.575946548016847e-05, + "loss": 1.2186, + "step": 382 + }, + { + "epoch": 0.7, + "grad_norm": 0.25131263043429275, + "learning_rate": 7.572118272477622e-05, + "loss": 1.2538, + "step": 383 + }, + { + "epoch": 0.7, + "grad_norm": 0.223974104870702, + "learning_rate": 7.568273770229546e-05, + "loss": 1.2165, + "step": 384 + }, + { + "epoch": 0.7, + "grad_norm": 0.25840356830252875, + "learning_rate": 7.564413058736663e-05, + "loss": 1.1848, + "step": 385 + }, + { + "epoch": 0.71, + "grad_norm": 0.2723156683076603, + "learning_rate": 7.560536155536641e-05, + "loss": 1.1982, + "step": 386 + }, + { + "epoch": 0.71, + "grad_norm": 0.265687427976889, + "learning_rate": 7.556643078240708e-05, + "loss": 1.231, + "step": 387 + }, + { + "epoch": 0.71, + "grad_norm": 0.25152762080976077, + "learning_rate": 7.552733844533562e-05, + "loss": 1.1974, + "step": 388 + }, + { + "epoch": 0.71, + "grad_norm": 0.2366049485053541, + "learning_rate": 7.548808472173292e-05, + "loss": 1.3119, + "step": 389 + }, + { + "epoch": 0.71, + "grad_norm": 0.22092196577077122, + "learning_rate": 7.5448669789913e-05, + "loss": 1.195, + "step": 390 + }, + { + "epoch": 0.72, + "grad_norm": 0.22667521540462374, + "learning_rate": 7.540909382892217e-05, + "loss": 1.1431, + "step": 391 + }, + { + "epoch": 0.72, + "grad_norm": 0.25432207282646513, + "learning_rate": 7.536935701853823e-05, + "loss": 1.2173, + "step": 392 + }, + { + "epoch": 0.72, + "grad_norm": 0.29950506457923864, + "learning_rate": 7.53294595392697e-05, + "loss": 1.1962, + "step": 393 + }, + { + "epoch": 0.72, + "grad_norm": 0.24735689607229913, + "learning_rate": 7.528940157235487e-05, + "loss": 1.2053, + "step": 394 + }, + { + "epoch": 0.72, + "grad_norm": 0.24394198607459663, + "learning_rate": 7.524918329976114e-05, + "loss": 1.1979, + "step": 395 + }, + { + "epoch": 0.72, + "grad_norm": 0.2630369372689188, + "learning_rate": 7.520880490418409e-05, + "loss": 1.2111, + "step": 396 + }, + { + "epoch": 0.73, + "grad_norm": 0.26275028416291457, + "learning_rate": 7.516826656904664e-05, + "loss": 1.2133, + "step": 397 + }, + { + "epoch": 0.73, + "grad_norm": 0.23938074620956928, + "learning_rate": 7.512756847849831e-05, + "loss": 1.1355, + "step": 398 + }, + { + "epoch": 0.73, + "grad_norm": 0.3724960610098138, + "learning_rate": 7.508671081741428e-05, + "loss": 1.2572, + "step": 399 + }, + { + "epoch": 0.73, + "grad_norm": 0.24161685847894723, + "learning_rate": 7.504569377139462e-05, + "loss": 1.1706, + "step": 400 + }, + { + "epoch": 0.73, + "grad_norm": 0.26121591322670523, + "learning_rate": 7.50045175267634e-05, + "loss": 1.2135, + "step": 401 + }, + { + "epoch": 0.74, + "grad_norm": 0.2465579498164775, + "learning_rate": 7.496318227056788e-05, + "loss": 1.1641, + "step": 402 + }, + { + "epoch": 0.74, + "grad_norm": 0.2556288696122787, + "learning_rate": 7.492168819057767e-05, + "loss": 1.2939, + "step": 403 + }, + { + "epoch": 0.74, + "grad_norm": 0.261481216336303, + "learning_rate": 7.488003547528382e-05, + "loss": 1.2026, + "step": 404 + }, + { + "epoch": 0.74, + "grad_norm": 0.2389415135676362, + "learning_rate": 7.483822431389799e-05, + "loss": 1.2131, + "step": 405 + }, + { + "epoch": 0.74, + "grad_norm": 0.2559201956627192, + "learning_rate": 7.479625489635162e-05, + "loss": 1.1246, + "step": 406 + }, + { + "epoch": 0.74, + "grad_norm": 0.27127932491822604, + "learning_rate": 7.475412741329504e-05, + "loss": 1.2429, + "step": 407 + }, + { + "epoch": 0.75, + "grad_norm": 0.27006004008695594, + "learning_rate": 7.47118420560966e-05, + "loss": 1.2388, + "step": 408 + }, + { + "epoch": 0.75, + "grad_norm": 0.23716823297200537, + "learning_rate": 7.466939901684182e-05, + "loss": 1.1264, + "step": 409 + }, + { + "epoch": 0.75, + "grad_norm": 0.2885373898669248, + "learning_rate": 7.462679848833252e-05, + "loss": 1.2786, + "step": 410 + }, + { + "epoch": 0.75, + "grad_norm": 0.49215227598639927, + "learning_rate": 7.458404066408588e-05, + "loss": 1.2386, + "step": 411 + }, + { + "epoch": 0.75, + "grad_norm": 0.24235735604947403, + "learning_rate": 7.454112573833368e-05, + "loss": 1.1423, + "step": 412 + }, + { + "epoch": 0.76, + "grad_norm": 0.2584614748054343, + "learning_rate": 7.449805390602127e-05, + "loss": 1.2669, + "step": 413 + }, + { + "epoch": 0.76, + "grad_norm": 0.23806123085998873, + "learning_rate": 7.445482536280684e-05, + "loss": 1.1763, + "step": 414 + }, + { + "epoch": 0.76, + "grad_norm": 0.24459517607786851, + "learning_rate": 7.441144030506043e-05, + "loss": 1.198, + "step": 415 + }, + { + "epoch": 0.76, + "grad_norm": 0.25801616402700395, + "learning_rate": 7.436789892986304e-05, + "loss": 1.2136, + "step": 416 + }, + { + "epoch": 0.76, + "grad_norm": 0.2814819942392514, + "learning_rate": 7.432420143500578e-05, + "loss": 1.2398, + "step": 417 + }, + { + "epoch": 0.76, + "grad_norm": 0.22134709322606153, + "learning_rate": 7.428034801898893e-05, + "loss": 1.1592, + "step": 418 + }, + { + "epoch": 0.77, + "grad_norm": 0.2899677536995633, + "learning_rate": 7.42363388810211e-05, + "loss": 1.2296, + "step": 419 + }, + { + "epoch": 0.77, + "grad_norm": 0.24005943230262294, + "learning_rate": 7.419217422101822e-05, + "loss": 1.2223, + "step": 420 + }, + { + "epoch": 0.77, + "grad_norm": 0.26417562369496167, + "learning_rate": 7.414785423960275e-05, + "loss": 1.2261, + "step": 421 + }, + { + "epoch": 0.77, + "grad_norm": 0.2580815883535521, + "learning_rate": 7.410337913810271e-05, + "loss": 1.2021, + "step": 422 + }, + { + "epoch": 0.77, + "grad_norm": 0.25242217589496435, + "learning_rate": 7.405874911855071e-05, + "loss": 1.239, + "step": 423 + }, + { + "epoch": 0.78, + "grad_norm": 0.21991733999839932, + "learning_rate": 7.401396438368315e-05, + "loss": 1.1716, + "step": 424 + }, + { + "epoch": 0.78, + "grad_norm": 0.40116538322720213, + "learning_rate": 7.396902513693924e-05, + "loss": 1.2773, + "step": 425 + }, + { + "epoch": 0.78, + "grad_norm": 0.277333939455099, + "learning_rate": 7.392393158246002e-05, + "loss": 1.2574, + "step": 426 + }, + { + "epoch": 0.78, + "grad_norm": 0.27146087746385755, + "learning_rate": 7.387868392508756e-05, + "loss": 1.2243, + "step": 427 + }, + { + "epoch": 0.78, + "grad_norm": 0.255881055620786, + "learning_rate": 7.38332823703639e-05, + "loss": 1.223, + "step": 428 + }, + { + "epoch": 0.78, + "grad_norm": 0.24807364856677255, + "learning_rate": 7.378772712453021e-05, + "loss": 1.1985, + "step": 429 + }, + { + "epoch": 0.79, + "grad_norm": 0.25746257617764423, + "learning_rate": 7.37420183945258e-05, + "loss": 1.2502, + "step": 430 + }, + { + "epoch": 0.79, + "grad_norm": 0.28851991049982234, + "learning_rate": 7.369615638798722e-05, + "loss": 1.2535, + "step": 431 + }, + { + "epoch": 0.79, + "grad_norm": 0.24113389811604363, + "learning_rate": 7.365014131324725e-05, + "loss": 1.2227, + "step": 432 + }, + { + "epoch": 0.79, + "grad_norm": 0.2414465151257969, + "learning_rate": 7.360397337933405e-05, + "loss": 1.1884, + "step": 433 + }, + { + "epoch": 0.79, + "grad_norm": 0.2735463134699831, + "learning_rate": 7.355765279597011e-05, + "loss": 1.2756, + "step": 434 + }, + { + "epoch": 0.8, + "grad_norm": 0.2588437452987293, + "learning_rate": 7.351117977357139e-05, + "loss": 1.2108, + "step": 435 + }, + { + "epoch": 0.8, + "grad_norm": 0.26573294117796553, + "learning_rate": 7.346455452324629e-05, + "loss": 1.1821, + "step": 436 + }, + { + "epoch": 0.8, + "grad_norm": 0.2555476577827304, + "learning_rate": 7.341777725679473e-05, + "loss": 1.1937, + "step": 437 + }, + { + "epoch": 0.8, + "grad_norm": 0.2867704132108098, + "learning_rate": 7.337084818670716e-05, + "loss": 1.2272, + "step": 438 + }, + { + "epoch": 0.8, + "grad_norm": 0.27726678115981157, + "learning_rate": 7.332376752616367e-05, + "loss": 1.2331, + "step": 439 + }, + { + "epoch": 0.8, + "grad_norm": 0.26955338021079955, + "learning_rate": 7.32765354890329e-05, + "loss": 1.1731, + "step": 440 + }, + { + "epoch": 0.81, + "grad_norm": 0.25250321202536524, + "learning_rate": 7.322915228987116e-05, + "loss": 1.2653, + "step": 441 + }, + { + "epoch": 0.81, + "grad_norm": 0.24748844179765395, + "learning_rate": 7.318161814392143e-05, + "loss": 1.24, + "step": 442 + }, + { + "epoch": 0.81, + "grad_norm": 0.28177805247356325, + "learning_rate": 7.313393326711239e-05, + "loss": 1.185, + "step": 443 + }, + { + "epoch": 0.81, + "grad_norm": 0.24093242000396312, + "learning_rate": 7.30860978760574e-05, + "loss": 1.1994, + "step": 444 + }, + { + "epoch": 0.81, + "grad_norm": 0.26277803901457075, + "learning_rate": 7.30381121880536e-05, + "loss": 1.212, + "step": 445 + }, + { + "epoch": 0.82, + "grad_norm": 0.2506524258682433, + "learning_rate": 7.298997642108079e-05, + "loss": 1.2421, + "step": 446 + }, + { + "epoch": 0.82, + "grad_norm": 0.2840599700015824, + "learning_rate": 7.294169079380061e-05, + "loss": 1.1818, + "step": 447 + }, + { + "epoch": 0.82, + "grad_norm": 0.24892184038117549, + "learning_rate": 7.289325552555538e-05, + "loss": 1.1916, + "step": 448 + }, + { + "epoch": 0.82, + "grad_norm": 0.2700898428541357, + "learning_rate": 7.284467083636722e-05, + "loss": 1.2517, + "step": 449 + }, + { + "epoch": 0.82, + "grad_norm": 0.2617848546539419, + "learning_rate": 7.279593694693698e-05, + "loss": 1.2063, + "step": 450 + }, + { + "epoch": 0.82, + "grad_norm": 0.2698278585334131, + "learning_rate": 7.274705407864332e-05, + "loss": 1.194, + "step": 451 + }, + { + "epoch": 0.83, + "grad_norm": 0.23678313024953834, + "learning_rate": 7.26980224535416e-05, + "loss": 1.2349, + "step": 452 + }, + { + "epoch": 0.83, + "grad_norm": 0.24851875792002978, + "learning_rate": 7.264884229436293e-05, + "loss": 1.1758, + "step": 453 + }, + { + "epoch": 0.83, + "grad_norm": 0.24122080121681125, + "learning_rate": 7.259951382451318e-05, + "loss": 1.1962, + "step": 454 + }, + { + "epoch": 0.83, + "grad_norm": 0.22741322959884405, + "learning_rate": 7.25500372680719e-05, + "loss": 1.1702, + "step": 455 + }, + { + "epoch": 0.83, + "grad_norm": 0.2297475610861458, + "learning_rate": 7.250041284979137e-05, + "loss": 1.1466, + "step": 456 + }, + { + "epoch": 0.84, + "grad_norm": 0.3057605989721467, + "learning_rate": 7.245064079509553e-05, + "loss": 1.246, + "step": 457 + }, + { + "epoch": 0.84, + "grad_norm": 0.2719638501597136, + "learning_rate": 7.240072133007899e-05, + "loss": 1.2184, + "step": 458 + }, + { + "epoch": 0.84, + "grad_norm": 0.2436807816414479, + "learning_rate": 7.235065468150593e-05, + "loss": 1.2324, + "step": 459 + }, + { + "epoch": 0.84, + "grad_norm": 0.23436349430255515, + "learning_rate": 7.23004410768092e-05, + "loss": 1.1813, + "step": 460 + }, + { + "epoch": 0.84, + "grad_norm": 0.2398940990211377, + "learning_rate": 7.22500807440892e-05, + "loss": 1.1924, + "step": 461 + }, + { + "epoch": 0.84, + "grad_norm": 0.2605716625062531, + "learning_rate": 7.219957391211281e-05, + "loss": 1.182, + "step": 462 + }, + { + "epoch": 0.85, + "grad_norm": 0.260462524570941, + "learning_rate": 7.214892081031244e-05, + "loss": 1.2136, + "step": 463 + }, + { + "epoch": 0.85, + "grad_norm": 0.21979766512306334, + "learning_rate": 7.209812166878491e-05, + "loss": 1.2066, + "step": 464 + }, + { + "epoch": 0.85, + "grad_norm": 0.23324453647530663, + "learning_rate": 7.204717671829051e-05, + "loss": 1.1657, + "step": 465 + }, + { + "epoch": 0.85, + "grad_norm": 0.2529434935507481, + "learning_rate": 7.199608619025177e-05, + "loss": 1.2093, + "step": 466 + }, + { + "epoch": 0.85, + "grad_norm": 0.25371701891720116, + "learning_rate": 7.194485031675265e-05, + "loss": 1.2225, + "step": 467 + }, + { + "epoch": 0.86, + "grad_norm": 0.23272423066292103, + "learning_rate": 7.189346933053725e-05, + "loss": 1.1721, + "step": 468 + }, + { + "epoch": 0.86, + "grad_norm": 0.25122928735587546, + "learning_rate": 7.184194346500892e-05, + "loss": 1.2537, + "step": 469 + }, + { + "epoch": 0.86, + "grad_norm": 0.2159270875490409, + "learning_rate": 7.179027295422913e-05, + "loss": 1.197, + "step": 470 + }, + { + "epoch": 0.86, + "grad_norm": 0.2633111059076544, + "learning_rate": 7.173845803291636e-05, + "loss": 1.1721, + "step": 471 + }, + { + "epoch": 0.86, + "grad_norm": 0.30555936322098703, + "learning_rate": 7.168649893644517e-05, + "loss": 1.3011, + "step": 472 + }, + { + "epoch": 0.87, + "grad_norm": 0.23492670111453726, + "learning_rate": 7.163439590084502e-05, + "loss": 1.1601, + "step": 473 + }, + { + "epoch": 0.87, + "grad_norm": 0.26602734263721806, + "learning_rate": 7.158214916279923e-05, + "loss": 1.2808, + "step": 474 + }, + { + "epoch": 0.87, + "grad_norm": 0.3182695007856262, + "learning_rate": 7.152975895964386e-05, + "loss": 1.2967, + "step": 475 + }, + { + "epoch": 0.87, + "grad_norm": 0.2785021674736721, + "learning_rate": 7.147722552936673e-05, + "loss": 1.1789, + "step": 476 + }, + { + "epoch": 0.87, + "grad_norm": 0.279474303138652, + "learning_rate": 7.142454911060627e-05, + "loss": 1.2596, + "step": 477 + }, + { + "epoch": 0.87, + "grad_norm": 0.2556980144910755, + "learning_rate": 7.137172994265044e-05, + "loss": 1.2426, + "step": 478 + }, + { + "epoch": 0.88, + "grad_norm": 0.3311256331993533, + "learning_rate": 7.131876826543565e-05, + "loss": 1.2059, + "step": 479 + }, + { + "epoch": 0.88, + "grad_norm": 0.26467296197775253, + "learning_rate": 7.12656643195457e-05, + "loss": 1.2482, + "step": 480 + }, + { + "epoch": 0.88, + "grad_norm": 0.27444885274652553, + "learning_rate": 7.121241834621064e-05, + "loss": 1.2528, + "step": 481 + }, + { + "epoch": 0.88, + "grad_norm": 0.2572283861115396, + "learning_rate": 7.115903058730567e-05, + "loss": 1.1849, + "step": 482 + }, + { + "epoch": 0.88, + "grad_norm": 0.2677065778235683, + "learning_rate": 7.11055012853501e-05, + "loss": 1.2011, + "step": 483 + }, + { + "epoch": 0.89, + "grad_norm": 0.29470622036742816, + "learning_rate": 7.105183068350619e-05, + "loss": 1.2398, + "step": 484 + }, + { + "epoch": 0.89, + "grad_norm": 0.27609230248969197, + "learning_rate": 7.099801902557811e-05, + "loss": 1.2259, + "step": 485 + }, + { + "epoch": 0.89, + "grad_norm": 0.24248634168099284, + "learning_rate": 7.094406655601073e-05, + "loss": 1.2282, + "step": 486 + }, + { + "epoch": 0.89, + "grad_norm": 0.2765941767688746, + "learning_rate": 7.088997351988865e-05, + "loss": 1.2319, + "step": 487 + }, + { + "epoch": 0.89, + "grad_norm": 0.29347776909858947, + "learning_rate": 7.083574016293493e-05, + "loss": 1.1765, + "step": 488 + }, + { + "epoch": 0.89, + "grad_norm": 0.285370295424537, + "learning_rate": 7.078136673151008e-05, + "loss": 1.26, + "step": 489 + }, + { + "epoch": 0.9, + "grad_norm": 0.29408734903836536, + "learning_rate": 7.072685347261093e-05, + "loss": 1.226, + "step": 490 + }, + { + "epoch": 0.9, + "grad_norm": 0.27437470239205813, + "learning_rate": 7.067220063386947e-05, + "loss": 1.1976, + "step": 491 + }, + { + "epoch": 0.9, + "grad_norm": 0.2680770258777871, + "learning_rate": 7.061740846355176e-05, + "loss": 1.1915, + "step": 492 + }, + { + "epoch": 0.9, + "grad_norm": 0.27200362879502954, + "learning_rate": 7.056247721055678e-05, + "loss": 1.2002, + "step": 493 + }, + { + "epoch": 0.9, + "grad_norm": 0.2637811092577037, + "learning_rate": 7.050740712441528e-05, + "loss": 1.287, + "step": 494 + }, + { + "epoch": 0.91, + "grad_norm": 0.24657959209271266, + "learning_rate": 7.045219845528875e-05, + "loss": 1.2284, + "step": 495 + }, + { + "epoch": 0.91, + "grad_norm": 0.25311992110358666, + "learning_rate": 7.039685145396812e-05, + "loss": 1.1616, + "step": 496 + }, + { + "epoch": 0.91, + "grad_norm": 0.2564633694193358, + "learning_rate": 7.034136637187275e-05, + "loss": 1.2067, + "step": 497 + }, + { + "epoch": 0.91, + "grad_norm": 0.2446797651174144, + "learning_rate": 7.028574346104926e-05, + "loss": 1.2284, + "step": 498 + }, + { + "epoch": 0.91, + "grad_norm": 0.2592751463399255, + "learning_rate": 7.022998297417034e-05, + "loss": 1.2371, + "step": 499 + }, + { + "epoch": 0.91, + "grad_norm": 0.2500713943206808, + "learning_rate": 7.017408516453365e-05, + "loss": 1.1061, + "step": 500 + }, + { + "epoch": 0.92, + "grad_norm": 0.2812266276040743, + "learning_rate": 7.011805028606064e-05, + "loss": 1.1949, + "step": 501 + }, + { + "epoch": 0.92, + "grad_norm": 0.298829667668083, + "learning_rate": 7.006187859329544e-05, + "loss": 1.2313, + "step": 502 + }, + { + "epoch": 0.92, + "grad_norm": 0.26518768159745104, + "learning_rate": 7.000557034140361e-05, + "loss": 1.2246, + "step": 503 + }, + { + "epoch": 0.92, + "grad_norm": 0.3037280360760458, + "learning_rate": 6.994912578617113e-05, + "loss": 1.1617, + "step": 504 + }, + { + "epoch": 0.92, + "grad_norm": 0.2726903109255714, + "learning_rate": 6.989254518400309e-05, + "loss": 1.2415, + "step": 505 + }, + { + "epoch": 0.93, + "grad_norm": 0.25568082003046966, + "learning_rate": 6.98358287919226e-05, + "loss": 1.1817, + "step": 506 + }, + { + "epoch": 0.93, + "grad_norm": 0.25633294893705044, + "learning_rate": 6.97789768675696e-05, + "loss": 1.2149, + "step": 507 + }, + { + "epoch": 0.93, + "grad_norm": 0.28291439435087123, + "learning_rate": 6.972198966919972e-05, + "loss": 1.1578, + "step": 508 + }, + { + "epoch": 0.93, + "grad_norm": 0.27195184756655516, + "learning_rate": 6.966486745568308e-05, + "loss": 1.2355, + "step": 509 + }, + { + "epoch": 0.93, + "grad_norm": 0.239159568376005, + "learning_rate": 6.960761048650312e-05, + "loss": 1.1688, + "step": 510 + }, + { + "epoch": 0.93, + "grad_norm": 0.22961475425949177, + "learning_rate": 6.955021902175543e-05, + "loss": 1.2094, + "step": 511 + }, + { + "epoch": 0.94, + "grad_norm": 0.27443773600741117, + "learning_rate": 6.949269332214651e-05, + "loss": 1.2559, + "step": 512 + }, + { + "epoch": 0.94, + "grad_norm": 0.26230551832002097, + "learning_rate": 6.94350336489927e-05, + "loss": 1.2121, + "step": 513 + }, + { + "epoch": 0.94, + "grad_norm": 0.2716742985303849, + "learning_rate": 6.937724026421892e-05, + "loss": 1.2444, + "step": 514 + }, + { + "epoch": 0.94, + "grad_norm": 0.2537850139439542, + "learning_rate": 6.931931343035742e-05, + "loss": 1.1327, + "step": 515 + }, + { + "epoch": 0.94, + "grad_norm": 0.28599587967496826, + "learning_rate": 6.926125341054676e-05, + "loss": 1.2236, + "step": 516 + }, + { + "epoch": 0.95, + "grad_norm": 0.26780654378470103, + "learning_rate": 6.920306046853043e-05, + "loss": 1.2295, + "step": 517 + }, + { + "epoch": 0.95, + "grad_norm": 0.23606296888412015, + "learning_rate": 6.914473486865577e-05, + "loss": 1.1543, + "step": 518 + }, + { + "epoch": 0.95, + "grad_norm": 0.34976881174240837, + "learning_rate": 6.90862768758727e-05, + "loss": 1.2067, + "step": 519 + }, + { + "epoch": 0.95, + "grad_norm": 0.2481257873494882, + "learning_rate": 6.902768675573258e-05, + "loss": 1.2188, + "step": 520 + }, + { + "epoch": 0.95, + "grad_norm": 0.2996395778117021, + "learning_rate": 6.896896477438699e-05, + "loss": 1.2326, + "step": 521 + }, + { + "epoch": 0.95, + "grad_norm": 0.8839768816333193, + "learning_rate": 6.891011119858643e-05, + "loss": 1.2435, + "step": 522 + }, + { + "epoch": 0.96, + "grad_norm": 0.2851882482058998, + "learning_rate": 6.885112629567927e-05, + "loss": 1.2644, + "step": 523 + }, + { + "epoch": 0.96, + "grad_norm": 0.2813663482913699, + "learning_rate": 6.879201033361035e-05, + "loss": 1.2309, + "step": 524 + }, + { + "epoch": 0.96, + "grad_norm": 0.3257551560135454, + "learning_rate": 6.873276358091996e-05, + "loss": 1.2755, + "step": 525 + }, + { + "epoch": 0.96, + "grad_norm": 0.28930479952494365, + "learning_rate": 6.867338630674247e-05, + "loss": 1.1962, + "step": 526 + }, + { + "epoch": 0.96, + "grad_norm": 0.3077462996938649, + "learning_rate": 6.861387878080511e-05, + "loss": 1.2402, + "step": 527 + }, + { + "epoch": 0.97, + "grad_norm": 0.2848900193452761, + "learning_rate": 6.855424127342688e-05, + "loss": 1.2748, + "step": 528 + }, + { + "epoch": 0.97, + "grad_norm": 0.4765938812802202, + "learning_rate": 6.849447405551718e-05, + "loss": 1.2226, + "step": 529 + }, + { + "epoch": 0.97, + "grad_norm": 0.53184473292579, + "learning_rate": 6.843457739857467e-05, + "loss": 1.2347, + "step": 530 + }, + { + "epoch": 0.97, + "grad_norm": 0.6416239346492343, + "learning_rate": 6.837455157468596e-05, + "loss": 1.2429, + "step": 531 + }, + { + "epoch": 0.97, + "grad_norm": 0.3188092712502773, + "learning_rate": 6.831439685652442e-05, + "loss": 1.216, + "step": 532 + }, + { + "epoch": 0.97, + "grad_norm": 0.3527495731006385, + "learning_rate": 6.825411351734895e-05, + "loss": 1.1682, + "step": 533 + }, + { + "epoch": 0.98, + "grad_norm": 0.29603753744741856, + "learning_rate": 6.819370183100274e-05, + "loss": 1.1434, + "step": 534 + }, + { + "epoch": 0.98, + "grad_norm": 0.5252450389976622, + "learning_rate": 6.813316207191198e-05, + "loss": 1.1943, + "step": 535 + }, + { + "epoch": 0.98, + "grad_norm": 0.32999419558659937, + "learning_rate": 6.807249451508466e-05, + "loss": 1.192, + "step": 536 + }, + { + "epoch": 0.98, + "grad_norm": 0.3650175469778724, + "learning_rate": 6.801169943610929e-05, + "loss": 1.2141, + "step": 537 + }, + { + "epoch": 0.98, + "grad_norm": 1.0643532150783557, + "learning_rate": 6.795077711115368e-05, + "loss": 1.2253, + "step": 538 + }, + { + "epoch": 0.99, + "grad_norm": 0.5041310609130145, + "learning_rate": 6.788972781696363e-05, + "loss": 1.278, + "step": 539 + }, + { + "epoch": 0.99, + "grad_norm": 0.5123058164360991, + "learning_rate": 6.782855183086177e-05, + "loss": 1.2231, + "step": 540 + }, + { + "epoch": 0.99, + "grad_norm": 0.3533015702394419, + "learning_rate": 6.776724943074619e-05, + "loss": 1.2072, + "step": 541 + }, + { + "epoch": 0.99, + "grad_norm": 0.30253964625417207, + "learning_rate": 6.770582089508927e-05, + "loss": 1.1382, + "step": 542 + }, + { + "epoch": 0.99, + "grad_norm": 0.348991618828202, + "learning_rate": 6.764426650293633e-05, + "loss": 1.2079, + "step": 543 + }, + { + "epoch": 0.99, + "grad_norm": 0.46017440578788743, + "learning_rate": 6.758258653390444e-05, + "loss": 1.1813, + "step": 544 + }, + { + "epoch": 1.0, + "grad_norm": 0.31962101755594885, + "learning_rate": 6.75207812681811e-05, + "loss": 1.1339, + "step": 545 + }, + { + "epoch": 1.0, + "grad_norm": 0.37092024548285923, + "learning_rate": 6.745885098652298e-05, + "loss": 1.2591, + "step": 546 + }, + { + "epoch": 1.0, + "grad_norm": 0.32347106450715835, + "learning_rate": 6.739679597025466e-05, + "loss": 1.2017, + "step": 547 + }, + { + "epoch": 1.0, + "grad_norm": 0.39250187112342494, + "learning_rate": 6.733461650126733e-05, + "loss": 1.0933, + "step": 548 + }, + { + "epoch": 1.0, + "grad_norm": 0.473522452217324, + "learning_rate": 6.727231286201752e-05, + "loss": 1.1124, + "step": 549 + }, + { + "epoch": 1.01, + "grad_norm": 0.4809062179622052, + "learning_rate": 6.720988533552582e-05, + "loss": 1.1585, + "step": 550 + }, + { + "epoch": 1.01, + "grad_norm": 0.3529662801059162, + "learning_rate": 6.714733420537559e-05, + "loss": 1.0501, + "step": 551 + }, + { + "epoch": 1.01, + "grad_norm": 0.5958247214391118, + "learning_rate": 6.708465975571168e-05, + "loss": 1.1086, + "step": 552 + }, + { + "epoch": 1.01, + "grad_norm": 0.5341364205022454, + "learning_rate": 6.70218622712391e-05, + "loss": 1.0518, + "step": 553 + }, + { + "epoch": 1.01, + "grad_norm": 0.3601805724462006, + "learning_rate": 6.695894203722181e-05, + "loss": 1.1779, + "step": 554 + }, + { + "epoch": 1.02, + "grad_norm": 0.43410190338280613, + "learning_rate": 6.68958993394813e-05, + "loss": 1.093, + "step": 555 + }, + { + "epoch": 1.02, + "grad_norm": 0.46217742572873594, + "learning_rate": 6.683273446439546e-05, + "loss": 1.0117, + "step": 556 + }, + { + "epoch": 1.02, + "grad_norm": 0.8591682373623357, + "learning_rate": 6.676944769889708e-05, + "loss": 1.1002, + "step": 557 + }, + { + "epoch": 1.02, + "grad_norm": 0.7383229487622726, + "learning_rate": 6.670603933047272e-05, + "loss": 1.0779, + "step": 558 + }, + { + "epoch": 1.02, + "grad_norm": 0.5965305891207813, + "learning_rate": 6.664250964716131e-05, + "loss": 1.0889, + "step": 559 + }, + { + "epoch": 1.02, + "grad_norm": 0.6030858606684543, + "learning_rate": 6.657885893755288e-05, + "loss": 1.0982, + "step": 560 + }, + { + "epoch": 1.03, + "grad_norm": 0.4644510682398409, + "learning_rate": 6.65150874907872e-05, + "loss": 1.1004, + "step": 561 + }, + { + "epoch": 1.03, + "grad_norm": 0.43943285132452564, + "learning_rate": 6.645119559655254e-05, + "loss": 1.0536, + "step": 562 + }, + { + "epoch": 1.03, + "grad_norm": 0.4456395978600012, + "learning_rate": 6.638718354508427e-05, + "loss": 1.0733, + "step": 563 + }, + { + "epoch": 1.03, + "grad_norm": 0.3303824433217466, + "learning_rate": 6.632305162716365e-05, + "loss": 1.0552, + "step": 564 + }, + { + "epoch": 1.03, + "grad_norm": 0.3617704823170143, + "learning_rate": 6.62588001341164e-05, + "loss": 1.1092, + "step": 565 + }, + { + "epoch": 1.04, + "grad_norm": 0.4465013349903427, + "learning_rate": 6.619442935781141e-05, + "loss": 1.0781, + "step": 566 + }, + { + "epoch": 1.04, + "grad_norm": 0.48516780613791277, + "learning_rate": 6.612993959065947e-05, + "loss": 1.0686, + "step": 567 + }, + { + "epoch": 1.04, + "grad_norm": 0.38867820318536633, + "learning_rate": 6.606533112561186e-05, + "loss": 1.1215, + "step": 568 + }, + { + "epoch": 1.04, + "grad_norm": 0.38566119820378336, + "learning_rate": 6.600060425615907e-05, + "loss": 1.1213, + "step": 569 + }, + { + "epoch": 1.04, + "grad_norm": 0.35534855445058544, + "learning_rate": 6.593575927632947e-05, + "loss": 1.0955, + "step": 570 + }, + { + "epoch": 1.04, + "grad_norm": 0.38124406233349717, + "learning_rate": 6.587079648068795e-05, + "loss": 1.0659, + "step": 571 + }, + { + "epoch": 1.05, + "grad_norm": 0.454750160923548, + "learning_rate": 6.580571616433457e-05, + "loss": 1.1149, + "step": 572 + }, + { + "epoch": 1.05, + "grad_norm": 0.35353190088025255, + "learning_rate": 6.574051862290325e-05, + "loss": 1.0388, + "step": 573 + }, + { + "epoch": 1.05, + "grad_norm": 0.3249395594793626, + "learning_rate": 6.567520415256045e-05, + "loss": 1.0784, + "step": 574 + }, + { + "epoch": 1.05, + "grad_norm": 0.40078898818247227, + "learning_rate": 6.560977305000375e-05, + "loss": 1.0859, + "step": 575 + }, + { + "epoch": 1.05, + "grad_norm": 0.4115264795060035, + "learning_rate": 6.554422561246054e-05, + "loss": 1.1828, + "step": 576 + }, + { + "epoch": 1.06, + "grad_norm": 0.30090229228069215, + "learning_rate": 6.54785621376867e-05, + "loss": 1.0901, + "step": 577 + }, + { + "epoch": 1.06, + "grad_norm": 0.28827860350299206, + "learning_rate": 6.541278292396523e-05, + "loss": 1.0277, + "step": 578 + }, + { + "epoch": 1.06, + "grad_norm": 0.34690404488996757, + "learning_rate": 6.534688827010484e-05, + "loss": 1.048, + "step": 579 + }, + { + "epoch": 1.06, + "grad_norm": 0.29943113556644785, + "learning_rate": 6.528087847543867e-05, + "loss": 1.0646, + "step": 580 + }, + { + "epoch": 1.06, + "grad_norm": 0.37318202575874415, + "learning_rate": 6.521475383982291e-05, + "loss": 1.1091, + "step": 581 + }, + { + "epoch": 1.06, + "grad_norm": 0.3049663659203959, + "learning_rate": 6.51485146636354e-05, + "loss": 1.0552, + "step": 582 + }, + { + "epoch": 1.07, + "grad_norm": 0.3342407867509692, + "learning_rate": 6.508216124777431e-05, + "loss": 1.2227, + "step": 583 + }, + { + "epoch": 1.07, + "grad_norm": 0.3348396047855952, + "learning_rate": 6.501569389365674e-05, + "loss": 1.0861, + "step": 584 + }, + { + "epoch": 1.07, + "grad_norm": 0.30951429367513383, + "learning_rate": 6.494911290321737e-05, + "loss": 1.0461, + "step": 585 + }, + { + "epoch": 1.07, + "grad_norm": 0.33898401361064606, + "learning_rate": 6.488241857890711e-05, + "loss": 1.0854, + "step": 586 + }, + { + "epoch": 1.07, + "grad_norm": 0.4901462068263497, + "learning_rate": 6.481561122369164e-05, + "loss": 1.1012, + "step": 587 + }, + { + "epoch": 1.08, + "grad_norm": 0.3179574879809652, + "learning_rate": 6.474869114105018e-05, + "loss": 1.0451, + "step": 588 + }, + { + "epoch": 1.08, + "grad_norm": 0.32159328915060714, + "learning_rate": 6.468165863497395e-05, + "loss": 1.0458, + "step": 589 + }, + { + "epoch": 1.08, + "grad_norm": 0.36462235008537297, + "learning_rate": 6.461451400996491e-05, + "loss": 1.1247, + "step": 590 + }, + { + "epoch": 1.08, + "grad_norm": 0.5373862753611778, + "learning_rate": 6.454725757103432e-05, + "loss": 1.0542, + "step": 591 + }, + { + "epoch": 1.08, + "grad_norm": 0.3160409270291303, + "learning_rate": 6.447988962370133e-05, + "loss": 1.0829, + "step": 592 + }, + { + "epoch": 1.08, + "grad_norm": 0.390452102978435, + "learning_rate": 6.441241047399169e-05, + "loss": 1.192, + "step": 593 + }, + { + "epoch": 1.09, + "grad_norm": 0.3802122712014928, + "learning_rate": 6.434482042843627e-05, + "loss": 1.1153, + "step": 594 + }, + { + "epoch": 1.09, + "grad_norm": 0.4081584328242501, + "learning_rate": 6.427711979406966e-05, + "loss": 1.1635, + "step": 595 + }, + { + "epoch": 1.09, + "grad_norm": 0.3791962989638633, + "learning_rate": 6.420930887842889e-05, + "loss": 1.1581, + "step": 596 + }, + { + "epoch": 1.09, + "grad_norm": 0.33239440056484193, + "learning_rate": 6.414138798955189e-05, + "loss": 1.0926, + "step": 597 + }, + { + "epoch": 1.09, + "grad_norm": 0.3279881540815014, + "learning_rate": 6.407335743597616e-05, + "loss": 1.1386, + "step": 598 + }, + { + "epoch": 1.1, + "grad_norm": 0.30309644763750837, + "learning_rate": 6.40052175267374e-05, + "loss": 1.0523, + "step": 599 + }, + { + "epoch": 1.1, + "grad_norm": 0.3349097308403333, + "learning_rate": 6.393696857136801e-05, + "loss": 1.0815, + "step": 600 + }, + { + "epoch": 1.1, + "grad_norm": 0.3288227593556618, + "learning_rate": 6.386861087989581e-05, + "loss": 1.015, + "step": 601 + }, + { + "epoch": 1.1, + "grad_norm": 0.36685586740843157, + "learning_rate": 6.380014476284255e-05, + "loss": 1.1232, + "step": 602 + }, + { + "epoch": 1.1, + "grad_norm": 0.3620977714204643, + "learning_rate": 6.373157053122243e-05, + "loss": 1.1138, + "step": 603 + }, + { + "epoch": 1.1, + "grad_norm": 0.3130587018197183, + "learning_rate": 6.366288849654091e-05, + "loss": 1.1255, + "step": 604 + }, + { + "epoch": 1.11, + "grad_norm": 0.3602737087072766, + "learning_rate": 6.359409897079303e-05, + "loss": 1.0282, + "step": 605 + }, + { + "epoch": 1.11, + "grad_norm": 0.31168852571991945, + "learning_rate": 6.352520226646222e-05, + "loss": 1.0779, + "step": 606 + }, + { + "epoch": 1.11, + "grad_norm": 0.3516045580189353, + "learning_rate": 6.345619869651871e-05, + "loss": 1.1028, + "step": 607 + }, + { + "epoch": 1.11, + "grad_norm": 0.3231857927563657, + "learning_rate": 6.33870885744182e-05, + "loss": 1.1202, + "step": 608 + }, + { + "epoch": 1.11, + "grad_norm": 0.30205205129701157, + "learning_rate": 6.331787221410041e-05, + "loss": 1.1369, + "step": 609 + }, + { + "epoch": 1.12, + "grad_norm": 0.3198359813888166, + "learning_rate": 6.32485499299877e-05, + "loss": 1.1763, + "step": 610 + }, + { + "epoch": 1.12, + "grad_norm": 0.3128641370321787, + "learning_rate": 6.31791220369835e-05, + "loss": 1.0223, + "step": 611 + }, + { + "epoch": 1.12, + "grad_norm": 0.2989105616213649, + "learning_rate": 6.31095888504711e-05, + "loss": 1.0358, + "step": 612 + }, + { + "epoch": 1.12, + "grad_norm": 0.3103537906853337, + "learning_rate": 6.303995068631203e-05, + "loss": 1.1261, + "step": 613 + }, + { + "epoch": 1.12, + "grad_norm": 0.28598715532508207, + "learning_rate": 6.297020786084467e-05, + "loss": 1.0629, + "step": 614 + }, + { + "epoch": 1.12, + "grad_norm": 0.29809789918093255, + "learning_rate": 6.290036069088288e-05, + "loss": 1.035, + "step": 615 + }, + { + "epoch": 1.13, + "grad_norm": 0.33765270252261453, + "learning_rate": 6.283040949371451e-05, + "loss": 1.1221, + "step": 616 + }, + { + "epoch": 1.13, + "grad_norm": 0.3424617501293415, + "learning_rate": 6.276035458709993e-05, + "loss": 1.155, + "step": 617 + }, + { + "epoch": 1.13, + "grad_norm": 0.3799189737987811, + "learning_rate": 6.269019628927067e-05, + "loss": 1.0701, + "step": 618 + }, + { + "epoch": 1.13, + "grad_norm": 0.3358898935253196, + "learning_rate": 6.261993491892791e-05, + "loss": 1.1649, + "step": 619 + }, + { + "epoch": 1.13, + "grad_norm": 0.31569979424117356, + "learning_rate": 6.254957079524099e-05, + "loss": 1.0633, + "step": 620 + }, + { + "epoch": 1.14, + "grad_norm": 0.3002168156888237, + "learning_rate": 6.247910423784609e-05, + "loss": 1.0846, + "step": 621 + }, + { + "epoch": 1.14, + "grad_norm": 0.3097238823450595, + "learning_rate": 6.24085355668447e-05, + "loss": 1.0808, + "step": 622 + }, + { + "epoch": 1.14, + "grad_norm": 0.3120312761417578, + "learning_rate": 6.233786510280212e-05, + "loss": 1.0142, + "step": 623 + }, + { + "epoch": 1.14, + "grad_norm": 0.3335343015064923, + "learning_rate": 6.22670931667461e-05, + "loss": 1.0674, + "step": 624 + }, + { + "epoch": 1.14, + "grad_norm": 0.3234062304634526, + "learning_rate": 6.219622008016533e-05, + "loss": 1.0981, + "step": 625 + }, + { + "epoch": 1.14, + "grad_norm": 0.32152678786547273, + "learning_rate": 6.212524616500798e-05, + "loss": 1.0244, + "step": 626 + }, + { + "epoch": 1.15, + "grad_norm": 0.39031977608147594, + "learning_rate": 6.205417174368023e-05, + "loss": 1.1205, + "step": 627 + }, + { + "epoch": 1.15, + "grad_norm": 0.3806189090017157, + "learning_rate": 6.198299713904485e-05, + "loss": 1.1134, + "step": 628 + }, + { + "epoch": 1.15, + "grad_norm": 0.2978349276971668, + "learning_rate": 6.191172267441967e-05, + "loss": 1.0088, + "step": 629 + }, + { + "epoch": 1.15, + "grad_norm": 0.3190354077382501, + "learning_rate": 6.184034867357617e-05, + "loss": 1.108, + "step": 630 + }, + { + "epoch": 1.15, + "grad_norm": 0.32633048665038994, + "learning_rate": 6.176887546073797e-05, + "loss": 1.0825, + "step": 631 + }, + { + "epoch": 1.16, + "grad_norm": 0.3428026413020903, + "learning_rate": 6.169730336057939e-05, + "loss": 1.0765, + "step": 632 + }, + { + "epoch": 1.16, + "grad_norm": 0.3475737151929015, + "learning_rate": 6.162563269822391e-05, + "loss": 1.0693, + "step": 633 + }, + { + "epoch": 1.16, + "grad_norm": 0.3870252154591392, + "learning_rate": 6.15538637992428e-05, + "loss": 1.1081, + "step": 634 + }, + { + "epoch": 1.16, + "grad_norm": 0.33597355193652834, + "learning_rate": 6.148199698965352e-05, + "loss": 1.0893, + "step": 635 + }, + { + "epoch": 1.16, + "grad_norm": 0.30805894179787247, + "learning_rate": 6.141003259591834e-05, + "loss": 1.0995, + "step": 636 + }, + { + "epoch": 1.17, + "grad_norm": 0.3025073882734066, + "learning_rate": 6.133797094494281e-05, + "loss": 1.0388, + "step": 637 + }, + { + "epoch": 1.17, + "grad_norm": 0.3524395196391662, + "learning_rate": 6.126581236407429e-05, + "loss": 1.1196, + "step": 638 + }, + { + "epoch": 1.17, + "grad_norm": 0.3377646188130345, + "learning_rate": 6.119355718110039e-05, + "loss": 1.0382, + "step": 639 + }, + { + "epoch": 1.17, + "grad_norm": 0.35508400659785483, + "learning_rate": 6.112120572424763e-05, + "loss": 1.1402, + "step": 640 + }, + { + "epoch": 1.17, + "grad_norm": 0.3454418793700457, + "learning_rate": 6.104875832217982e-05, + "loss": 1.1032, + "step": 641 + }, + { + "epoch": 1.17, + "grad_norm": 0.32629806837059866, + "learning_rate": 6.097621530399661e-05, + "loss": 1.0959, + "step": 642 + }, + { + "epoch": 1.18, + "grad_norm": 0.3329536837751315, + "learning_rate": 6.090357699923202e-05, + "loss": 1.0467, + "step": 643 + }, + { + "epoch": 1.18, + "grad_norm": 0.32302233828349475, + "learning_rate": 6.083084373785287e-05, + "loss": 1.0858, + "step": 644 + }, + { + "epoch": 1.18, + "grad_norm": 0.3310358826507611, + "learning_rate": 6.075801585025739e-05, + "loss": 1.0715, + "step": 645 + }, + { + "epoch": 1.18, + "grad_norm": 0.319322035854079, + "learning_rate": 6.068509366727362e-05, + "loss": 1.177, + "step": 646 + }, + { + "epoch": 1.18, + "grad_norm": 0.3065230667302707, + "learning_rate": 6.061207752015797e-05, + "loss": 1.0649, + "step": 647 + }, + { + "epoch": 1.19, + "grad_norm": 0.29926795565748227, + "learning_rate": 6.053896774059368e-05, + "loss": 1.1325, + "step": 648 + }, + { + "epoch": 1.19, + "grad_norm": 0.3556069634279046, + "learning_rate": 6.046576466068931e-05, + "loss": 1.1366, + "step": 649 + }, + { + "epoch": 1.19, + "grad_norm": 0.3189191131461966, + "learning_rate": 6.039246861297727e-05, + "loss": 1.0693, + "step": 650 + }, + { + "epoch": 1.19, + "grad_norm": 0.3347197156648834, + "learning_rate": 6.031907993041227e-05, + "loss": 1.1009, + "step": 651 + }, + { + "epoch": 1.19, + "grad_norm": 0.32274156348185445, + "learning_rate": 6.0245598946369826e-05, + "loss": 1.1675, + "step": 652 + }, + { + "epoch": 1.19, + "grad_norm": 0.35534089035455224, + "learning_rate": 6.017202599464476e-05, + "loss": 1.1723, + "step": 653 + }, + { + "epoch": 1.2, + "grad_norm": 0.3106026578570133, + "learning_rate": 6.009836140944965e-05, + "loss": 1.0954, + "step": 654 + }, + { + "epoch": 1.2, + "grad_norm": 0.3309144454564729, + "learning_rate": 6.002460552541331e-05, + "loss": 1.0209, + "step": 655 + }, + { + "epoch": 1.2, + "grad_norm": 0.3023619281400003, + "learning_rate": 5.9950758677579345e-05, + "loss": 1.0363, + "step": 656 + }, + { + "epoch": 1.2, + "grad_norm": 0.3311182880219704, + "learning_rate": 5.987682120140451e-05, + "loss": 1.0515, + "step": 657 + }, + { + "epoch": 1.2, + "grad_norm": 0.33396486010030413, + "learning_rate": 5.980279343275729e-05, + "loss": 1.1251, + "step": 658 + }, + { + "epoch": 1.21, + "grad_norm": 0.3465764556678002, + "learning_rate": 5.97286757079163e-05, + "loss": 1.165, + "step": 659 + }, + { + "epoch": 1.21, + "grad_norm": 0.304193441363374, + "learning_rate": 5.965446836356882e-05, + "loss": 1.0228, + "step": 660 + }, + { + "epoch": 1.21, + "grad_norm": 0.3415149030413082, + "learning_rate": 5.9580171736809224e-05, + "loss": 1.0742, + "step": 661 + }, + { + "epoch": 1.21, + "grad_norm": 0.33138658321132064, + "learning_rate": 5.950578616513746e-05, + "loss": 1.0843, + "step": 662 + }, + { + "epoch": 1.21, + "grad_norm": 0.30774403421162994, + "learning_rate": 5.943131198645752e-05, + "loss": 1.065, + "step": 663 + }, + { + "epoch": 1.21, + "grad_norm": 0.3428877492183819, + "learning_rate": 5.9356749539075885e-05, + "loss": 1.1101, + "step": 664 + }, + { + "epoch": 1.22, + "grad_norm": 0.3621290546130101, + "learning_rate": 5.928209916170003e-05, + "loss": 1.1372, + "step": 665 + }, + { + "epoch": 1.22, + "grad_norm": 0.3482375945469884, + "learning_rate": 5.9207361193436865e-05, + "loss": 1.132, + "step": 666 + }, + { + "epoch": 1.22, + "grad_norm": 0.31754384974068384, + "learning_rate": 5.9132535973791156e-05, + "loss": 1.148, + "step": 667 + }, + { + "epoch": 1.22, + "grad_norm": 0.36003834782050365, + "learning_rate": 5.9057623842664044e-05, + "loss": 1.1099, + "step": 668 + }, + { + "epoch": 1.22, + "grad_norm": 0.2963701622969662, + "learning_rate": 5.8982625140351464e-05, + "loss": 1.0755, + "step": 669 + }, + { + "epoch": 1.23, + "grad_norm": 0.32579569606066516, + "learning_rate": 5.8907540207542616e-05, + "loss": 1.0809, + "step": 670 + }, + { + "epoch": 1.23, + "grad_norm": 0.4247563451753457, + "learning_rate": 5.8832369385318416e-05, + "loss": 1.097, + "step": 671 + }, + { + "epoch": 1.23, + "grad_norm": 0.33076932102169776, + "learning_rate": 5.875711301514992e-05, + "loss": 1.1078, + "step": 672 + }, + { + "epoch": 1.23, + "grad_norm": 0.3609238032332309, + "learning_rate": 5.8681771438896815e-05, + "loss": 1.1031, + "step": 673 + }, + { + "epoch": 1.23, + "grad_norm": 0.325159585649425, + "learning_rate": 5.860634499880583e-05, + "loss": 1.0707, + "step": 674 + }, + { + "epoch": 1.23, + "grad_norm": 0.4620687271068983, + "learning_rate": 5.853083403750922e-05, + "loss": 1.1017, + "step": 675 + }, + { + "epoch": 1.24, + "grad_norm": 0.33485279064365936, + "learning_rate": 5.845523889802316e-05, + "loss": 1.0989, + "step": 676 + }, + { + "epoch": 1.24, + "grad_norm": 0.30952573170841513, + "learning_rate": 5.8379559923746214e-05, + "loss": 1.0393, + "step": 677 + }, + { + "epoch": 1.24, + "grad_norm": 0.33498605810588283, + "learning_rate": 5.830379745845781e-05, + "loss": 1.1259, + "step": 678 + }, + { + "epoch": 1.24, + "grad_norm": 0.35771921163037307, + "learning_rate": 5.822795184631659e-05, + "loss": 1.0815, + "step": 679 + }, + { + "epoch": 1.24, + "grad_norm": 0.3329650192347647, + "learning_rate": 5.815202343185894e-05, + "loss": 1.1344, + "step": 680 + }, + { + "epoch": 1.25, + "grad_norm": 0.3356634465845771, + "learning_rate": 5.807601255999736e-05, + "loss": 1.1297, + "step": 681 + }, + { + "epoch": 1.25, + "grad_norm": 0.3289442034151235, + "learning_rate": 5.7999919576018934e-05, + "loss": 1.022, + "step": 682 + }, + { + "epoch": 1.25, + "grad_norm": 0.3207007334784113, + "learning_rate": 5.7923744825583745e-05, + "loss": 1.0571, + "step": 683 + }, + { + "epoch": 1.25, + "grad_norm": 0.3582460325329284, + "learning_rate": 5.7847488654723304e-05, + "loss": 1.0778, + "step": 684 + }, + { + "epoch": 1.25, + "grad_norm": 0.3563317666176927, + "learning_rate": 5.777115140983899e-05, + "loss": 1.1003, + "step": 685 + }, + { + "epoch": 1.25, + "grad_norm": 3.4694912945702105, + "learning_rate": 5.769473343770047e-05, + "loss": 1.121, + "step": 686 + }, + { + "epoch": 1.26, + "grad_norm": 0.43002349520483113, + "learning_rate": 5.761823508544411e-05, + "loss": 1.0765, + "step": 687 + }, + { + "epoch": 1.26, + "grad_norm": 0.39467783104839754, + "learning_rate": 5.754165670057142e-05, + "loss": 1.0788, + "step": 688 + }, + { + "epoch": 1.26, + "grad_norm": 0.39629029674867916, + "learning_rate": 5.7464998630947464e-05, + "loss": 1.0812, + "step": 689 + }, + { + "epoch": 1.26, + "grad_norm": 0.3880152093965208, + "learning_rate": 5.738826122479929e-05, + "loss": 1.1228, + "step": 690 + }, + { + "epoch": 1.26, + "grad_norm": 0.3777874121959188, + "learning_rate": 5.7311444830714324e-05, + "loss": 1.0907, + "step": 691 + }, + { + "epoch": 1.27, + "grad_norm": 0.38004041653523696, + "learning_rate": 5.723454979763882e-05, + "loss": 1.1263, + "step": 692 + }, + { + "epoch": 1.27, + "grad_norm": 0.37049672627797636, + "learning_rate": 5.7157576474876246e-05, + "loss": 1.1438, + "step": 693 + }, + { + "epoch": 1.27, + "grad_norm": 0.32973606103437614, + "learning_rate": 5.7080525212085725e-05, + "loss": 1.0553, + "step": 694 + }, + { + "epoch": 1.27, + "grad_norm": 0.31674639252070325, + "learning_rate": 5.700339635928038e-05, + "loss": 1.06, + "step": 695 + }, + { + "epoch": 1.27, + "grad_norm": 0.32282199426553837, + "learning_rate": 5.692619026682588e-05, + "loss": 1.0841, + "step": 696 + }, + { + "epoch": 1.27, + "grad_norm": 0.4810882958061859, + "learning_rate": 5.684890728543869e-05, + "loss": 1.0803, + "step": 697 + }, + { + "epoch": 1.28, + "grad_norm": 0.3995638550178378, + "learning_rate": 5.6771547766184566e-05, + "loss": 1.1187, + "step": 698 + }, + { + "epoch": 1.28, + "grad_norm": 0.35264932960583484, + "learning_rate": 5.669411206047699e-05, + "loss": 1.0641, + "step": 699 + }, + { + "epoch": 1.28, + "grad_norm": 0.35240640524733, + "learning_rate": 5.661660052007547e-05, + "loss": 1.076, + "step": 700 + }, + { + "epoch": 1.28, + "grad_norm": 0.3540694609860389, + "learning_rate": 5.653901349708401e-05, + "loss": 1.1369, + "step": 701 + }, + { + "epoch": 1.28, + "grad_norm": 0.3196055112925304, + "learning_rate": 5.646135134394955e-05, + "loss": 1.0677, + "step": 702 + }, + { + "epoch": 1.29, + "grad_norm": 0.4214141007955914, + "learning_rate": 5.6383614413460266e-05, + "loss": 1.1139, + "step": 703 + }, + { + "epoch": 1.29, + "grad_norm": 0.3625611311798579, + "learning_rate": 5.630580305874402e-05, + "loss": 1.1845, + "step": 704 + }, + { + "epoch": 1.29, + "grad_norm": 0.3425208672181188, + "learning_rate": 5.62279176332668e-05, + "loss": 1.174, + "step": 705 + }, + { + "epoch": 1.29, + "grad_norm": 0.3108419862818321, + "learning_rate": 5.6149958490830996e-05, + "loss": 1.0331, + "step": 706 + }, + { + "epoch": 1.29, + "grad_norm": 0.3274644181571904, + "learning_rate": 5.607192598557394e-05, + "loss": 1.0664, + "step": 707 + }, + { + "epoch": 1.29, + "grad_norm": 0.346218197215145, + "learning_rate": 5.599382047196617e-05, + "loss": 1.2088, + "step": 708 + }, + { + "epoch": 1.3, + "grad_norm": 0.328497632267458, + "learning_rate": 5.591564230480989e-05, + "loss": 1.0287, + "step": 709 + }, + { + "epoch": 1.3, + "grad_norm": 0.3708173720611468, + "learning_rate": 5.583739183923732e-05, + "loss": 1.0883, + "step": 710 + }, + { + "epoch": 1.3, + "grad_norm": 0.3631427403535479, + "learning_rate": 5.575906943070915e-05, + "loss": 1.1155, + "step": 711 + }, + { + "epoch": 1.3, + "grad_norm": 0.3305201458598695, + "learning_rate": 5.5680675435012834e-05, + "loss": 1.0958, + "step": 712 + }, + { + "epoch": 1.3, + "grad_norm": 0.34978833532083714, + "learning_rate": 5.5602210208261036e-05, + "loss": 1.1437, + "step": 713 + }, + { + "epoch": 1.31, + "grad_norm": 0.3510553882510229, + "learning_rate": 5.552367410688999e-05, + "loss": 1.0941, + "step": 714 + }, + { + "epoch": 1.31, + "grad_norm": 0.3523747462465078, + "learning_rate": 5.544506748765789e-05, + "loss": 1.1289, + "step": 715 + }, + { + "epoch": 1.31, + "grad_norm": 0.38262637783927445, + "learning_rate": 5.5366390707643266e-05, + "loss": 1.099, + "step": 716 + }, + { + "epoch": 1.31, + "grad_norm": 0.38620065989073454, + "learning_rate": 5.528764412424334e-05, + "loss": 1.083, + "step": 717 + }, + { + "epoch": 1.31, + "grad_norm": 0.3401355276121096, + "learning_rate": 5.520882809517245e-05, + "loss": 1.028, + "step": 718 + }, + { + "epoch": 1.32, + "grad_norm": 0.3392061008943934, + "learning_rate": 5.512994297846039e-05, + "loss": 1.1083, + "step": 719 + }, + { + "epoch": 1.32, + "grad_norm": 0.34219480421015414, + "learning_rate": 5.505098913245077e-05, + "loss": 1.1108, + "step": 720 + }, + { + "epoch": 1.32, + "grad_norm": 0.3275058061553761, + "learning_rate": 5.497196691579945e-05, + "loss": 1.111, + "step": 721 + }, + { + "epoch": 1.32, + "grad_norm": 0.36800249746509384, + "learning_rate": 5.489287668747283e-05, + "loss": 1.1221, + "step": 722 + }, + { + "epoch": 1.32, + "grad_norm": 0.4129005533101575, + "learning_rate": 5.481371880674628e-05, + "loss": 1.0966, + "step": 723 + }, + { + "epoch": 1.32, + "grad_norm": 0.36563906596251655, + "learning_rate": 5.4734493633202505e-05, + "loss": 1.0927, + "step": 724 + }, + { + "epoch": 1.33, + "grad_norm": 0.3614650536839971, + "learning_rate": 5.465520152672986e-05, + "loss": 1.13, + "step": 725 + }, + { + "epoch": 1.33, + "grad_norm": 0.36419665098633497, + "learning_rate": 5.4575842847520765e-05, + "loss": 1.1183, + "step": 726 + }, + { + "epoch": 1.33, + "grad_norm": 0.34490689807258995, + "learning_rate": 5.449641795607005e-05, + "loss": 1.0919, + "step": 727 + }, + { + "epoch": 1.33, + "grad_norm": 0.3627643746876298, + "learning_rate": 5.441692721317334e-05, + "loss": 1.0411, + "step": 728 + }, + { + "epoch": 1.33, + "grad_norm": 0.323620411949565, + "learning_rate": 5.433737097992537e-05, + "loss": 1.0725, + "step": 729 + }, + { + "epoch": 1.34, + "grad_norm": 0.3521599501824965, + "learning_rate": 5.425774961771838e-05, + "loss": 1.0926, + "step": 730 + }, + { + "epoch": 1.34, + "grad_norm": 0.3302390546764222, + "learning_rate": 5.417806348824047e-05, + "loss": 1.0468, + "step": 731 + }, + { + "epoch": 1.34, + "grad_norm": 0.3833325802616019, + "learning_rate": 5.4098312953473956e-05, + "loss": 1.1291, + "step": 732 + }, + { + "epoch": 1.34, + "grad_norm": 0.3708621126835512, + "learning_rate": 5.401849837569372e-05, + "loss": 1.0887, + "step": 733 + }, + { + "epoch": 1.34, + "grad_norm": 0.3625834373416278, + "learning_rate": 5.393862011746555e-05, + "loss": 1.0981, + "step": 734 + }, + { + "epoch": 1.34, + "grad_norm": 0.3583343965080617, + "learning_rate": 5.385867854164451e-05, + "loss": 1.1021, + "step": 735 + }, + { + "epoch": 1.35, + "grad_norm": 0.34598320594096066, + "learning_rate": 5.377867401137332e-05, + "loss": 1.1376, + "step": 736 + }, + { + "epoch": 1.35, + "grad_norm": 0.3046382791315433, + "learning_rate": 5.369860689008066e-05, + "loss": 1.0206, + "step": 737 + }, + { + "epoch": 1.35, + "grad_norm": 0.34464948380043725, + "learning_rate": 5.3618477541479505e-05, + "loss": 1.1084, + "step": 738 + }, + { + "epoch": 1.35, + "grad_norm": 0.3203242519627101, + "learning_rate": 5.353828632956557e-05, + "loss": 1.0731, + "step": 739 + }, + { + "epoch": 1.35, + "grad_norm": 0.3431169960355163, + "learning_rate": 5.3458033618615516e-05, + "loss": 1.091, + "step": 740 + }, + { + "epoch": 1.36, + "grad_norm": 0.33492074521678705, + "learning_rate": 5.337771977318543e-05, + "loss": 1.1112, + "step": 741 + }, + { + "epoch": 1.36, + "grad_norm": 0.32576546585541344, + "learning_rate": 5.3297345158109086e-05, + "loss": 1.0993, + "step": 742 + }, + { + "epoch": 1.36, + "grad_norm": 0.3410007245037574, + "learning_rate": 5.3216910138496286e-05, + "loss": 1.094, + "step": 743 + }, + { + "epoch": 1.36, + "grad_norm": 0.34891180680896833, + "learning_rate": 5.313641507973128e-05, + "loss": 1.1331, + "step": 744 + }, + { + "epoch": 1.36, + "grad_norm": 0.37135766946717214, + "learning_rate": 5.3055860347471006e-05, + "loss": 1.1, + "step": 745 + }, + { + "epoch": 1.36, + "grad_norm": 0.3465019415478411, + "learning_rate": 5.297524630764349e-05, + "loss": 1.1256, + "step": 746 + }, + { + "epoch": 1.37, + "grad_norm": 0.37035388481626563, + "learning_rate": 5.289457332644615e-05, + "loss": 1.0366, + "step": 747 + }, + { + "epoch": 1.37, + "grad_norm": 0.33853883270759155, + "learning_rate": 5.281384177034421e-05, + "loss": 1.0547, + "step": 748 + }, + { + "epoch": 1.37, + "grad_norm": 0.364306618627317, + "learning_rate": 5.2733052006068897e-05, + "loss": 1.0768, + "step": 749 + }, + { + "epoch": 1.37, + "grad_norm": 0.4021754315731627, + "learning_rate": 5.2652204400615916e-05, + "loss": 1.1382, + "step": 750 + } + ], + "logging_steps": 1.0, + "max_steps": 1638, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 50, + "total_flos": 777666642837504.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-750/training_args.bin b/checkpoint-750/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..c5d2416a3b70bb5260978ec9996f00154a724ba7 --- /dev/null +++ b/checkpoint-750/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b22e8f9d51a16d03a2c506fa3d1eafa8f4b1ae992992c2086a4d435ffd97387e +size 6712 diff --git a/checkpoint-750/zero_to_fp32.py b/checkpoint-750/zero_to_fp32.py new file mode 100755 index 0000000000000000000000000000000000000000..24cc342e78d1a006c782b3a4cd68d9ce786d8fd8 --- /dev/null +++ b/checkpoint-750/zero_to_fp32.py @@ -0,0 +1,604 @@ +#!/usr/bin/env python + +# Copyright (c) Microsoft Corporation. +# SPDX-License-Identifier: Apache-2.0 + +# DeepSpeed Team + +# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets +# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in +# the future. Once extracted, the weights don't require DeepSpeed and can be used in any +# application. +# +# example: python zero_to_fp32.py . pytorch_model.bin + +import argparse +import torch +import glob +import math +import os +import re +from collections import OrderedDict +from dataclasses import dataclass + +# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with +# DeepSpeed data structures it has to be available in the current python environment. +from deepspeed.utils import logger +from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS, + FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES, + FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS) + + +@dataclass +class zero_model_state: + buffers: dict() + param_shapes: dict() + shared_params: list + ds_version: int + frozen_param_shapes: dict() + frozen_param_fragments: dict() + + +debug = 0 + +# load to cpu +device = torch.device('cpu') + + +def atoi(text): + return int(text) if text.isdigit() else text + + +def natural_keys(text): + ''' + alist.sort(key=natural_keys) sorts in human order + http://nedbatchelder.com/blog/200712/human_sorting.html + (See Toothy's implementation in the comments) + ''' + return [atoi(c) for c in re.split(r'(\d+)', text)] + + +def get_model_state_file(checkpoint_dir, zero_stage): + if not os.path.isdir(checkpoint_dir): + raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist") + + # there should be only one file + if zero_stage <= 2: + file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt") + elif zero_stage == 3: + file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt") + + if not os.path.exists(file): + raise FileNotFoundError(f"can't find model states file at '{file}'") + + return file + + +def get_checkpoint_files(checkpoint_dir, glob_pattern): + # XXX: need to test that this simple glob rule works for multi-node setup too + ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys) + + if len(ckpt_files) == 0: + raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'") + + return ckpt_files + + +def get_optim_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt") + + +def get_model_state_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_model_states.pt") + + +def parse_model_states(files): + zero_model_states = [] + for file in files: + state_dict = torch.load(file, map_location=device) + + if BUFFER_NAMES not in state_dict: + raise ValueError(f"{file} is not a model state checkpoint") + buffer_names = state_dict[BUFFER_NAMES] + if debug: + print("Found buffers:", buffer_names) + + # recover just the buffers while restoring them to fp32 if they were saved in fp16 + buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names} + param_shapes = state_dict[PARAM_SHAPES] + + # collect parameters that are included in param_shapes + param_names = [] + for s in param_shapes: + for name in s.keys(): + param_names.append(name) + + # update with frozen parameters + frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None) + if frozen_param_shapes is not None: + if debug: + print(f"Found frozen_param_shapes: {frozen_param_shapes}") + param_names += list(frozen_param_shapes.keys()) + + # handle shared params + shared_params = [[k, v] for k, v in state_dict["shared_params"].items()] + + ds_version = state_dict.get(DS_VERSION, None) + + frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None) + + z_model_state = zero_model_state(buffers=buffers, + param_shapes=param_shapes, + shared_params=shared_params, + ds_version=ds_version, + frozen_param_shapes=frozen_param_shapes, + frozen_param_fragments=frozen_param_fragments) + zero_model_states.append(z_model_state) + + return zero_model_states + + +def parse_optim_states(files, ds_checkpoint_dir): + + total_files = len(files) + state_dicts = [] + for f in files: + state_dict = torch.load(f, map_location=device) + # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights + # and also handle the case where it was already removed by another helper script + state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None) + state_dicts.append(state_dict) + + if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]: + raise ValueError(f"{files[0]} is not a zero checkpoint") + zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE] + world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT] + + # For ZeRO-2 each param group can have different partition_count as data parallelism for expert + # parameters can be different from data parallelism for non-expert parameters. So we can just + # use the max of the partition_count to get the dp world_size. + + if type(world_size) is list: + world_size = max(world_size) + + if world_size != total_files: + raise ValueError( + f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. " + "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes." + ) + + # the groups are named differently in each stage + if zero_stage <= 2: + fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS + elif zero_stage == 3: + fp32_groups_key = FP32_FLAT_GROUPS + else: + raise ValueError(f"unknown zero stage {zero_stage}") + + if zero_stage <= 2: + fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))] + elif zero_stage == 3: + # if there is more than one param group, there will be multiple flattened tensors - one + # flattened tensor per group - for simplicity merge them into a single tensor + # + # XXX: could make the script more memory efficient for when there are multiple groups - it + # will require matching the sub-lists of param_shapes for each param group flattened tensor + + fp32_flat_groups = [ + torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts)) + ] + + return zero_stage, world_size, fp32_flat_groups + + +def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters): + """ + Returns fp32 state_dict reconstructed from ds checkpoint + + Args: + - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are) + + """ + print(f"Processing zero checkpoint '{ds_checkpoint_dir}'") + + optim_files = get_optim_files(ds_checkpoint_dir) + zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir) + print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}") + + model_files = get_model_state_files(ds_checkpoint_dir) + + zero_model_states = parse_model_states(model_files) + print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}') + + if zero_stage <= 2: + return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + elif zero_stage == 3: + return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + + +def _zero2_merge_frozen_params(state_dict, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + frozen_param_fragments = zero_model_states[0].frozen_param_fragments + + if debug: + num_elem = sum(s.numel() for s in frozen_param_shapes.values()) + print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in frozen_param_fragments.values()]) + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + state_dict[name] = frozen_param_fragments[name] + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _has_callable(obj, fn): + attr = getattr(obj, fn, None) + return callable(attr) + + +def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + + # Reconstruction protocol: + # + # XXX: document this + + if debug: + for i in range(world_size): + for j in range(len(fp32_flat_groups[0])): + print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}") + + # XXX: memory usage doubles here (zero2) + num_param_groups = len(fp32_flat_groups[0]) + merged_single_partition_of_fp32_groups = [] + for i in range(num_param_groups): + merged_partitions = [sd[i] for sd in fp32_flat_groups] + full_single_fp32_vector = torch.cat(merged_partitions, 0) + merged_single_partition_of_fp32_groups.append(full_single_fp32_vector) + avail_numel = sum( + [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups]) + + if debug: + wanted_params = sum([len(shapes) for shapes in param_shapes]) + wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes]) + # not asserting if there is a mismatch due to possible padding + print(f"Have {avail_numel} numels to process.") + print(f"Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + total_numel = 0 + total_params = 0 + for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups): + offset = 0 + avail_numel = full_single_fp32_vector.numel() + for name, shape in shapes.items(): + + unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape) + total_numel += unpartitioned_numel + total_params += 1 + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape) + offset += unpartitioned_numel + + # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and + # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex + # paddings performed in the code it's almost impossible to predict the exact numbers w/o the + # live optimizer object, so we are checking that the numbers are within the right range + align_to = 2 * world_size + + def zero2_align(x): + return align_to * math.ceil(x / align_to) + + if debug: + print(f"original offset={offset}, avail_numel={avail_numel}") + + offset = zero2_align(offset) + avail_numel = zero2_align(avail_numel) + + if debug: + print(f"aligned offset={offset}, avail_numel={avail_numel}") + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero2_merge_frozen_params(state_dict, zero_model_states) + + _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def zero3_partitioned_param_info(unpartitioned_numel, world_size): + remainder = unpartitioned_numel % world_size + padding_numel = (world_size - remainder) if remainder else 0 + partitioned_numel = math.ceil(unpartitioned_numel / world_size) + return partitioned_numel, padding_numel + + +def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + if debug: + for i in range(world_size): + num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values()) + print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in zero_model_states[0].frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states) + state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape) + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + avail_numel = fp32_flat_groups[0].numel() * world_size + # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each + # param, re-consolidating each param, while dealing with padding if any + + # merge list of dicts, preserving order + param_shapes = {k: v for d in param_shapes for k, v in d.items()} + + if debug: + for i in range(world_size): + print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}") + + wanted_params = len(param_shapes) + wanted_numel = sum(shape.numel() for shape in param_shapes.values()) + # not asserting if there is a mismatch due to possible padding + avail_numel = fp32_flat_groups[0].numel() * world_size + print(f"Trainable params: Have {avail_numel} numels to process.") + print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + offset = 0 + total_numel = 0 + total_params = 0 + for name, shape in param_shapes.items(): + + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + total_params += 1 + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + # XXX: memory usage doubles here + state_dict[name] = torch.cat( + tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)), + 0).narrow(0, 0, unpartitioned_numel).view(shape) + offset += partitioned_numel + + offset *= world_size + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero3_merge_frozen_params(state_dict, world_size, zero_model_states) + + _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with + ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example + via a model hub. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + + Returns: + - pytorch ``state_dict`` + + Note: this approach may not work if your application doesn't have sufficient free CPU memory and + you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with + the checkpoint. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + # do the training and checkpoint saving + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu + model = model.cpu() # move to cpu + model.load_state_dict(state_dict) + # submit to model hub or save the model to share with others + + In this example the ``model`` will no longer be usable in the deepspeed context of the same + application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead. + + """ + if tag is None: + latest_path = os.path.join(checkpoint_dir, 'latest') + if os.path.isfile(latest_path): + with open(latest_path, 'r') as fd: + tag = fd.read().strip() + else: + raise ValueError(f"Unable to find 'latest' file at {latest_path}") + + ds_checkpoint_dir = os.path.join(checkpoint_dir, tag) + + if not os.path.isdir(ds_checkpoint_dir): + raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist") + + return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters) + + +def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be + loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + """ + + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters) + print(f"Saving fp32 state dict to {output_file}") + torch.save(state_dict, output_file) + + +def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None): + """ + 1. Put the provided model to cpu + 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` + 3. Load it into the provided model + + Args: + - ``model``: the model object to update + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + + Returns: + - ``model`: modified model + + Make sure you have plenty of CPU memory available before you call this function. If you don't + have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it + conveniently placed for you in the checkpoint folder. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint + model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir) + # submit to model hub or save the model to share with others + + Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context + of the same application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + """ + logger.info(f"Extracting fp32 weights") + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag) + + logger.info(f"Overwriting model with fp32 weights") + model = model.cpu() + model.load_state_dict(state_dict, strict=False) + + return model + + +if __name__ == "__main__": + + parser = argparse.ArgumentParser() + parser.add_argument("checkpoint_dir", + type=str, + help="path to the desired checkpoint folder, e.g., path/checkpoint-12") + parser.add_argument( + "output_file", + type=str, + help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)") + parser.add_argument("-t", + "--tag", + type=str, + default=None, + help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1") + parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters") + parser.add_argument("-d", "--debug", action='store_true', help="enable debug") + args = parser.parse_args() + + debug = args.debug + + convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, + args.output_file, + tag=args.tag, + exclude_frozen_parameters=args.exclude_frozen_parameters) diff --git a/checkpoint-800/README.md b/checkpoint-800/README.md new file mode 100644 index 0000000000000000000000000000000000000000..16b1eacdd9353dec380a08ee77ce6ed5ab50f12e --- /dev/null +++ b/checkpoint-800/README.md @@ -0,0 +1,202 @@ +--- +library_name: peft +base_model: gotzmann/uni +--- + +# Model Card for Model ID + + + + + +## Model Details + +### Model Description + + + + + +- **Developed by:** [More Information Needed] +- **Funded by [optional]:** [More Information Needed] +- **Shared by [optional]:** [More Information Needed] +- **Model type:** [More Information Needed] +- **Language(s) (NLP):** [More Information Needed] +- **License:** [More Information Needed] +- **Finetuned from model [optional]:** [More Information Needed] + +### Model Sources [optional] + + + +- **Repository:** [More Information Needed] +- **Paper [optional]:** [More Information Needed] +- **Demo [optional]:** [More Information Needed] + +## Uses + + + +### Direct Use + + + +[More Information Needed] + +### Downstream Use [optional] + + + +[More Information Needed] + +### Out-of-Scope Use + + + +[More Information Needed] + +## Bias, Risks, and Limitations + + + +[More Information Needed] + +### Recommendations + + + +Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations. + +## How to Get Started with the Model + +Use the code below to get started with the model. + +[More Information Needed] + +## Training Details + +### Training Data + + + +[More Information Needed] + +### Training Procedure + + + +#### Preprocessing [optional] + +[More Information Needed] + + +#### Training Hyperparameters + +- **Training regime:** [More Information Needed] + +#### Speeds, Sizes, Times [optional] + + + +[More Information Needed] + +## Evaluation + + + +### Testing Data, Factors & Metrics + +#### Testing Data + + + +[More Information Needed] + +#### Factors + + + +[More Information Needed] + +#### Metrics + + + +[More Information Needed] + +### Results + +[More Information Needed] + +#### Summary + + + +## Model Examination [optional] + + + +[More Information Needed] + +## Environmental Impact + + + +Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700). + +- **Hardware Type:** [More Information Needed] +- **Hours used:** [More Information Needed] +- **Cloud Provider:** [More Information Needed] +- **Compute Region:** [More Information Needed] +- **Carbon Emitted:** [More Information Needed] + +## Technical Specifications [optional] + +### Model Architecture and Objective + +[More Information Needed] + +### Compute Infrastructure + +[More Information Needed] + +#### Hardware + +[More Information Needed] + +#### Software + +[More Information Needed] + +## Citation [optional] + + + +**BibTeX:** + +[More Information Needed] + +**APA:** + +[More Information Needed] + +## Glossary [optional] + + + +[More Information Needed] + +## More Information [optional] + +[More Information Needed] + +## Model Card Authors [optional] + +[More Information Needed] + +## Model Card Contact + +[More Information Needed] +### Framework versions + +- PEFT 0.10.0 \ No newline at end of file diff --git a/checkpoint-800/adapter_config.json b/checkpoint-800/adapter_config.json new file mode 100644 index 0000000000000000000000000000000000000000..3cd6dba5d79f7ca21fd4ad465cbbcac1e0960476 --- /dev/null +++ b/checkpoint-800/adapter_config.json @@ -0,0 +1,31 @@ +{ + "alpha_pattern": {}, + "auto_mapping": null, + "base_model_name_or_path": "gotzmann/uni", + "bias": "none", + "fan_in_fan_out": false, + "inference_mode": true, + "init_lora_weights": true, + "layer_replication": null, + "layers_pattern": null, + "layers_to_transform": null, + "loftq_config": {}, + "lora_alpha": 128, + "lora_dropout": 0.0, + "megatron_config": null, + "megatron_core": "megatron.core", + "modules_to_save": null, + "peft_type": "LORA", + "r": 128, + "rank_pattern": {}, + "revision": null, + "target_modules": [ + "v_proj", + "k_proj", + "q_proj", + "o_proj" + ], + "task_type": "CAUSAL_LM", + "use_dora": false, + "use_rslora": true +} \ No newline at end of file diff --git a/checkpoint-800/adapter_model.safetensors b/checkpoint-800/adapter_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..7b809fa170dcf5c1fe34a2567be2a0ca08b1fad4 --- /dev/null +++ b/checkpoint-800/adapter_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:957309eb8fa80fd16965595c1dc6a97c9d8900131039bb55d965e13653175a9d +size 1048664848 diff --git a/checkpoint-800/global_step800/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt b/checkpoint-800/global_step800/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..e714c3aa6e55b3db5dd59e64f6d144c157a1e7aa --- /dev/null +++ b/checkpoint-800/global_step800/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:c55c954a6d1a20b8fbfe142a705e1bb1ac59400f37f3fee03c0dbed4d8d3832f +size 787270042 diff --git a/checkpoint-800/global_step800/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt b/checkpoint-800/global_step800/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..e990ad16e430b3995cf0928d5fd1dc370fdfb28b --- /dev/null +++ b/checkpoint-800/global_step800/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:c9b4722a1d11c57d0721b138b81650ec52f6e7aefb835a66cfc5d29f2d60a794 +size 787270042 diff --git a/checkpoint-800/global_step800/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt b/checkpoint-800/global_step800/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..ce01e60191d4eb91eed67fcdeada29c1500c1415 --- /dev/null +++ b/checkpoint-800/global_step800/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:48f5e2d23e98ffb6690a97e988075882dc973e879d8d4a8408403adf6dd85355 +size 787270042 diff --git a/checkpoint-800/global_step800/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt b/checkpoint-800/global_step800/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..fa8ad470cd371e12d220443d3f268d7ea9a2a322 --- /dev/null +++ b/checkpoint-800/global_step800/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2950e806e878460d1e46b94f05543dd19cf944318887735bc0cbca1e07e0ba58 +size 787270042 diff --git a/checkpoint-800/global_step800/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt b/checkpoint-800/global_step800/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..2e74300dda38ec1c331372108eec284b90e148a2 --- /dev/null +++ b/checkpoint-800/global_step800/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:279bbcae36e6e4ee14fee00c972840eaf4b1a59370003cef1c74026c31c378ff +size 787270042 diff --git a/checkpoint-800/global_step800/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt b/checkpoint-800/global_step800/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..87593dc746f6b784f2fd57d6c1322fbc79323e94 --- /dev/null +++ b/checkpoint-800/global_step800/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1aa9b1733767642452406caf7e6205632eb28e244cc44660fe32442be16a48b6 +size 787270042 diff --git a/checkpoint-800/global_step800/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt b/checkpoint-800/global_step800/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..180241eba094120819bef8ea4134e63de403681f --- /dev/null +++ b/checkpoint-800/global_step800/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3e890fac1e94ff55df91ae0554dc5c6f7520d375caab14aac7f60b135601cad2 +size 787270042 diff --git a/checkpoint-800/global_step800/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt b/checkpoint-800/global_step800/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..1f3142cbf64f6387bbb75e6fd9c665ec778785a8 --- /dev/null +++ b/checkpoint-800/global_step800/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9b09008e7481896f3e820578250007c53aa36ce5b02f98beef749b855be4783b +size 787270042 diff --git a/checkpoint-800/global_step800/zero_pp_rank_0_mp_rank_00_model_states.pt b/checkpoint-800/global_step800/zero_pp_rank_0_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..9b11d0d44bbae33e5bc118045048e8845a542c64 --- /dev/null +++ b/checkpoint-800/global_step800/zero_pp_rank_0_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c577fddf8e24a7ffa762f5719315f724538aa93576ecfcbf0032d25212a3ce4a +size 653742 diff --git a/checkpoint-800/global_step800/zero_pp_rank_1_mp_rank_00_model_states.pt b/checkpoint-800/global_step800/zero_pp_rank_1_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..2e4cd0e5ea0b57263d79c4bf375f9da70b304278 --- /dev/null +++ b/checkpoint-800/global_step800/zero_pp_rank_1_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:20b0c382515475979998f4d222f0b235c61aaa493c54354fb5551eeac507f756 +size 653742 diff --git a/checkpoint-800/global_step800/zero_pp_rank_2_mp_rank_00_model_states.pt b/checkpoint-800/global_step800/zero_pp_rank_2_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..1513b1cdd8ff350ab853d1624ca6573d1192d585 --- /dev/null +++ b/checkpoint-800/global_step800/zero_pp_rank_2_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8e6cca9ad652cb5a3e1a28b18d9545a9750a3004db04fd38e1ef2ee5eccfb562 +size 653742 diff --git a/checkpoint-800/global_step800/zero_pp_rank_3_mp_rank_00_model_states.pt b/checkpoint-800/global_step800/zero_pp_rank_3_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..ac86b1cc4e75211cd593546784ef5b815018c809 --- /dev/null +++ b/checkpoint-800/global_step800/zero_pp_rank_3_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3c1b5dcaa027488d8d6bc110cd54cfb85ddf3527cd94aa9429f04aebc6fca723 +size 653742 diff --git a/checkpoint-800/global_step800/zero_pp_rank_4_mp_rank_00_model_states.pt b/checkpoint-800/global_step800/zero_pp_rank_4_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..fdf666fe4e3517fb692d423d5887d312a29dcac6 --- /dev/null +++ b/checkpoint-800/global_step800/zero_pp_rank_4_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dc878f32f7ef72294ca96f298fa1c7fcfbf2b6e425304525a7e2fa197bbd5d55 +size 653742 diff --git a/checkpoint-800/global_step800/zero_pp_rank_5_mp_rank_00_model_states.pt b/checkpoint-800/global_step800/zero_pp_rank_5_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..245a928f1cc458a8716ad046bb0d58abd995ed61 --- /dev/null +++ b/checkpoint-800/global_step800/zero_pp_rank_5_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:350a480253a8dcec7cbf69b13ae08b32017dd165f7e112f0c1999a7c50e1cdeb +size 653742 diff --git a/checkpoint-800/global_step800/zero_pp_rank_6_mp_rank_00_model_states.pt b/checkpoint-800/global_step800/zero_pp_rank_6_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..f05c3303bcd651cc9fe267280eb2f663de2479e8 --- /dev/null +++ b/checkpoint-800/global_step800/zero_pp_rank_6_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ccd1e8726e8abaa96f95daf2437a82f4c3b6295defc800b6d6ada66800c7ad59 +size 653742 diff --git a/checkpoint-800/global_step800/zero_pp_rank_7_mp_rank_00_model_states.pt b/checkpoint-800/global_step800/zero_pp_rank_7_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..8229754e8cf24d583b62eb943576567db0283a1f --- /dev/null +++ b/checkpoint-800/global_step800/zero_pp_rank_7_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4233f2bed13c6017800b385c409efe11bc60406831614dc2aff33f60e00c00f4 +size 653742 diff --git a/checkpoint-800/latest b/checkpoint-800/latest new file mode 100644 index 0000000000000000000000000000000000000000..57729c0be88118cbd582c8c68b4149cee821f0b4 --- /dev/null +++ b/checkpoint-800/latest @@ -0,0 +1 @@ +global_step800 \ No newline at end of file diff --git a/checkpoint-800/rng_state_0.pth b/checkpoint-800/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..4e5b7e2ec90fdb824c8932464c1d9068330655a7 --- /dev/null +++ b/checkpoint-800/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:36d2a2034ebb05cb71c510897f2795b31164e50f17b270bc25d2be3ad9a17b22 +size 15984 diff --git a/checkpoint-800/rng_state_1.pth b/checkpoint-800/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..7d8d7722fc72cab6d492b76cb99c8177dcc47544 --- /dev/null +++ b/checkpoint-800/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:060dfdb1c49102cbdc8868a6031e68787601b4ccd782f3fb9b137e20c1fd2c7a +size 15984 diff --git a/checkpoint-800/rng_state_2.pth b/checkpoint-800/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..3c9f84eff30cfa9ea1feedaf262d61fb12e4cba7 --- /dev/null +++ b/checkpoint-800/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:af01895cb66e616591f2e4baa8dcd8151530eab133c73571ccb31c74f35422ce +size 15984 diff --git a/checkpoint-800/rng_state_3.pth b/checkpoint-800/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..6eebfb928f8e91eff0ea1645a20b5aa4465c705b --- /dev/null +++ b/checkpoint-800/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:677921992b1e0cef3aee776f245975003d22f51d9bd6ed20f248ded1deb72fa9 +size 15984 diff --git a/checkpoint-800/rng_state_4.pth b/checkpoint-800/rng_state_4.pth new file mode 100644 index 0000000000000000000000000000000000000000..0866030a266c6d003cc378a9418a723f69e8ab99 --- /dev/null +++ b/checkpoint-800/rng_state_4.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d69353c629541c690c5471f8ec05fdab2bfecf3d37afaa436bc45939da6db68f +size 15984 diff --git a/checkpoint-800/rng_state_5.pth b/checkpoint-800/rng_state_5.pth new file mode 100644 index 0000000000000000000000000000000000000000..554638d77107f832d7aa51c61645ee2d6c48a36d --- /dev/null +++ b/checkpoint-800/rng_state_5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8e40ba6668cc03c9162c68a933d164bf38ae2d196a9a6fec03ae615491201185 +size 15984 diff --git a/checkpoint-800/rng_state_6.pth b/checkpoint-800/rng_state_6.pth new file mode 100644 index 0000000000000000000000000000000000000000..964331b65172a1bcac03e4673415fa787f724268 --- /dev/null +++ b/checkpoint-800/rng_state_6.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:870968fea834e24b2e099cf3e4fe1e3fb8caf38d8f8e5b790d7d47386d4d05f5 +size 15984 diff --git a/checkpoint-800/rng_state_7.pth b/checkpoint-800/rng_state_7.pth new file mode 100644 index 0000000000000000000000000000000000000000..cd4754d65217d0f9d1f2d3334397df7a8a079652 --- /dev/null +++ b/checkpoint-800/rng_state_7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e9e19618bee7c6ef43256fea25abe19bca88535eb1e7dc213cde8929ae4e8180 +size 15984 diff --git a/checkpoint-800/scheduler.pt b/checkpoint-800/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..c8f56f167203bea14f45dcb6f31e429f6898e67a --- /dev/null +++ b/checkpoint-800/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9bf60fed2d5b3076be79b7c1fa2a43fe5a2f69188a45ae24f4bfafe186448b99 +size 1064 diff --git a/checkpoint-800/special_tokens_map.json b/checkpoint-800/special_tokens_map.json new file mode 100644 index 0000000000000000000000000000000000000000..72ecfeeb7e14d244c936169d2ed139eeae235ef1 --- /dev/null +++ b/checkpoint-800/special_tokens_map.json @@ -0,0 +1,24 @@ +{ + "bos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "eos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "pad_token": "", + "unk_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + } +} diff --git a/checkpoint-800/tokenizer.model b/checkpoint-800/tokenizer.model new file mode 100644 index 0000000000000000000000000000000000000000..6c00c742ce03c627d6cd5b795984876fa49fa899 --- /dev/null +++ b/checkpoint-800/tokenizer.model @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347 +size 499723 diff --git a/checkpoint-800/tokenizer_config.json b/checkpoint-800/tokenizer_config.json new file mode 100644 index 0000000000000000000000000000000000000000..bb5a9f09d8c0f3c32c66fc6118fe5c76c5c6fd90 --- /dev/null +++ b/checkpoint-800/tokenizer_config.json @@ -0,0 +1,45 @@ +{ + "add_bos_token": true, + "add_eos_token": false, + "add_prefix_space": false, + "added_tokens_decoder": { + "0": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "1": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "2": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + } + }, + "bos_token": "", + "chat_template": "{% if messages[0]['role'] == 'system' %}{% set system_message = messages[0]['content'] %}{% endif %}{% if system_message is defined %}{{ '' + '### System:\\n\\n' + system_message }}{% endif %}{% for message in messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ '\\n\\n### Human:\\n\\n' + content }}{% elif message['role'] == 'assistant' %}{{ '\\n\\n### Assistant:\\n\\n' + content + '' }}{% endif %}{% endfor %}", + "clean_up_tokenization_spaces": false, + "eos_token": "", + "legacy": false, + "model_max_length": 1000000000000000019884624838656, + "pad_token": "", + "padding_side": "right", + "sp_model_kwargs": {}, + "spaces_between_special_tokens": false, + "split_special_tokens": false, + "tokenizer_class": "LlamaTokenizer", + "unk_token": "", + "use_default_system_prompt": false +} diff --git a/checkpoint-800/trainer_state.json b/checkpoint-800/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..a03f0713c43233653146668b3dbe1287e237e421 --- /dev/null +++ b/checkpoint-800/trainer_state.json @@ -0,0 +1,5621 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 1.4631915866483767, + "eval_steps": 500, + "global_step": 800, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "grad_norm": 0.849355824164473, + "learning_rate": 4.878048780487805e-07, + "loss": 1.3655, + "step": 1 + }, + { + "epoch": 0.0, + "grad_norm": 10.01567518957158, + "learning_rate": 9.75609756097561e-07, + "loss": 1.5767, + "step": 2 + }, + { + "epoch": 0.01, + "grad_norm": 0.6466000875559635, + "learning_rate": 1.4634146341463414e-06, + "loss": 1.3913, + "step": 3 + }, + { + "epoch": 0.01, + "grad_norm": 0.6644565932010504, + "learning_rate": 1.951219512195122e-06, + "loss": 1.3218, + "step": 4 + }, + { + "epoch": 0.01, + "grad_norm": 0.571354207588475, + "learning_rate": 2.4390243902439027e-06, + "loss": 1.3597, + "step": 5 + }, + { + "epoch": 0.01, + "grad_norm": 0.31036262839244955, + "learning_rate": 2.926829268292683e-06, + "loss": 1.2832, + "step": 6 + }, + { + "epoch": 0.01, + "grad_norm": 0.2622135027188184, + "learning_rate": 3.414634146341464e-06, + "loss": 1.2161, + "step": 7 + }, + { + "epoch": 0.01, + "grad_norm": 0.296824630261661, + "learning_rate": 3.902439024390244e-06, + "loss": 1.2985, + "step": 8 + }, + { + "epoch": 0.02, + "grad_norm": 0.2557267467361569, + "learning_rate": 4.390243902439025e-06, + "loss": 1.3175, + "step": 9 + }, + { + "epoch": 0.02, + "grad_norm": 0.23418939513890769, + "learning_rate": 4.8780487804878055e-06, + "loss": 1.2617, + "step": 10 + }, + { + "epoch": 0.02, + "grad_norm": 0.2364760983285843, + "learning_rate": 5.365853658536586e-06, + "loss": 1.3103, + "step": 11 + }, + { + "epoch": 0.02, + "grad_norm": 0.23893034721889, + "learning_rate": 5.853658536585366e-06, + "loss": 1.2405, + "step": 12 + }, + { + "epoch": 0.02, + "grad_norm": 0.25563593295485887, + "learning_rate": 6.341463414634147e-06, + "loss": 1.2831, + "step": 13 + }, + { + "epoch": 0.03, + "grad_norm": 0.23239975352661665, + "learning_rate": 6.829268292682928e-06, + "loss": 1.3125, + "step": 14 + }, + { + "epoch": 0.03, + "grad_norm": 0.3092813858209507, + "learning_rate": 7.317073170731707e-06, + "loss": 1.2422, + "step": 15 + }, + { + "epoch": 0.03, + "grad_norm": 0.282563380367434, + "learning_rate": 7.804878048780489e-06, + "loss": 1.2453, + "step": 16 + }, + { + "epoch": 0.03, + "grad_norm": 0.22065680088315018, + "learning_rate": 8.292682926829268e-06, + "loss": 1.2491, + "step": 17 + }, + { + "epoch": 0.03, + "grad_norm": 0.22777800877980184, + "learning_rate": 8.78048780487805e-06, + "loss": 1.2655, + "step": 18 + }, + { + "epoch": 0.03, + "grad_norm": 0.22145212540177928, + "learning_rate": 9.268292682926831e-06, + "loss": 1.2413, + "step": 19 + }, + { + "epoch": 0.04, + "grad_norm": 0.22482351883112714, + "learning_rate": 9.756097560975611e-06, + "loss": 1.2653, + "step": 20 + }, + { + "epoch": 0.04, + "grad_norm": 0.20823080508385733, + "learning_rate": 1.024390243902439e-05, + "loss": 1.2374, + "step": 21 + }, + { + "epoch": 0.04, + "grad_norm": 0.26025492562935737, + "learning_rate": 1.0731707317073172e-05, + "loss": 1.2065, + "step": 22 + }, + { + "epoch": 0.04, + "grad_norm": 0.2150252124176173, + "learning_rate": 1.1219512195121953e-05, + "loss": 1.2782, + "step": 23 + }, + { + "epoch": 0.04, + "grad_norm": 0.2505915177425618, + "learning_rate": 1.1707317073170731e-05, + "loss": 1.2742, + "step": 24 + }, + { + "epoch": 0.05, + "grad_norm": 0.20129223044786942, + "learning_rate": 1.2195121951219513e-05, + "loss": 1.3366, + "step": 25 + }, + { + "epoch": 0.05, + "grad_norm": 0.1973508510397107, + "learning_rate": 1.2682926829268294e-05, + "loss": 1.2476, + "step": 26 + }, + { + "epoch": 0.05, + "grad_norm": 0.27103325392437194, + "learning_rate": 1.3170731707317076e-05, + "loss": 1.2325, + "step": 27 + }, + { + "epoch": 0.05, + "grad_norm": 0.17954976411006285, + "learning_rate": 1.3658536585365855e-05, + "loss": 1.2523, + "step": 28 + }, + { + "epoch": 0.05, + "grad_norm": 0.22216997851088888, + "learning_rate": 1.4146341463414635e-05, + "loss": 1.3297, + "step": 29 + }, + { + "epoch": 0.05, + "grad_norm": 0.2071458864548587, + "learning_rate": 1.4634146341463415e-05, + "loss": 1.2127, + "step": 30 + }, + { + "epoch": 0.06, + "grad_norm": 0.18039422081622164, + "learning_rate": 1.5121951219512196e-05, + "loss": 1.2509, + "step": 31 + }, + { + "epoch": 0.06, + "grad_norm": 0.18631254372974412, + "learning_rate": 1.5609756097560978e-05, + "loss": 1.2247, + "step": 32 + }, + { + "epoch": 0.06, + "grad_norm": 0.18843872523649827, + "learning_rate": 1.6097560975609757e-05, + "loss": 1.195, + "step": 33 + }, + { + "epoch": 0.06, + "grad_norm": 0.2163847267778325, + "learning_rate": 1.6585365853658537e-05, + "loss": 1.2179, + "step": 34 + }, + { + "epoch": 0.06, + "grad_norm": 0.19687688475496104, + "learning_rate": 1.7073170731707317e-05, + "loss": 1.2763, + "step": 35 + }, + { + "epoch": 0.07, + "grad_norm": 0.20409643064887947, + "learning_rate": 1.75609756097561e-05, + "loss": 1.253, + "step": 36 + }, + { + "epoch": 0.07, + "grad_norm": 0.1879182661759335, + "learning_rate": 1.804878048780488e-05, + "loss": 1.2586, + "step": 37 + }, + { + "epoch": 0.07, + "grad_norm": 0.19400648948514373, + "learning_rate": 1.8536585365853663e-05, + "loss": 1.2154, + "step": 38 + }, + { + "epoch": 0.07, + "grad_norm": 0.1878879343148452, + "learning_rate": 1.902439024390244e-05, + "loss": 1.2304, + "step": 39 + }, + { + "epoch": 0.07, + "grad_norm": 0.17687475469924052, + "learning_rate": 1.9512195121951222e-05, + "loss": 1.2351, + "step": 40 + }, + { + "epoch": 0.07, + "grad_norm": 0.18223935625384885, + "learning_rate": 2e-05, + "loss": 1.2222, + "step": 41 + }, + { + "epoch": 0.08, + "grad_norm": 0.1943061629408338, + "learning_rate": 2.048780487804878e-05, + "loss": 1.2044, + "step": 42 + }, + { + "epoch": 0.08, + "grad_norm": 0.17027514338700078, + "learning_rate": 2.0975609756097564e-05, + "loss": 1.1548, + "step": 43 + }, + { + "epoch": 0.08, + "grad_norm": 0.18553769630586192, + "learning_rate": 2.1463414634146344e-05, + "loss": 1.2721, + "step": 44 + }, + { + "epoch": 0.08, + "grad_norm": 0.19732826914228765, + "learning_rate": 2.1951219512195124e-05, + "loss": 1.3097, + "step": 45 + }, + { + "epoch": 0.08, + "grad_norm": 0.18714230986631472, + "learning_rate": 2.2439024390243907e-05, + "loss": 1.2662, + "step": 46 + }, + { + "epoch": 0.09, + "grad_norm": 0.19988987568002223, + "learning_rate": 2.2926829268292683e-05, + "loss": 1.2904, + "step": 47 + }, + { + "epoch": 0.09, + "grad_norm": 0.17744650133390918, + "learning_rate": 2.3414634146341463e-05, + "loss": 1.1825, + "step": 48 + }, + { + "epoch": 0.09, + "grad_norm": 0.16576734763834533, + "learning_rate": 2.3902439024390246e-05, + "loss": 1.1858, + "step": 49 + }, + { + "epoch": 0.09, + "grad_norm": 0.179591794065527, + "learning_rate": 2.4390243902439026e-05, + "loss": 1.2711, + "step": 50 + }, + { + "epoch": 0.09, + "grad_norm": 0.17923464471176911, + "learning_rate": 2.4878048780487805e-05, + "loss": 1.2289, + "step": 51 + }, + { + "epoch": 0.1, + "grad_norm": 0.18991742907836837, + "learning_rate": 2.536585365853659e-05, + "loss": 1.3097, + "step": 52 + }, + { + "epoch": 0.1, + "grad_norm": 0.19849796137254636, + "learning_rate": 2.5853658536585368e-05, + "loss": 1.2489, + "step": 53 + }, + { + "epoch": 0.1, + "grad_norm": 0.17452371110976383, + "learning_rate": 2.634146341463415e-05, + "loss": 1.2461, + "step": 54 + }, + { + "epoch": 0.1, + "grad_norm": 0.17671022353085036, + "learning_rate": 2.682926829268293e-05, + "loss": 1.153, + "step": 55 + }, + { + "epoch": 0.1, + "grad_norm": 0.36820559192096686, + "learning_rate": 2.731707317073171e-05, + "loss": 1.2431, + "step": 56 + }, + { + "epoch": 0.1, + "grad_norm": 0.20331468526494198, + "learning_rate": 2.7804878048780487e-05, + "loss": 1.2575, + "step": 57 + }, + { + "epoch": 0.11, + "grad_norm": 0.2402486598118377, + "learning_rate": 2.829268292682927e-05, + "loss": 1.2538, + "step": 58 + }, + { + "epoch": 0.11, + "grad_norm": 0.2549409484173144, + "learning_rate": 2.878048780487805e-05, + "loss": 1.2065, + "step": 59 + }, + { + "epoch": 0.11, + "grad_norm": 0.2053105349872685, + "learning_rate": 2.926829268292683e-05, + "loss": 1.2094, + "step": 60 + }, + { + "epoch": 0.11, + "grad_norm": 0.17971910872957886, + "learning_rate": 2.9756097560975613e-05, + "loss": 1.228, + "step": 61 + }, + { + "epoch": 0.11, + "grad_norm": 0.1885853654992973, + "learning_rate": 3.0243902439024392e-05, + "loss": 1.2286, + "step": 62 + }, + { + "epoch": 0.12, + "grad_norm": 0.1848524571968613, + "learning_rate": 3.073170731707317e-05, + "loss": 1.2718, + "step": 63 + }, + { + "epoch": 0.12, + "grad_norm": 0.18734105883548513, + "learning_rate": 3.1219512195121955e-05, + "loss": 1.2357, + "step": 64 + }, + { + "epoch": 0.12, + "grad_norm": 0.17774668052121825, + "learning_rate": 3.170731707317074e-05, + "loss": 1.1509, + "step": 65 + }, + { + "epoch": 0.12, + "grad_norm": 0.17890968008080646, + "learning_rate": 3.2195121951219514e-05, + "loss": 1.1924, + "step": 66 + }, + { + "epoch": 0.12, + "grad_norm": 0.18249273371332375, + "learning_rate": 3.268292682926829e-05, + "loss": 1.2545, + "step": 67 + }, + { + "epoch": 0.12, + "grad_norm": 0.21064122671902577, + "learning_rate": 3.3170731707317074e-05, + "loss": 1.2832, + "step": 68 + }, + { + "epoch": 0.13, + "grad_norm": 0.1820064171955093, + "learning_rate": 3.365853658536586e-05, + "loss": 1.2071, + "step": 69 + }, + { + "epoch": 0.13, + "grad_norm": 0.16996662800553433, + "learning_rate": 3.414634146341463e-05, + "loss": 1.2073, + "step": 70 + }, + { + "epoch": 0.13, + "grad_norm": 0.1618669302922445, + "learning_rate": 3.4634146341463416e-05, + "loss": 1.1289, + "step": 71 + }, + { + "epoch": 0.13, + "grad_norm": 0.18948744950985544, + "learning_rate": 3.51219512195122e-05, + "loss": 1.2915, + "step": 72 + }, + { + "epoch": 0.13, + "grad_norm": 0.18326143691603383, + "learning_rate": 3.5609756097560976e-05, + "loss": 1.2238, + "step": 73 + }, + { + "epoch": 0.14, + "grad_norm": 0.17410704510700503, + "learning_rate": 3.609756097560976e-05, + "loss": 1.1784, + "step": 74 + }, + { + "epoch": 0.14, + "grad_norm": 0.1983667344995625, + "learning_rate": 3.658536585365854e-05, + "loss": 1.2452, + "step": 75 + }, + { + "epoch": 0.14, + "grad_norm": 0.3416310763369357, + "learning_rate": 3.7073170731707325e-05, + "loss": 1.1972, + "step": 76 + }, + { + "epoch": 0.14, + "grad_norm": 0.2776466983511955, + "learning_rate": 3.75609756097561e-05, + "loss": 1.3121, + "step": 77 + }, + { + "epoch": 0.14, + "grad_norm": 0.20026129636576834, + "learning_rate": 3.804878048780488e-05, + "loss": 1.2436, + "step": 78 + }, + { + "epoch": 0.14, + "grad_norm": 0.21064549243917835, + "learning_rate": 3.853658536585366e-05, + "loss": 1.2064, + "step": 79 + }, + { + "epoch": 0.15, + "grad_norm": 0.22119482175714267, + "learning_rate": 3.9024390243902444e-05, + "loss": 1.2715, + "step": 80 + }, + { + "epoch": 0.15, + "grad_norm": 0.23047133748844142, + "learning_rate": 3.951219512195122e-05, + "loss": 1.2888, + "step": 81 + }, + { + "epoch": 0.15, + "grad_norm": 0.18741863156973176, + "learning_rate": 4e-05, + "loss": 1.248, + "step": 82 + }, + { + "epoch": 0.15, + "grad_norm": 0.1747859810629604, + "learning_rate": 4.0487804878048786e-05, + "loss": 1.1683, + "step": 83 + }, + { + "epoch": 0.15, + "grad_norm": 0.1896944798413341, + "learning_rate": 4.097560975609756e-05, + "loss": 1.2155, + "step": 84 + }, + { + "epoch": 0.16, + "grad_norm": 0.18724128114363303, + "learning_rate": 4.1463414634146346e-05, + "loss": 1.2273, + "step": 85 + }, + { + "epoch": 0.16, + "grad_norm": 0.17368125504855478, + "learning_rate": 4.195121951219513e-05, + "loss": 1.224, + "step": 86 + }, + { + "epoch": 0.16, + "grad_norm": 0.18371141013625703, + "learning_rate": 4.2439024390243905e-05, + "loss": 1.2294, + "step": 87 + }, + { + "epoch": 0.16, + "grad_norm": 0.1791029365673714, + "learning_rate": 4.292682926829269e-05, + "loss": 1.2895, + "step": 88 + }, + { + "epoch": 0.16, + "grad_norm": 0.20259974283859655, + "learning_rate": 4.341463414634147e-05, + "loss": 1.1841, + "step": 89 + }, + { + "epoch": 0.16, + "grad_norm": 0.17457456183272174, + "learning_rate": 4.390243902439025e-05, + "loss": 1.2357, + "step": 90 + }, + { + "epoch": 0.17, + "grad_norm": 0.1815824380789748, + "learning_rate": 4.439024390243903e-05, + "loss": 1.2304, + "step": 91 + }, + { + "epoch": 0.17, + "grad_norm": 0.17566480599583392, + "learning_rate": 4.4878048780487814e-05, + "loss": 1.242, + "step": 92 + }, + { + "epoch": 0.17, + "grad_norm": 0.18422975005984474, + "learning_rate": 4.536585365853658e-05, + "loss": 1.2177, + "step": 93 + }, + { + "epoch": 0.17, + "grad_norm": 0.16796781877940678, + "learning_rate": 4.5853658536585366e-05, + "loss": 1.1482, + "step": 94 + }, + { + "epoch": 0.17, + "grad_norm": 0.18636131653783305, + "learning_rate": 4.634146341463415e-05, + "loss": 1.1758, + "step": 95 + }, + { + "epoch": 0.18, + "grad_norm": 0.1823665700289814, + "learning_rate": 4.6829268292682926e-05, + "loss": 1.289, + "step": 96 + }, + { + "epoch": 0.18, + "grad_norm": 0.1719900691262439, + "learning_rate": 4.731707317073171e-05, + "loss": 1.1626, + "step": 97 + }, + { + "epoch": 0.18, + "grad_norm": 0.17937994168039778, + "learning_rate": 4.780487804878049e-05, + "loss": 1.175, + "step": 98 + }, + { + "epoch": 0.18, + "grad_norm": 0.16631851422106986, + "learning_rate": 4.829268292682927e-05, + "loss": 1.2177, + "step": 99 + }, + { + "epoch": 0.18, + "grad_norm": 0.19143696232800309, + "learning_rate": 4.878048780487805e-05, + "loss": 1.3071, + "step": 100 + }, + { + "epoch": 0.18, + "grad_norm": 0.17859506638780318, + "learning_rate": 4.9268292682926835e-05, + "loss": 1.2351, + "step": 101 + }, + { + "epoch": 0.19, + "grad_norm": 0.18381520321248196, + "learning_rate": 4.975609756097561e-05, + "loss": 1.2342, + "step": 102 + }, + { + "epoch": 0.19, + "grad_norm": 0.17968218683773912, + "learning_rate": 5.0243902439024394e-05, + "loss": 1.2074, + "step": 103 + }, + { + "epoch": 0.19, + "grad_norm": 0.18139489969339018, + "learning_rate": 5.073170731707318e-05, + "loss": 1.1558, + "step": 104 + }, + { + "epoch": 0.19, + "grad_norm": 0.17366624842514394, + "learning_rate": 5.121951219512195e-05, + "loss": 1.1897, + "step": 105 + }, + { + "epoch": 0.19, + "grad_norm": 0.16034845455223745, + "learning_rate": 5.1707317073170736e-05, + "loss": 1.179, + "step": 106 + }, + { + "epoch": 0.2, + "grad_norm": 0.17583069577827776, + "learning_rate": 5.219512195121952e-05, + "loss": 1.1856, + "step": 107 + }, + { + "epoch": 0.2, + "grad_norm": 0.1853758076989552, + "learning_rate": 5.26829268292683e-05, + "loss": 1.2072, + "step": 108 + }, + { + "epoch": 0.2, + "grad_norm": 0.19597443965936462, + "learning_rate": 5.317073170731708e-05, + "loss": 1.2271, + "step": 109 + }, + { + "epoch": 0.2, + "grad_norm": 0.1899206334098331, + "learning_rate": 5.365853658536586e-05, + "loss": 1.1961, + "step": 110 + }, + { + "epoch": 0.2, + "grad_norm": 0.17463763837757018, + "learning_rate": 5.4146341463414645e-05, + "loss": 1.2049, + "step": 111 + }, + { + "epoch": 0.2, + "grad_norm": 0.20431371701229986, + "learning_rate": 5.463414634146342e-05, + "loss": 1.2891, + "step": 112 + }, + { + "epoch": 0.21, + "grad_norm": 0.1814475107638498, + "learning_rate": 5.51219512195122e-05, + "loss": 1.2346, + "step": 113 + }, + { + "epoch": 0.21, + "grad_norm": 0.1883849423207823, + "learning_rate": 5.5609756097560974e-05, + "loss": 1.244, + "step": 114 + }, + { + "epoch": 0.21, + "grad_norm": 0.1857258128640568, + "learning_rate": 5.609756097560976e-05, + "loss": 1.2669, + "step": 115 + }, + { + "epoch": 0.21, + "grad_norm": 0.1740768514118401, + "learning_rate": 5.658536585365854e-05, + "loss": 1.2414, + "step": 116 + }, + { + "epoch": 0.21, + "grad_norm": 0.1919320335584178, + "learning_rate": 5.7073170731707317e-05, + "loss": 1.2886, + "step": 117 + }, + { + "epoch": 0.22, + "grad_norm": 0.18288775167828136, + "learning_rate": 5.75609756097561e-05, + "loss": 1.1875, + "step": 118 + }, + { + "epoch": 0.22, + "grad_norm": 0.18208588867750863, + "learning_rate": 5.804878048780488e-05, + "loss": 1.2388, + "step": 119 + }, + { + "epoch": 0.22, + "grad_norm": 0.1743260015658331, + "learning_rate": 5.853658536585366e-05, + "loss": 1.1762, + "step": 120 + }, + { + "epoch": 0.22, + "grad_norm": 0.17856046291517946, + "learning_rate": 5.902439024390244e-05, + "loss": 1.2888, + "step": 121 + }, + { + "epoch": 0.22, + "grad_norm": 0.17493794870966536, + "learning_rate": 5.9512195121951225e-05, + "loss": 1.2222, + "step": 122 + }, + { + "epoch": 0.22, + "grad_norm": 0.1909202655203384, + "learning_rate": 6.000000000000001e-05, + "loss": 1.2414, + "step": 123 + }, + { + "epoch": 0.23, + "grad_norm": 0.18345819482834988, + "learning_rate": 6.0487804878048785e-05, + "loss": 1.2756, + "step": 124 + }, + { + "epoch": 0.23, + "grad_norm": 0.2057069352956621, + "learning_rate": 6.097560975609757e-05, + "loss": 1.261, + "step": 125 + }, + { + "epoch": 0.23, + "grad_norm": 0.299775882469108, + "learning_rate": 6.146341463414634e-05, + "loss": 1.2566, + "step": 126 + }, + { + "epoch": 0.23, + "grad_norm": 0.1869687633018095, + "learning_rate": 6.195121951219513e-05, + "loss": 1.3039, + "step": 127 + }, + { + "epoch": 0.23, + "grad_norm": 0.17747149926197442, + "learning_rate": 6.243902439024391e-05, + "loss": 1.2524, + "step": 128 + }, + { + "epoch": 0.24, + "grad_norm": 0.17885157788044242, + "learning_rate": 6.29268292682927e-05, + "loss": 1.2455, + "step": 129 + }, + { + "epoch": 0.24, + "grad_norm": 0.17617298187845123, + "learning_rate": 6.341463414634148e-05, + "loss": 1.2009, + "step": 130 + }, + { + "epoch": 0.24, + "grad_norm": 0.20164176323497066, + "learning_rate": 6.390243902439025e-05, + "loss": 1.2634, + "step": 131 + }, + { + "epoch": 0.24, + "grad_norm": 0.20459903417307612, + "learning_rate": 6.439024390243903e-05, + "loss": 1.1963, + "step": 132 + }, + { + "epoch": 0.24, + "grad_norm": 0.1863755486334296, + "learning_rate": 6.487804878048781e-05, + "loss": 1.2387, + "step": 133 + }, + { + "epoch": 0.25, + "grad_norm": 0.19265866140295207, + "learning_rate": 6.536585365853658e-05, + "loss": 1.2688, + "step": 134 + }, + { + "epoch": 0.25, + "grad_norm": 0.1823425868969493, + "learning_rate": 6.585365853658536e-05, + "loss": 1.2041, + "step": 135 + }, + { + "epoch": 0.25, + "grad_norm": 0.2016853266472781, + "learning_rate": 6.634146341463415e-05, + "loss": 1.1223, + "step": 136 + }, + { + "epoch": 0.25, + "grad_norm": 0.17282675192463448, + "learning_rate": 6.682926829268293e-05, + "loss": 1.1879, + "step": 137 + }, + { + "epoch": 0.25, + "grad_norm": 0.17398811693399288, + "learning_rate": 6.731707317073171e-05, + "loss": 1.2682, + "step": 138 + }, + { + "epoch": 0.25, + "grad_norm": 0.18516916965434696, + "learning_rate": 6.78048780487805e-05, + "loss": 1.1666, + "step": 139 + }, + { + "epoch": 0.26, + "grad_norm": 0.1852213129647933, + "learning_rate": 6.829268292682927e-05, + "loss": 1.2501, + "step": 140 + }, + { + "epoch": 0.26, + "grad_norm": 0.17915948766591883, + "learning_rate": 6.878048780487805e-05, + "loss": 1.2264, + "step": 141 + }, + { + "epoch": 0.26, + "grad_norm": 0.21599939417233183, + "learning_rate": 6.926829268292683e-05, + "loss": 1.2376, + "step": 142 + }, + { + "epoch": 0.26, + "grad_norm": 0.17839304459521851, + "learning_rate": 6.975609756097562e-05, + "loss": 1.2353, + "step": 143 + }, + { + "epoch": 0.26, + "grad_norm": 0.20826913231380875, + "learning_rate": 7.02439024390244e-05, + "loss": 1.1901, + "step": 144 + }, + { + "epoch": 0.27, + "grad_norm": 0.20788894913361589, + "learning_rate": 7.073170731707318e-05, + "loss": 1.2577, + "step": 145 + }, + { + "epoch": 0.27, + "grad_norm": 0.18420055842301297, + "learning_rate": 7.121951219512195e-05, + "loss": 1.1393, + "step": 146 + }, + { + "epoch": 0.27, + "grad_norm": 0.19903048468685589, + "learning_rate": 7.170731707317073e-05, + "loss": 1.2321, + "step": 147 + }, + { + "epoch": 0.27, + "grad_norm": 0.19074116314985748, + "learning_rate": 7.219512195121952e-05, + "loss": 1.1912, + "step": 148 + }, + { + "epoch": 0.27, + "grad_norm": 0.2353816469403903, + "learning_rate": 7.26829268292683e-05, + "loss": 1.28, + "step": 149 + }, + { + "epoch": 0.27, + "grad_norm": 0.21634875684769345, + "learning_rate": 7.317073170731708e-05, + "loss": 1.3312, + "step": 150 + }, + { + "epoch": 0.28, + "grad_norm": 0.18290969006743918, + "learning_rate": 7.365853658536587e-05, + "loss": 1.2214, + "step": 151 + }, + { + "epoch": 0.28, + "grad_norm": 0.18484243897545208, + "learning_rate": 7.414634146341465e-05, + "loss": 1.1895, + "step": 152 + }, + { + "epoch": 0.28, + "grad_norm": 0.21882343112978872, + "learning_rate": 7.463414634146342e-05, + "loss": 1.2219, + "step": 153 + }, + { + "epoch": 0.28, + "grad_norm": 0.19868284379241205, + "learning_rate": 7.51219512195122e-05, + "loss": 1.2176, + "step": 154 + }, + { + "epoch": 0.28, + "grad_norm": 0.20912516312950613, + "learning_rate": 7.560975609756097e-05, + "loss": 1.242, + "step": 155 + }, + { + "epoch": 0.29, + "grad_norm": 0.23811880045549916, + "learning_rate": 7.609756097560976e-05, + "loss": 1.2838, + "step": 156 + }, + { + "epoch": 0.29, + "grad_norm": 0.19511077122033713, + "learning_rate": 7.658536585365854e-05, + "loss": 1.1594, + "step": 157 + }, + { + "epoch": 0.29, + "grad_norm": 0.20094129399534238, + "learning_rate": 7.707317073170732e-05, + "loss": 1.2966, + "step": 158 + }, + { + "epoch": 0.29, + "grad_norm": 0.19366245038292418, + "learning_rate": 7.75609756097561e-05, + "loss": 1.2246, + "step": 159 + }, + { + "epoch": 0.29, + "grad_norm": 0.19409570223867306, + "learning_rate": 7.804878048780489e-05, + "loss": 1.2312, + "step": 160 + }, + { + "epoch": 0.29, + "grad_norm": 0.2087258457033805, + "learning_rate": 7.853658536585366e-05, + "loss": 1.2169, + "step": 161 + }, + { + "epoch": 0.3, + "grad_norm": 0.18765223996270428, + "learning_rate": 7.902439024390244e-05, + "loss": 1.2383, + "step": 162 + }, + { + "epoch": 0.3, + "grad_norm": 0.20734180224147242, + "learning_rate": 7.951219512195122e-05, + "loss": 1.2587, + "step": 163 + }, + { + "epoch": 0.3, + "grad_norm": 0.24690929540287834, + "learning_rate": 8e-05, + "loss": 1.1951, + "step": 164 + }, + { + "epoch": 0.3, + "grad_norm": 0.2003538797619543, + "learning_rate": 7.999990914797545e-05, + "loss": 1.1982, + "step": 165 + }, + { + "epoch": 0.3, + "grad_norm": 0.22469075613510484, + "learning_rate": 7.99996365923145e-05, + "loss": 1.2355, + "step": 166 + }, + { + "epoch": 0.31, + "grad_norm": 0.21870100788336058, + "learning_rate": 7.999918233425526e-05, + "loss": 1.1103, + "step": 167 + }, + { + "epoch": 0.31, + "grad_norm": 0.20939989594131886, + "learning_rate": 7.999854637586122e-05, + "loss": 1.1966, + "step": 168 + }, + { + "epoch": 0.31, + "grad_norm": 0.43108211416237796, + "learning_rate": 7.999772872002132e-05, + "loss": 1.2882, + "step": 169 + }, + { + "epoch": 0.31, + "grad_norm": 0.27045413432174487, + "learning_rate": 7.999672937044984e-05, + "loss": 1.2399, + "step": 170 + }, + { + "epoch": 0.31, + "grad_norm": 0.19700483036740515, + "learning_rate": 7.999554833168642e-05, + "loss": 1.202, + "step": 171 + }, + { + "epoch": 0.31, + "grad_norm": 0.3335979493370708, + "learning_rate": 7.999418560909604e-05, + "loss": 1.1995, + "step": 172 + }, + { + "epoch": 0.32, + "grad_norm": 0.3165803974474567, + "learning_rate": 7.999264120886902e-05, + "loss": 1.1569, + "step": 173 + }, + { + "epoch": 0.32, + "grad_norm": 0.1951699080346223, + "learning_rate": 7.999091513802093e-05, + "loss": 1.1778, + "step": 174 + }, + { + "epoch": 0.32, + "grad_norm": 0.2087559121749787, + "learning_rate": 7.998900740439265e-05, + "loss": 1.1736, + "step": 175 + }, + { + "epoch": 0.32, + "grad_norm": 0.20345180977460478, + "learning_rate": 7.998691801665024e-05, + "loss": 1.2281, + "step": 176 + }, + { + "epoch": 0.32, + "grad_norm": 0.24617644827252333, + "learning_rate": 7.998464698428495e-05, + "loss": 1.2072, + "step": 177 + }, + { + "epoch": 0.33, + "grad_norm": 0.2469050959356265, + "learning_rate": 7.998219431761318e-05, + "loss": 1.2242, + "step": 178 + }, + { + "epoch": 0.33, + "grad_norm": 0.19529317748460623, + "learning_rate": 7.997956002777642e-05, + "loss": 1.2567, + "step": 179 + }, + { + "epoch": 0.33, + "grad_norm": 0.19048389491381376, + "learning_rate": 7.99767441267412e-05, + "loss": 1.2982, + "step": 180 + }, + { + "epoch": 0.33, + "grad_norm": 0.2085799116493225, + "learning_rate": 7.997374662729904e-05, + "loss": 1.1254, + "step": 181 + }, + { + "epoch": 0.33, + "grad_norm": 0.20636853256378995, + "learning_rate": 7.997056754306636e-05, + "loss": 1.2435, + "step": 182 + }, + { + "epoch": 0.33, + "grad_norm": 0.20590016382290252, + "learning_rate": 7.99672068884845e-05, + "loss": 1.2658, + "step": 183 + }, + { + "epoch": 0.34, + "grad_norm": 0.1931166169764433, + "learning_rate": 7.996366467881955e-05, + "loss": 1.1637, + "step": 184 + }, + { + "epoch": 0.34, + "grad_norm": 0.18873318157988098, + "learning_rate": 7.995994093016237e-05, + "loss": 1.1335, + "step": 185 + }, + { + "epoch": 0.34, + "grad_norm": 0.19210254625199108, + "learning_rate": 7.995603565942846e-05, + "loss": 1.1928, + "step": 186 + }, + { + "epoch": 0.34, + "grad_norm": 0.2130986479765664, + "learning_rate": 7.995194888435792e-05, + "loss": 1.2158, + "step": 187 + }, + { + "epoch": 0.34, + "grad_norm": 0.22003854501814088, + "learning_rate": 7.994768062351532e-05, + "loss": 1.2288, + "step": 188 + }, + { + "epoch": 0.35, + "grad_norm": 0.20330803191993058, + "learning_rate": 7.994323089628968e-05, + "loss": 1.2426, + "step": 189 + }, + { + "epoch": 0.35, + "grad_norm": 0.20567314642208634, + "learning_rate": 7.993859972289434e-05, + "loss": 1.2649, + "step": 190 + }, + { + "epoch": 0.35, + "grad_norm": 0.21556663727342962, + "learning_rate": 7.993378712436686e-05, + "loss": 1.2545, + "step": 191 + }, + { + "epoch": 0.35, + "grad_norm": 0.20309165469109888, + "learning_rate": 7.992879312256897e-05, + "loss": 1.3338, + "step": 192 + }, + { + "epoch": 0.35, + "grad_norm": 0.19574356669421325, + "learning_rate": 7.992361774018641e-05, + "loss": 1.278, + "step": 193 + }, + { + "epoch": 0.35, + "grad_norm": 0.2763613746722313, + "learning_rate": 7.991826100072891e-05, + "loss": 1.2571, + "step": 194 + }, + { + "epoch": 0.36, + "grad_norm": 0.19346552479915102, + "learning_rate": 7.991272292852996e-05, + "loss": 1.2027, + "step": 195 + }, + { + "epoch": 0.36, + "grad_norm": 0.2281167812123908, + "learning_rate": 7.990700354874683e-05, + "loss": 1.2586, + "step": 196 + }, + { + "epoch": 0.36, + "grad_norm": 0.19699013712137542, + "learning_rate": 7.990110288736042e-05, + "loss": 1.1371, + "step": 197 + }, + { + "epoch": 0.36, + "grad_norm": 0.21768209981475933, + "learning_rate": 7.989502097117503e-05, + "loss": 1.2522, + "step": 198 + }, + { + "epoch": 0.36, + "grad_norm": 0.21335427847754582, + "learning_rate": 7.988875782781838e-05, + "loss": 1.2437, + "step": 199 + }, + { + "epoch": 0.37, + "grad_norm": 0.21856710629066897, + "learning_rate": 7.988231348574147e-05, + "loss": 1.2135, + "step": 200 + }, + { + "epoch": 0.37, + "grad_norm": 0.20482062658774797, + "learning_rate": 7.987568797421836e-05, + "loss": 1.1755, + "step": 201 + }, + { + "epoch": 0.37, + "grad_norm": 0.2017756813960897, + "learning_rate": 7.986888132334608e-05, + "loss": 1.1699, + "step": 202 + }, + { + "epoch": 0.37, + "grad_norm": 0.20496443848153809, + "learning_rate": 7.986189356404458e-05, + "loss": 1.2125, + "step": 203 + }, + { + "epoch": 0.37, + "grad_norm": 0.2134603800558358, + "learning_rate": 7.985472472805643e-05, + "loss": 1.2391, + "step": 204 + }, + { + "epoch": 0.37, + "grad_norm": 0.2364175573420861, + "learning_rate": 7.98473748479468e-05, + "loss": 1.2384, + "step": 205 + }, + { + "epoch": 0.38, + "grad_norm": 0.1872419861598724, + "learning_rate": 7.983984395710326e-05, + "loss": 1.1457, + "step": 206 + }, + { + "epoch": 0.38, + "grad_norm": 0.28222194007095774, + "learning_rate": 7.983213208973566e-05, + "loss": 1.2952, + "step": 207 + }, + { + "epoch": 0.38, + "grad_norm": 0.1916094851162064, + "learning_rate": 7.982423928087593e-05, + "loss": 1.1763, + "step": 208 + }, + { + "epoch": 0.38, + "grad_norm": 0.18446245256166657, + "learning_rate": 7.981616556637795e-05, + "loss": 1.1863, + "step": 209 + }, + { + "epoch": 0.38, + "grad_norm": 0.195191961022491, + "learning_rate": 7.980791098291737e-05, + "loss": 1.2036, + "step": 210 + }, + { + "epoch": 0.39, + "grad_norm": 0.2652439657825496, + "learning_rate": 7.979947556799151e-05, + "loss": 1.2834, + "step": 211 + }, + { + "epoch": 0.39, + "grad_norm": 0.24308438957843412, + "learning_rate": 7.979085935991906e-05, + "loss": 1.234, + "step": 212 + }, + { + "epoch": 0.39, + "grad_norm": 0.21294701043622016, + "learning_rate": 7.978206239784004e-05, + "loss": 1.3006, + "step": 213 + }, + { + "epoch": 0.39, + "grad_norm": 0.25809277041859524, + "learning_rate": 7.977308472171553e-05, + "loss": 1.2272, + "step": 214 + }, + { + "epoch": 0.39, + "grad_norm": 0.193463860107294, + "learning_rate": 7.976392637232754e-05, + "loss": 1.2295, + "step": 215 + }, + { + "epoch": 0.4, + "grad_norm": 0.2150023760609626, + "learning_rate": 7.975458739127877e-05, + "loss": 1.2135, + "step": 216 + }, + { + "epoch": 0.4, + "grad_norm": 0.22590495955605894, + "learning_rate": 7.974506782099253e-05, + "loss": 1.2532, + "step": 217 + }, + { + "epoch": 0.4, + "grad_norm": 0.21023744668403702, + "learning_rate": 7.973536770471242e-05, + "loss": 1.2472, + "step": 218 + }, + { + "epoch": 0.4, + "grad_norm": 0.2345749799511543, + "learning_rate": 7.972548708650218e-05, + "loss": 1.1791, + "step": 219 + }, + { + "epoch": 0.4, + "grad_norm": 0.2158876734005217, + "learning_rate": 7.971542601124553e-05, + "loss": 1.2483, + "step": 220 + }, + { + "epoch": 0.4, + "grad_norm": 0.29455339949432446, + "learning_rate": 7.970518452464593e-05, + "loss": 1.2894, + "step": 221 + }, + { + "epoch": 0.41, + "grad_norm": 0.23983708730626851, + "learning_rate": 7.969476267322636e-05, + "loss": 1.271, + "step": 222 + }, + { + "epoch": 0.41, + "grad_norm": 0.1922400905426158, + "learning_rate": 7.968416050432912e-05, + "loss": 1.2139, + "step": 223 + }, + { + "epoch": 0.41, + "grad_norm": 0.2238136844422931, + "learning_rate": 7.967337806611568e-05, + "loss": 1.2655, + "step": 224 + }, + { + "epoch": 0.41, + "grad_norm": 0.21230292828267672, + "learning_rate": 7.966241540756631e-05, + "loss": 1.2406, + "step": 225 + }, + { + "epoch": 0.41, + "grad_norm": 0.26656119419070456, + "learning_rate": 7.965127257848004e-05, + "loss": 1.2595, + "step": 226 + }, + { + "epoch": 0.42, + "grad_norm": 0.22381385502992684, + "learning_rate": 7.963994962947426e-05, + "loss": 1.1737, + "step": 227 + }, + { + "epoch": 0.42, + "grad_norm": 0.20056702203994298, + "learning_rate": 7.962844661198462e-05, + "loss": 1.1969, + "step": 228 + }, + { + "epoch": 0.42, + "grad_norm": 0.20148701321526885, + "learning_rate": 7.961676357826478e-05, + "loss": 1.2151, + "step": 229 + }, + { + "epoch": 0.42, + "grad_norm": 0.20034834807028637, + "learning_rate": 7.960490058138604e-05, + "loss": 1.1455, + "step": 230 + }, + { + "epoch": 0.42, + "grad_norm": 0.21050838521846033, + "learning_rate": 7.959285767523732e-05, + "loss": 1.2223, + "step": 231 + }, + { + "epoch": 0.42, + "grad_norm": 0.20904772138969777, + "learning_rate": 7.95806349145247e-05, + "loss": 1.2534, + "step": 232 + }, + { + "epoch": 0.43, + "grad_norm": 0.20307877304792957, + "learning_rate": 7.956823235477134e-05, + "loss": 1.1352, + "step": 233 + }, + { + "epoch": 0.43, + "grad_norm": 0.20501105270897094, + "learning_rate": 7.95556500523171e-05, + "loss": 1.2031, + "step": 234 + }, + { + "epoch": 0.43, + "grad_norm": 0.19800586972038586, + "learning_rate": 7.954288806431838e-05, + "loss": 1.2567, + "step": 235 + }, + { + "epoch": 0.43, + "grad_norm": 0.2175102450594135, + "learning_rate": 7.952994644874777e-05, + "loss": 1.2538, + "step": 236 + }, + { + "epoch": 0.43, + "grad_norm": 0.22698189300067595, + "learning_rate": 7.951682526439391e-05, + "loss": 1.3088, + "step": 237 + }, + { + "epoch": 0.44, + "grad_norm": 0.19208392014975315, + "learning_rate": 7.950352457086109e-05, + "loss": 1.2336, + "step": 238 + }, + { + "epoch": 0.44, + "grad_norm": 0.27004086334319655, + "learning_rate": 7.949004442856905e-05, + "loss": 1.2012, + "step": 239 + }, + { + "epoch": 0.44, + "grad_norm": 0.23420974954538043, + "learning_rate": 7.947638489875272e-05, + "loss": 1.2244, + "step": 240 + }, + { + "epoch": 0.44, + "grad_norm": 0.20514399124802024, + "learning_rate": 7.946254604346186e-05, + "loss": 1.2548, + "step": 241 + }, + { + "epoch": 0.44, + "grad_norm": 0.19334973602372896, + "learning_rate": 7.944852792556092e-05, + "loss": 1.2104, + "step": 242 + }, + { + "epoch": 0.44, + "grad_norm": 0.1992640714537956, + "learning_rate": 7.943433060872858e-05, + "loss": 1.2628, + "step": 243 + }, + { + "epoch": 0.45, + "grad_norm": 0.203284617090413, + "learning_rate": 7.941995415745761e-05, + "loss": 1.2002, + "step": 244 + }, + { + "epoch": 0.45, + "grad_norm": 0.22795306969682058, + "learning_rate": 7.94053986370545e-05, + "loss": 1.2215, + "step": 245 + }, + { + "epoch": 0.45, + "grad_norm": 0.20789041346838505, + "learning_rate": 7.939066411363915e-05, + "loss": 1.0998, + "step": 246 + }, + { + "epoch": 0.45, + "grad_norm": 0.22354868884742066, + "learning_rate": 7.937575065414464e-05, + "loss": 1.2564, + "step": 247 + }, + { + "epoch": 0.45, + "grad_norm": 0.21176392726647736, + "learning_rate": 7.936065832631687e-05, + "loss": 1.2816, + "step": 248 + }, + { + "epoch": 0.46, + "grad_norm": 0.19967179557235587, + "learning_rate": 7.934538719871427e-05, + "loss": 1.1961, + "step": 249 + }, + { + "epoch": 0.46, + "grad_norm": 0.210819577350627, + "learning_rate": 7.932993734070747e-05, + "loss": 1.2167, + "step": 250 + }, + { + "epoch": 0.46, + "grad_norm": 0.21537794551756187, + "learning_rate": 7.931430882247903e-05, + "loss": 1.2341, + "step": 251 + }, + { + "epoch": 0.46, + "grad_norm": 0.22850872387256574, + "learning_rate": 7.929850171502304e-05, + "loss": 1.1686, + "step": 252 + }, + { + "epoch": 0.46, + "grad_norm": 0.22380366415076383, + "learning_rate": 7.928251609014493e-05, + "loss": 1.1462, + "step": 253 + }, + { + "epoch": 0.46, + "grad_norm": 0.22426923149036065, + "learning_rate": 7.926635202046102e-05, + "loss": 1.1792, + "step": 254 + }, + { + "epoch": 0.47, + "grad_norm": 0.42082703321103965, + "learning_rate": 7.925000957939822e-05, + "loss": 1.2718, + "step": 255 + }, + { + "epoch": 0.47, + "grad_norm": 0.2235432774854074, + "learning_rate": 7.92334888411937e-05, + "loss": 1.2598, + "step": 256 + }, + { + "epoch": 0.47, + "grad_norm": 0.281644028934108, + "learning_rate": 7.92167898808946e-05, + "loss": 1.2205, + "step": 257 + }, + { + "epoch": 0.47, + "grad_norm": 0.2037705143888748, + "learning_rate": 7.919991277435763e-05, + "loss": 1.1737, + "step": 258 + }, + { + "epoch": 0.47, + "grad_norm": 0.20917419230028977, + "learning_rate": 7.918285759824879e-05, + "loss": 1.2035, + "step": 259 + }, + { + "epoch": 0.48, + "grad_norm": 0.20510847570635518, + "learning_rate": 7.916562443004292e-05, + "loss": 1.2135, + "step": 260 + }, + { + "epoch": 0.48, + "grad_norm": 0.25172483071092466, + "learning_rate": 7.914821334802342e-05, + "loss": 1.2218, + "step": 261 + }, + { + "epoch": 0.48, + "grad_norm": 0.21102706700634313, + "learning_rate": 7.91306244312819e-05, + "loss": 1.1738, + "step": 262 + }, + { + "epoch": 0.48, + "grad_norm": 0.22626060872645815, + "learning_rate": 7.911285775971781e-05, + "loss": 1.238, + "step": 263 + }, + { + "epoch": 0.48, + "grad_norm": 0.22448567539778486, + "learning_rate": 7.909491341403805e-05, + "loss": 1.2404, + "step": 264 + }, + { + "epoch": 0.48, + "grad_norm": 0.2019099786139193, + "learning_rate": 7.907679147575661e-05, + "loss": 1.213, + "step": 265 + }, + { + "epoch": 0.49, + "grad_norm": 0.24307234839096267, + "learning_rate": 7.905849202719422e-05, + "loss": 1.2322, + "step": 266 + }, + { + "epoch": 0.49, + "grad_norm": 0.19801890521743487, + "learning_rate": 7.904001515147802e-05, + "loss": 1.2448, + "step": 267 + }, + { + "epoch": 0.49, + "grad_norm": 0.2102742273575385, + "learning_rate": 7.902136093254106e-05, + "loss": 1.1657, + "step": 268 + }, + { + "epoch": 0.49, + "grad_norm": 0.2173464476815016, + "learning_rate": 7.900252945512201e-05, + "loss": 1.2549, + "step": 269 + }, + { + "epoch": 0.49, + "grad_norm": 0.20957275458699595, + "learning_rate": 7.898352080476479e-05, + "loss": 1.2536, + "step": 270 + }, + { + "epoch": 0.5, + "grad_norm": 0.20691966388952363, + "learning_rate": 7.896433506781811e-05, + "loss": 1.2661, + "step": 271 + }, + { + "epoch": 0.5, + "grad_norm": 0.2276662275112648, + "learning_rate": 7.894497233143509e-05, + "loss": 1.2409, + "step": 272 + }, + { + "epoch": 0.5, + "grad_norm": 0.23854109569301263, + "learning_rate": 7.892543268357297e-05, + "loss": 1.2681, + "step": 273 + }, + { + "epoch": 0.5, + "grad_norm": 0.2233864156677627, + "learning_rate": 7.890571621299252e-05, + "loss": 1.1687, + "step": 274 + }, + { + "epoch": 0.5, + "grad_norm": 0.20114129147925475, + "learning_rate": 7.888582300925787e-05, + "loss": 1.2184, + "step": 275 + }, + { + "epoch": 0.5, + "grad_norm": 0.2154654670569462, + "learning_rate": 7.886575316273586e-05, + "loss": 1.1982, + "step": 276 + }, + { + "epoch": 0.51, + "grad_norm": 0.2292982209343639, + "learning_rate": 7.884550676459583e-05, + "loss": 1.2129, + "step": 277 + }, + { + "epoch": 0.51, + "grad_norm": 0.21302713135229548, + "learning_rate": 7.882508390680908e-05, + "loss": 1.1605, + "step": 278 + }, + { + "epoch": 0.51, + "grad_norm": 0.2123661020671048, + "learning_rate": 7.88044846821485e-05, + "loss": 1.2308, + "step": 279 + }, + { + "epoch": 0.51, + "grad_norm": 0.2080577410800404, + "learning_rate": 7.878370918418818e-05, + "loss": 1.2195, + "step": 280 + }, + { + "epoch": 0.51, + "grad_norm": 0.19663901881127385, + "learning_rate": 7.876275750730289e-05, + "loss": 1.1591, + "step": 281 + }, + { + "epoch": 0.52, + "grad_norm": 0.20534502031312163, + "learning_rate": 7.874162974666776e-05, + "loss": 1.2664, + "step": 282 + }, + { + "epoch": 0.52, + "grad_norm": 0.23240445399513837, + "learning_rate": 7.872032599825779e-05, + "loss": 1.2151, + "step": 283 + }, + { + "epoch": 0.52, + "grad_norm": 0.2672527316717507, + "learning_rate": 7.86988463588474e-05, + "loss": 1.2406, + "step": 284 + }, + { + "epoch": 0.52, + "grad_norm": 0.19893903058743695, + "learning_rate": 7.867719092601003e-05, + "loss": 1.1291, + "step": 285 + }, + { + "epoch": 0.52, + "grad_norm": 0.33275268109930917, + "learning_rate": 7.865535979811768e-05, + "loss": 1.1406, + "step": 286 + }, + { + "epoch": 0.52, + "grad_norm": 0.2373619455690358, + "learning_rate": 7.863335307434045e-05, + "loss": 1.2799, + "step": 287 + }, + { + "epoch": 0.53, + "grad_norm": 0.263235735390858, + "learning_rate": 7.861117085464612e-05, + "loss": 1.2415, + "step": 288 + }, + { + "epoch": 0.53, + "grad_norm": 0.25884281780784324, + "learning_rate": 7.858881323979965e-05, + "loss": 1.3919, + "step": 289 + }, + { + "epoch": 0.53, + "grad_norm": 0.25426288332255736, + "learning_rate": 7.85662803313628e-05, + "loss": 1.174, + "step": 290 + }, + { + "epoch": 0.53, + "grad_norm": 0.26655405527881243, + "learning_rate": 7.854357223169356e-05, + "loss": 1.2806, + "step": 291 + }, + { + "epoch": 0.53, + "grad_norm": 0.20909844432349833, + "learning_rate": 7.852068904394579e-05, + "loss": 1.2627, + "step": 292 + }, + { + "epoch": 0.54, + "grad_norm": 0.21307115068935759, + "learning_rate": 7.849763087206866e-05, + "loss": 1.1879, + "step": 293 + }, + { + "epoch": 0.54, + "grad_norm": 0.25009949471398946, + "learning_rate": 7.847439782080628e-05, + "loss": 1.2881, + "step": 294 + }, + { + "epoch": 0.54, + "grad_norm": 0.20960783418679174, + "learning_rate": 7.845098999569712e-05, + "loss": 1.2723, + "step": 295 + }, + { + "epoch": 0.54, + "grad_norm": 0.24968832437925104, + "learning_rate": 7.842740750307362e-05, + "loss": 1.2029, + "step": 296 + }, + { + "epoch": 0.54, + "grad_norm": 0.22981196585125677, + "learning_rate": 7.84036504500616e-05, + "loss": 1.1695, + "step": 297 + }, + { + "epoch": 0.55, + "grad_norm": 0.2320606844751365, + "learning_rate": 7.837971894457991e-05, + "loss": 1.2317, + "step": 298 + }, + { + "epoch": 0.55, + "grad_norm": 0.23051459673906124, + "learning_rate": 7.835561309533981e-05, + "loss": 1.2046, + "step": 299 + }, + { + "epoch": 0.55, + "grad_norm": 0.2510027231060586, + "learning_rate": 7.833133301184457e-05, + "loss": 1.199, + "step": 300 + }, + { + "epoch": 0.55, + "grad_norm": 0.23601180466018787, + "learning_rate": 7.830687880438895e-05, + "loss": 1.1755, + "step": 301 + }, + { + "epoch": 0.55, + "grad_norm": 0.24740820934385369, + "learning_rate": 7.828225058405864e-05, + "loss": 1.2054, + "step": 302 + }, + { + "epoch": 0.55, + "grad_norm": 0.23065372979111173, + "learning_rate": 7.825744846272984e-05, + "loss": 1.2066, + "step": 303 + }, + { + "epoch": 0.56, + "grad_norm": 0.22385077334838213, + "learning_rate": 7.823247255306866e-05, + "loss": 1.2147, + "step": 304 + }, + { + "epoch": 0.56, + "grad_norm": 0.42981213948386104, + "learning_rate": 7.820732296853074e-05, + "loss": 1.2314, + "step": 305 + }, + { + "epoch": 0.56, + "grad_norm": 0.21122844902751076, + "learning_rate": 7.818199982336058e-05, + "loss": 1.1462, + "step": 306 + }, + { + "epoch": 0.56, + "grad_norm": 0.23374869692118933, + "learning_rate": 7.815650323259117e-05, + "loss": 1.2051, + "step": 307 + }, + { + "epoch": 0.56, + "grad_norm": 0.21662363795962128, + "learning_rate": 7.813083331204332e-05, + "loss": 1.1575, + "step": 308 + }, + { + "epoch": 0.57, + "grad_norm": 0.2088315773384112, + "learning_rate": 7.810499017832526e-05, + "loss": 1.1316, + "step": 309 + }, + { + "epoch": 0.57, + "grad_norm": 0.2095238410730976, + "learning_rate": 7.807897394883203e-05, + "loss": 1.2087, + "step": 310 + }, + { + "epoch": 0.57, + "grad_norm": 0.22672932127256515, + "learning_rate": 7.805278474174499e-05, + "loss": 1.2512, + "step": 311 + }, + { + "epoch": 0.57, + "grad_norm": 0.21873052340922736, + "learning_rate": 7.802642267603126e-05, + "loss": 1.1909, + "step": 312 + }, + { + "epoch": 0.57, + "grad_norm": 0.219814521916342, + "learning_rate": 7.79998878714432e-05, + "loss": 1.1669, + "step": 313 + }, + { + "epoch": 0.57, + "grad_norm": 0.3049426027257317, + "learning_rate": 7.797318044851786e-05, + "loss": 1.1797, + "step": 314 + }, + { + "epoch": 0.58, + "grad_norm": 0.22309435690065985, + "learning_rate": 7.794630052857638e-05, + "loss": 1.1417, + "step": 315 + }, + { + "epoch": 0.58, + "grad_norm": 0.3891885169154885, + "learning_rate": 7.791924823372354e-05, + "loss": 1.2369, + "step": 316 + }, + { + "epoch": 0.58, + "grad_norm": 0.24780269452456372, + "learning_rate": 7.789202368684711e-05, + "loss": 1.2521, + "step": 317 + }, + { + "epoch": 0.58, + "grad_norm": 0.21660460720269362, + "learning_rate": 7.786462701161738e-05, + "loss": 1.2151, + "step": 318 + }, + { + "epoch": 0.58, + "grad_norm": 0.23635409466561857, + "learning_rate": 7.783705833248649e-05, + "loss": 1.2363, + "step": 319 + }, + { + "epoch": 0.59, + "grad_norm": 0.2616135839903218, + "learning_rate": 7.780931777468797e-05, + "loss": 1.2428, + "step": 320 + }, + { + "epoch": 0.59, + "grad_norm": 0.21461059159245083, + "learning_rate": 7.77814054642361e-05, + "loss": 1.1434, + "step": 321 + }, + { + "epoch": 0.59, + "grad_norm": 0.25348824286656163, + "learning_rate": 7.775332152792539e-05, + "loss": 1.2368, + "step": 322 + }, + { + "epoch": 0.59, + "grad_norm": 0.22275034726331247, + "learning_rate": 7.772506609332995e-05, + "loss": 1.1827, + "step": 323 + }, + { + "epoch": 0.59, + "grad_norm": 0.25030821228147526, + "learning_rate": 7.769663928880298e-05, + "loss": 1.2428, + "step": 324 + }, + { + "epoch": 0.59, + "grad_norm": 0.22251804398745534, + "learning_rate": 7.766804124347608e-05, + "loss": 1.1889, + "step": 325 + }, + { + "epoch": 0.6, + "grad_norm": 0.23381455520411995, + "learning_rate": 7.763927208725879e-05, + "loss": 1.2115, + "step": 326 + }, + { + "epoch": 0.6, + "grad_norm": 0.27341902651946226, + "learning_rate": 7.761033195083791e-05, + "loss": 1.2535, + "step": 327 + }, + { + "epoch": 0.6, + "grad_norm": 0.24862471659814522, + "learning_rate": 7.758122096567694e-05, + "loss": 1.2128, + "step": 328 + }, + { + "epoch": 0.6, + "grad_norm": 0.2251357082045494, + "learning_rate": 7.755193926401547e-05, + "loss": 1.2334, + "step": 329 + }, + { + "epoch": 0.6, + "grad_norm": 0.3173274941622932, + "learning_rate": 7.752248697886857e-05, + "loss": 1.226, + "step": 330 + }, + { + "epoch": 0.61, + "grad_norm": 0.23056440717672175, + "learning_rate": 7.74928642440263e-05, + "loss": 1.2339, + "step": 331 + }, + { + "epoch": 0.61, + "grad_norm": 0.2801507500859342, + "learning_rate": 7.746307119405286e-05, + "loss": 1.287, + "step": 332 + }, + { + "epoch": 0.61, + "grad_norm": 0.2267818430426272, + "learning_rate": 7.743310796428622e-05, + "loss": 1.1916, + "step": 333 + }, + { + "epoch": 0.61, + "grad_norm": 0.2777329160365585, + "learning_rate": 7.74029746908374e-05, + "loss": 1.252, + "step": 334 + }, + { + "epoch": 0.61, + "grad_norm": 0.25289169762353, + "learning_rate": 7.737267151058983e-05, + "loss": 1.2153, + "step": 335 + }, + { + "epoch": 0.61, + "grad_norm": 0.2424670686901653, + "learning_rate": 7.734219856119875e-05, + "loss": 1.2227, + "step": 336 + }, + { + "epoch": 0.62, + "grad_norm": 0.22747092217441645, + "learning_rate": 7.731155598109067e-05, + "loss": 1.19, + "step": 337 + }, + { + "epoch": 0.62, + "grad_norm": 0.2307810940100189, + "learning_rate": 7.728074390946257e-05, + "loss": 1.1818, + "step": 338 + }, + { + "epoch": 0.62, + "grad_norm": 0.2583402574655623, + "learning_rate": 7.724976248628142e-05, + "loss": 1.1608, + "step": 339 + }, + { + "epoch": 0.62, + "grad_norm": 0.22140209760890694, + "learning_rate": 7.721861185228347e-05, + "loss": 1.1245, + "step": 340 + }, + { + "epoch": 0.62, + "grad_norm": 0.25859310758244686, + "learning_rate": 7.718729214897362e-05, + "loss": 1.2247, + "step": 341 + }, + { + "epoch": 0.63, + "grad_norm": 0.26371179531372124, + "learning_rate": 7.715580351862482e-05, + "loss": 1.2128, + "step": 342 + }, + { + "epoch": 0.63, + "grad_norm": 0.26575541302851047, + "learning_rate": 7.712414610427733e-05, + "loss": 1.2443, + "step": 343 + }, + { + "epoch": 0.63, + "grad_norm": 0.269978305197599, + "learning_rate": 7.709232004973816e-05, + "loss": 1.2231, + "step": 344 + }, + { + "epoch": 0.63, + "grad_norm": 0.26583998705977047, + "learning_rate": 7.70603254995804e-05, + "loss": 1.2476, + "step": 345 + }, + { + "epoch": 0.63, + "grad_norm": 0.24256062164066097, + "learning_rate": 7.702816259914253e-05, + "loss": 1.2901, + "step": 346 + }, + { + "epoch": 0.63, + "grad_norm": 0.3463123472658915, + "learning_rate": 7.699583149452779e-05, + "loss": 1.3277, + "step": 347 + }, + { + "epoch": 0.64, + "grad_norm": 0.2269096590531878, + "learning_rate": 7.696333233260345e-05, + "loss": 1.2047, + "step": 348 + }, + { + "epoch": 0.64, + "grad_norm": 0.25136883001050025, + "learning_rate": 7.693066526100031e-05, + "loss": 1.1619, + "step": 349 + }, + { + "epoch": 0.64, + "grad_norm": 0.2565112571116145, + "learning_rate": 7.68978304281118e-05, + "loss": 1.2389, + "step": 350 + }, + { + "epoch": 0.64, + "grad_norm": 0.22175779550828703, + "learning_rate": 7.686482798309349e-05, + "loss": 1.2238, + "step": 351 + }, + { + "epoch": 0.64, + "grad_norm": 0.22588304332216555, + "learning_rate": 7.683165807586234e-05, + "loss": 1.174, + "step": 352 + }, + { + "epoch": 0.65, + "grad_norm": 0.24889474296529737, + "learning_rate": 7.6798320857096e-05, + "loss": 1.2366, + "step": 353 + }, + { + "epoch": 0.65, + "grad_norm": 0.27339703806525034, + "learning_rate": 7.676481647823214e-05, + "loss": 1.2356, + "step": 354 + }, + { + "epoch": 0.65, + "grad_norm": 0.23424666722888365, + "learning_rate": 7.673114509146782e-05, + "loss": 1.2089, + "step": 355 + }, + { + "epoch": 0.65, + "grad_norm": 0.27978285392461766, + "learning_rate": 7.66973068497587e-05, + "loss": 1.2609, + "step": 356 + }, + { + "epoch": 0.65, + "grad_norm": 0.2509423350138824, + "learning_rate": 7.666330190681844e-05, + "loss": 1.1777, + "step": 357 + }, + { + "epoch": 0.65, + "grad_norm": 0.23007730927468031, + "learning_rate": 7.662913041711793e-05, + "loss": 1.154, + "step": 358 + }, + { + "epoch": 0.66, + "grad_norm": 0.2438648674953112, + "learning_rate": 7.659479253588462e-05, + "loss": 1.2257, + "step": 359 + }, + { + "epoch": 0.66, + "grad_norm": 0.28816093242092233, + "learning_rate": 7.65602884191018e-05, + "loss": 1.2558, + "step": 360 + }, + { + "epoch": 0.66, + "grad_norm": 0.24972815300596035, + "learning_rate": 7.652561822350793e-05, + "loss": 1.2837, + "step": 361 + }, + { + "epoch": 0.66, + "grad_norm": 0.2543189139697063, + "learning_rate": 7.649078210659587e-05, + "loss": 1.2193, + "step": 362 + }, + { + "epoch": 0.66, + "grad_norm": 0.2237937956718952, + "learning_rate": 7.645578022661224e-05, + "loss": 1.2237, + "step": 363 + }, + { + "epoch": 0.67, + "grad_norm": 0.29742029408787396, + "learning_rate": 7.642061274255657e-05, + "loss": 1.2116, + "step": 364 + }, + { + "epoch": 0.67, + "grad_norm": 0.2462883147335493, + "learning_rate": 7.638527981418075e-05, + "loss": 1.1827, + "step": 365 + }, + { + "epoch": 0.67, + "grad_norm": 0.2647802498907096, + "learning_rate": 7.634978160198817e-05, + "loss": 1.2739, + "step": 366 + }, + { + "epoch": 0.67, + "grad_norm": 0.22360398779217264, + "learning_rate": 7.631411826723306e-05, + "loss": 1.2185, + "step": 367 + }, + { + "epoch": 0.67, + "grad_norm": 0.2635048004593543, + "learning_rate": 7.627828997191973e-05, + "loss": 1.2317, + "step": 368 + }, + { + "epoch": 0.67, + "grad_norm": 0.2764803449917684, + "learning_rate": 7.624229687880184e-05, + "loss": 1.1923, + "step": 369 + }, + { + "epoch": 0.68, + "grad_norm": 0.25724943233414527, + "learning_rate": 7.620613915138166e-05, + "loss": 1.2218, + "step": 370 + }, + { + "epoch": 0.68, + "grad_norm": 0.2858318045794755, + "learning_rate": 7.61698169539093e-05, + "loss": 1.1496, + "step": 371 + }, + { + "epoch": 0.68, + "grad_norm": 0.23547216647460364, + "learning_rate": 7.613333045138206e-05, + "loss": 1.1905, + "step": 372 + }, + { + "epoch": 0.68, + "grad_norm": 0.22984814903684375, + "learning_rate": 7.609667980954355e-05, + "loss": 1.2009, + "step": 373 + }, + { + "epoch": 0.68, + "grad_norm": 0.2551903754079084, + "learning_rate": 7.605986519488301e-05, + "loss": 1.2042, + "step": 374 + }, + { + "epoch": 0.69, + "grad_norm": 0.2508257410125616, + "learning_rate": 7.602288677463457e-05, + "loss": 1.2468, + "step": 375 + }, + { + "epoch": 0.69, + "grad_norm": 0.25324577774935964, + "learning_rate": 7.598574471677644e-05, + "loss": 1.2603, + "step": 376 + }, + { + "epoch": 0.69, + "grad_norm": 0.35888776531769967, + "learning_rate": 7.59484391900302e-05, + "loss": 1.1929, + "step": 377 + }, + { + "epoch": 0.69, + "grad_norm": 0.22048517191014724, + "learning_rate": 7.591097036385994e-05, + "loss": 1.1783, + "step": 378 + }, + { + "epoch": 0.69, + "grad_norm": 0.2781160412746083, + "learning_rate": 7.587333840847162e-05, + "loss": 1.3397, + "step": 379 + }, + { + "epoch": 0.7, + "grad_norm": 0.24033046830332258, + "learning_rate": 7.583554349481222e-05, + "loss": 1.2436, + "step": 380 + }, + { + "epoch": 0.7, + "grad_norm": 0.26413762380260003, + "learning_rate": 7.579758579456893e-05, + "loss": 1.1917, + "step": 381 + }, + { + "epoch": 0.7, + "grad_norm": 0.2390937887338632, + "learning_rate": 7.575946548016847e-05, + "loss": 1.2186, + "step": 382 + }, + { + "epoch": 0.7, + "grad_norm": 0.25131263043429275, + "learning_rate": 7.572118272477622e-05, + "loss": 1.2538, + "step": 383 + }, + { + "epoch": 0.7, + "grad_norm": 0.223974104870702, + "learning_rate": 7.568273770229546e-05, + "loss": 1.2165, + "step": 384 + }, + { + "epoch": 0.7, + "grad_norm": 0.25840356830252875, + "learning_rate": 7.564413058736663e-05, + "loss": 1.1848, + "step": 385 + }, + { + "epoch": 0.71, + "grad_norm": 0.2723156683076603, + "learning_rate": 7.560536155536641e-05, + "loss": 1.1982, + "step": 386 + }, + { + "epoch": 0.71, + "grad_norm": 0.265687427976889, + "learning_rate": 7.556643078240708e-05, + "loss": 1.231, + "step": 387 + }, + { + "epoch": 0.71, + "grad_norm": 0.25152762080976077, + "learning_rate": 7.552733844533562e-05, + "loss": 1.1974, + "step": 388 + }, + { + "epoch": 0.71, + "grad_norm": 0.2366049485053541, + "learning_rate": 7.548808472173292e-05, + "loss": 1.3119, + "step": 389 + }, + { + "epoch": 0.71, + "grad_norm": 0.22092196577077122, + "learning_rate": 7.5448669789913e-05, + "loss": 1.195, + "step": 390 + }, + { + "epoch": 0.72, + "grad_norm": 0.22667521540462374, + "learning_rate": 7.540909382892217e-05, + "loss": 1.1431, + "step": 391 + }, + { + "epoch": 0.72, + "grad_norm": 0.25432207282646513, + "learning_rate": 7.536935701853823e-05, + "loss": 1.2173, + "step": 392 + }, + { + "epoch": 0.72, + "grad_norm": 0.29950506457923864, + "learning_rate": 7.53294595392697e-05, + "loss": 1.1962, + "step": 393 + }, + { + "epoch": 0.72, + "grad_norm": 0.24735689607229913, + "learning_rate": 7.528940157235487e-05, + "loss": 1.2053, + "step": 394 + }, + { + "epoch": 0.72, + "grad_norm": 0.24394198607459663, + "learning_rate": 7.524918329976114e-05, + "loss": 1.1979, + "step": 395 + }, + { + "epoch": 0.72, + "grad_norm": 0.2630369372689188, + "learning_rate": 7.520880490418409e-05, + "loss": 1.2111, + "step": 396 + }, + { + "epoch": 0.73, + "grad_norm": 0.26275028416291457, + "learning_rate": 7.516826656904664e-05, + "loss": 1.2133, + "step": 397 + }, + { + "epoch": 0.73, + "grad_norm": 0.23938074620956928, + "learning_rate": 7.512756847849831e-05, + "loss": 1.1355, + "step": 398 + }, + { + "epoch": 0.73, + "grad_norm": 0.3724960610098138, + "learning_rate": 7.508671081741428e-05, + "loss": 1.2572, + "step": 399 + }, + { + "epoch": 0.73, + "grad_norm": 0.24161685847894723, + "learning_rate": 7.504569377139462e-05, + "loss": 1.1706, + "step": 400 + }, + { + "epoch": 0.73, + "grad_norm": 0.26121591322670523, + "learning_rate": 7.50045175267634e-05, + "loss": 1.2135, + "step": 401 + }, + { + "epoch": 0.74, + "grad_norm": 0.2465579498164775, + "learning_rate": 7.496318227056788e-05, + "loss": 1.1641, + "step": 402 + }, + { + "epoch": 0.74, + "grad_norm": 0.2556288696122787, + "learning_rate": 7.492168819057767e-05, + "loss": 1.2939, + "step": 403 + }, + { + "epoch": 0.74, + "grad_norm": 0.261481216336303, + "learning_rate": 7.488003547528382e-05, + "loss": 1.2026, + "step": 404 + }, + { + "epoch": 0.74, + "grad_norm": 0.2389415135676362, + "learning_rate": 7.483822431389799e-05, + "loss": 1.2131, + "step": 405 + }, + { + "epoch": 0.74, + "grad_norm": 0.2559201956627192, + "learning_rate": 7.479625489635162e-05, + "loss": 1.1246, + "step": 406 + }, + { + "epoch": 0.74, + "grad_norm": 0.27127932491822604, + "learning_rate": 7.475412741329504e-05, + "loss": 1.2429, + "step": 407 + }, + { + "epoch": 0.75, + "grad_norm": 0.27006004008695594, + "learning_rate": 7.47118420560966e-05, + "loss": 1.2388, + "step": 408 + }, + { + "epoch": 0.75, + "grad_norm": 0.23716823297200537, + "learning_rate": 7.466939901684182e-05, + "loss": 1.1264, + "step": 409 + }, + { + "epoch": 0.75, + "grad_norm": 0.2885373898669248, + "learning_rate": 7.462679848833252e-05, + "loss": 1.2786, + "step": 410 + }, + { + "epoch": 0.75, + "grad_norm": 0.49215227598639927, + "learning_rate": 7.458404066408588e-05, + "loss": 1.2386, + "step": 411 + }, + { + "epoch": 0.75, + "grad_norm": 0.24235735604947403, + "learning_rate": 7.454112573833368e-05, + "loss": 1.1423, + "step": 412 + }, + { + "epoch": 0.76, + "grad_norm": 0.2584614748054343, + "learning_rate": 7.449805390602127e-05, + "loss": 1.2669, + "step": 413 + }, + { + "epoch": 0.76, + "grad_norm": 0.23806123085998873, + "learning_rate": 7.445482536280684e-05, + "loss": 1.1763, + "step": 414 + }, + { + "epoch": 0.76, + "grad_norm": 0.24459517607786851, + "learning_rate": 7.441144030506043e-05, + "loss": 1.198, + "step": 415 + }, + { + "epoch": 0.76, + "grad_norm": 0.25801616402700395, + "learning_rate": 7.436789892986304e-05, + "loss": 1.2136, + "step": 416 + }, + { + "epoch": 0.76, + "grad_norm": 0.2814819942392514, + "learning_rate": 7.432420143500578e-05, + "loss": 1.2398, + "step": 417 + }, + { + "epoch": 0.76, + "grad_norm": 0.22134709322606153, + "learning_rate": 7.428034801898893e-05, + "loss": 1.1592, + "step": 418 + }, + { + "epoch": 0.77, + "grad_norm": 0.2899677536995633, + "learning_rate": 7.42363388810211e-05, + "loss": 1.2296, + "step": 419 + }, + { + "epoch": 0.77, + "grad_norm": 0.24005943230262294, + "learning_rate": 7.419217422101822e-05, + "loss": 1.2223, + "step": 420 + }, + { + "epoch": 0.77, + "grad_norm": 0.26417562369496167, + "learning_rate": 7.414785423960275e-05, + "loss": 1.2261, + "step": 421 + }, + { + "epoch": 0.77, + "grad_norm": 0.2580815883535521, + "learning_rate": 7.410337913810271e-05, + "loss": 1.2021, + "step": 422 + }, + { + "epoch": 0.77, + "grad_norm": 0.25242217589496435, + "learning_rate": 7.405874911855071e-05, + "loss": 1.239, + "step": 423 + }, + { + "epoch": 0.78, + "grad_norm": 0.21991733999839932, + "learning_rate": 7.401396438368315e-05, + "loss": 1.1716, + "step": 424 + }, + { + "epoch": 0.78, + "grad_norm": 0.40116538322720213, + "learning_rate": 7.396902513693924e-05, + "loss": 1.2773, + "step": 425 + }, + { + "epoch": 0.78, + "grad_norm": 0.277333939455099, + "learning_rate": 7.392393158246002e-05, + "loss": 1.2574, + "step": 426 + }, + { + "epoch": 0.78, + "grad_norm": 0.27146087746385755, + "learning_rate": 7.387868392508756e-05, + "loss": 1.2243, + "step": 427 + }, + { + "epoch": 0.78, + "grad_norm": 0.255881055620786, + "learning_rate": 7.38332823703639e-05, + "loss": 1.223, + "step": 428 + }, + { + "epoch": 0.78, + "grad_norm": 0.24807364856677255, + "learning_rate": 7.378772712453021e-05, + "loss": 1.1985, + "step": 429 + }, + { + "epoch": 0.79, + "grad_norm": 0.25746257617764423, + "learning_rate": 7.37420183945258e-05, + "loss": 1.2502, + "step": 430 + }, + { + "epoch": 0.79, + "grad_norm": 0.28851991049982234, + "learning_rate": 7.369615638798722e-05, + "loss": 1.2535, + "step": 431 + }, + { + "epoch": 0.79, + "grad_norm": 0.24113389811604363, + "learning_rate": 7.365014131324725e-05, + "loss": 1.2227, + "step": 432 + }, + { + "epoch": 0.79, + "grad_norm": 0.2414465151257969, + "learning_rate": 7.360397337933405e-05, + "loss": 1.1884, + "step": 433 + }, + { + "epoch": 0.79, + "grad_norm": 0.2735463134699831, + "learning_rate": 7.355765279597011e-05, + "loss": 1.2756, + "step": 434 + }, + { + "epoch": 0.8, + "grad_norm": 0.2588437452987293, + "learning_rate": 7.351117977357139e-05, + "loss": 1.2108, + "step": 435 + }, + { + "epoch": 0.8, + "grad_norm": 0.26573294117796553, + "learning_rate": 7.346455452324629e-05, + "loss": 1.1821, + "step": 436 + }, + { + "epoch": 0.8, + "grad_norm": 0.2555476577827304, + "learning_rate": 7.341777725679473e-05, + "loss": 1.1937, + "step": 437 + }, + { + "epoch": 0.8, + "grad_norm": 0.2867704132108098, + "learning_rate": 7.337084818670716e-05, + "loss": 1.2272, + "step": 438 + }, + { + "epoch": 0.8, + "grad_norm": 0.27726678115981157, + "learning_rate": 7.332376752616367e-05, + "loss": 1.2331, + "step": 439 + }, + { + "epoch": 0.8, + "grad_norm": 0.26955338021079955, + "learning_rate": 7.32765354890329e-05, + "loss": 1.1731, + "step": 440 + }, + { + "epoch": 0.81, + "grad_norm": 0.25250321202536524, + "learning_rate": 7.322915228987116e-05, + "loss": 1.2653, + "step": 441 + }, + { + "epoch": 0.81, + "grad_norm": 0.24748844179765395, + "learning_rate": 7.318161814392143e-05, + "loss": 1.24, + "step": 442 + }, + { + "epoch": 0.81, + "grad_norm": 0.28177805247356325, + "learning_rate": 7.313393326711239e-05, + "loss": 1.185, + "step": 443 + }, + { + "epoch": 0.81, + "grad_norm": 0.24093242000396312, + "learning_rate": 7.30860978760574e-05, + "loss": 1.1994, + "step": 444 + }, + { + "epoch": 0.81, + "grad_norm": 0.26277803901457075, + "learning_rate": 7.30381121880536e-05, + "loss": 1.212, + "step": 445 + }, + { + "epoch": 0.82, + "grad_norm": 0.2506524258682433, + "learning_rate": 7.298997642108079e-05, + "loss": 1.2421, + "step": 446 + }, + { + "epoch": 0.82, + "grad_norm": 0.2840599700015824, + "learning_rate": 7.294169079380061e-05, + "loss": 1.1818, + "step": 447 + }, + { + "epoch": 0.82, + "grad_norm": 0.24892184038117549, + "learning_rate": 7.289325552555538e-05, + "loss": 1.1916, + "step": 448 + }, + { + "epoch": 0.82, + "grad_norm": 0.2700898428541357, + "learning_rate": 7.284467083636722e-05, + "loss": 1.2517, + "step": 449 + }, + { + "epoch": 0.82, + "grad_norm": 0.2617848546539419, + "learning_rate": 7.279593694693698e-05, + "loss": 1.2063, + "step": 450 + }, + { + "epoch": 0.82, + "grad_norm": 0.2698278585334131, + "learning_rate": 7.274705407864332e-05, + "loss": 1.194, + "step": 451 + }, + { + "epoch": 0.83, + "grad_norm": 0.23678313024953834, + "learning_rate": 7.26980224535416e-05, + "loss": 1.2349, + "step": 452 + }, + { + "epoch": 0.83, + "grad_norm": 0.24851875792002978, + "learning_rate": 7.264884229436293e-05, + "loss": 1.1758, + "step": 453 + }, + { + "epoch": 0.83, + "grad_norm": 0.24122080121681125, + "learning_rate": 7.259951382451318e-05, + "loss": 1.1962, + "step": 454 + }, + { + "epoch": 0.83, + "grad_norm": 0.22741322959884405, + "learning_rate": 7.25500372680719e-05, + "loss": 1.1702, + "step": 455 + }, + { + "epoch": 0.83, + "grad_norm": 0.2297475610861458, + "learning_rate": 7.250041284979137e-05, + "loss": 1.1466, + "step": 456 + }, + { + "epoch": 0.84, + "grad_norm": 0.3057605989721467, + "learning_rate": 7.245064079509553e-05, + "loss": 1.246, + "step": 457 + }, + { + "epoch": 0.84, + "grad_norm": 0.2719638501597136, + "learning_rate": 7.240072133007899e-05, + "loss": 1.2184, + "step": 458 + }, + { + "epoch": 0.84, + "grad_norm": 0.2436807816414479, + "learning_rate": 7.235065468150593e-05, + "loss": 1.2324, + "step": 459 + }, + { + "epoch": 0.84, + "grad_norm": 0.23436349430255515, + "learning_rate": 7.23004410768092e-05, + "loss": 1.1813, + "step": 460 + }, + { + "epoch": 0.84, + "grad_norm": 0.2398940990211377, + "learning_rate": 7.22500807440892e-05, + "loss": 1.1924, + "step": 461 + }, + { + "epoch": 0.84, + "grad_norm": 0.2605716625062531, + "learning_rate": 7.219957391211281e-05, + "loss": 1.182, + "step": 462 + }, + { + "epoch": 0.85, + "grad_norm": 0.260462524570941, + "learning_rate": 7.214892081031244e-05, + "loss": 1.2136, + "step": 463 + }, + { + "epoch": 0.85, + "grad_norm": 0.21979766512306334, + "learning_rate": 7.209812166878491e-05, + "loss": 1.2066, + "step": 464 + }, + { + "epoch": 0.85, + "grad_norm": 0.23324453647530663, + "learning_rate": 7.204717671829051e-05, + "loss": 1.1657, + "step": 465 + }, + { + "epoch": 0.85, + "grad_norm": 0.2529434935507481, + "learning_rate": 7.199608619025177e-05, + "loss": 1.2093, + "step": 466 + }, + { + "epoch": 0.85, + "grad_norm": 0.25371701891720116, + "learning_rate": 7.194485031675265e-05, + "loss": 1.2225, + "step": 467 + }, + { + "epoch": 0.86, + "grad_norm": 0.23272423066292103, + "learning_rate": 7.189346933053725e-05, + "loss": 1.1721, + "step": 468 + }, + { + "epoch": 0.86, + "grad_norm": 0.25122928735587546, + "learning_rate": 7.184194346500892e-05, + "loss": 1.2537, + "step": 469 + }, + { + "epoch": 0.86, + "grad_norm": 0.2159270875490409, + "learning_rate": 7.179027295422913e-05, + "loss": 1.197, + "step": 470 + }, + { + "epoch": 0.86, + "grad_norm": 0.2633111059076544, + "learning_rate": 7.173845803291636e-05, + "loss": 1.1721, + "step": 471 + }, + { + "epoch": 0.86, + "grad_norm": 0.30555936322098703, + "learning_rate": 7.168649893644517e-05, + "loss": 1.3011, + "step": 472 + }, + { + "epoch": 0.87, + "grad_norm": 0.23492670111453726, + "learning_rate": 7.163439590084502e-05, + "loss": 1.1601, + "step": 473 + }, + { + "epoch": 0.87, + "grad_norm": 0.26602734263721806, + "learning_rate": 7.158214916279923e-05, + "loss": 1.2808, + "step": 474 + }, + { + "epoch": 0.87, + "grad_norm": 0.3182695007856262, + "learning_rate": 7.152975895964386e-05, + "loss": 1.2967, + "step": 475 + }, + { + "epoch": 0.87, + "grad_norm": 0.2785021674736721, + "learning_rate": 7.147722552936673e-05, + "loss": 1.1789, + "step": 476 + }, + { + "epoch": 0.87, + "grad_norm": 0.279474303138652, + "learning_rate": 7.142454911060627e-05, + "loss": 1.2596, + "step": 477 + }, + { + "epoch": 0.87, + "grad_norm": 0.2556980144910755, + "learning_rate": 7.137172994265044e-05, + "loss": 1.2426, + "step": 478 + }, + { + "epoch": 0.88, + "grad_norm": 0.3311256331993533, + "learning_rate": 7.131876826543565e-05, + "loss": 1.2059, + "step": 479 + }, + { + "epoch": 0.88, + "grad_norm": 0.26467296197775253, + "learning_rate": 7.12656643195457e-05, + "loss": 1.2482, + "step": 480 + }, + { + "epoch": 0.88, + "grad_norm": 0.27444885274652553, + "learning_rate": 7.121241834621064e-05, + "loss": 1.2528, + "step": 481 + }, + { + "epoch": 0.88, + "grad_norm": 0.2572283861115396, + "learning_rate": 7.115903058730567e-05, + "loss": 1.1849, + "step": 482 + }, + { + "epoch": 0.88, + "grad_norm": 0.2677065778235683, + "learning_rate": 7.11055012853501e-05, + "loss": 1.2011, + "step": 483 + }, + { + "epoch": 0.89, + "grad_norm": 0.29470622036742816, + "learning_rate": 7.105183068350619e-05, + "loss": 1.2398, + "step": 484 + }, + { + "epoch": 0.89, + "grad_norm": 0.27609230248969197, + "learning_rate": 7.099801902557811e-05, + "loss": 1.2259, + "step": 485 + }, + { + "epoch": 0.89, + "grad_norm": 0.24248634168099284, + "learning_rate": 7.094406655601073e-05, + "loss": 1.2282, + "step": 486 + }, + { + "epoch": 0.89, + "grad_norm": 0.2765941767688746, + "learning_rate": 7.088997351988865e-05, + "loss": 1.2319, + "step": 487 + }, + { + "epoch": 0.89, + "grad_norm": 0.29347776909858947, + "learning_rate": 7.083574016293493e-05, + "loss": 1.1765, + "step": 488 + }, + { + "epoch": 0.89, + "grad_norm": 0.285370295424537, + "learning_rate": 7.078136673151008e-05, + "loss": 1.26, + "step": 489 + }, + { + "epoch": 0.9, + "grad_norm": 0.29408734903836536, + "learning_rate": 7.072685347261093e-05, + "loss": 1.226, + "step": 490 + }, + { + "epoch": 0.9, + "grad_norm": 0.27437470239205813, + "learning_rate": 7.067220063386947e-05, + "loss": 1.1976, + "step": 491 + }, + { + "epoch": 0.9, + "grad_norm": 0.2680770258777871, + "learning_rate": 7.061740846355176e-05, + "loss": 1.1915, + "step": 492 + }, + { + "epoch": 0.9, + "grad_norm": 0.27200362879502954, + "learning_rate": 7.056247721055678e-05, + "loss": 1.2002, + "step": 493 + }, + { + "epoch": 0.9, + "grad_norm": 0.2637811092577037, + "learning_rate": 7.050740712441528e-05, + "loss": 1.287, + "step": 494 + }, + { + "epoch": 0.91, + "grad_norm": 0.24657959209271266, + "learning_rate": 7.045219845528875e-05, + "loss": 1.2284, + "step": 495 + }, + { + "epoch": 0.91, + "grad_norm": 0.25311992110358666, + "learning_rate": 7.039685145396812e-05, + "loss": 1.1616, + "step": 496 + }, + { + "epoch": 0.91, + "grad_norm": 0.2564633694193358, + "learning_rate": 7.034136637187275e-05, + "loss": 1.2067, + "step": 497 + }, + { + "epoch": 0.91, + "grad_norm": 0.2446797651174144, + "learning_rate": 7.028574346104926e-05, + "loss": 1.2284, + "step": 498 + }, + { + "epoch": 0.91, + "grad_norm": 0.2592751463399255, + "learning_rate": 7.022998297417034e-05, + "loss": 1.2371, + "step": 499 + }, + { + "epoch": 0.91, + "grad_norm": 0.2500713943206808, + "learning_rate": 7.017408516453365e-05, + "loss": 1.1061, + "step": 500 + }, + { + "epoch": 0.92, + "grad_norm": 0.2812266276040743, + "learning_rate": 7.011805028606064e-05, + "loss": 1.1949, + "step": 501 + }, + { + "epoch": 0.92, + "grad_norm": 0.298829667668083, + "learning_rate": 7.006187859329544e-05, + "loss": 1.2313, + "step": 502 + }, + { + "epoch": 0.92, + "grad_norm": 0.26518768159745104, + "learning_rate": 7.000557034140361e-05, + "loss": 1.2246, + "step": 503 + }, + { + "epoch": 0.92, + "grad_norm": 0.3037280360760458, + "learning_rate": 6.994912578617113e-05, + "loss": 1.1617, + "step": 504 + }, + { + "epoch": 0.92, + "grad_norm": 0.2726903109255714, + "learning_rate": 6.989254518400309e-05, + "loss": 1.2415, + "step": 505 + }, + { + "epoch": 0.93, + "grad_norm": 0.25568082003046966, + "learning_rate": 6.98358287919226e-05, + "loss": 1.1817, + "step": 506 + }, + { + "epoch": 0.93, + "grad_norm": 0.25633294893705044, + "learning_rate": 6.97789768675696e-05, + "loss": 1.2149, + "step": 507 + }, + { + "epoch": 0.93, + "grad_norm": 0.28291439435087123, + "learning_rate": 6.972198966919972e-05, + "loss": 1.1578, + "step": 508 + }, + { + "epoch": 0.93, + "grad_norm": 0.27195184756655516, + "learning_rate": 6.966486745568308e-05, + "loss": 1.2355, + "step": 509 + }, + { + "epoch": 0.93, + "grad_norm": 0.239159568376005, + "learning_rate": 6.960761048650312e-05, + "loss": 1.1688, + "step": 510 + }, + { + "epoch": 0.93, + "grad_norm": 0.22961475425949177, + "learning_rate": 6.955021902175543e-05, + "loss": 1.2094, + "step": 511 + }, + { + "epoch": 0.94, + "grad_norm": 0.27443773600741117, + "learning_rate": 6.949269332214651e-05, + "loss": 1.2559, + "step": 512 + }, + { + "epoch": 0.94, + "grad_norm": 0.26230551832002097, + "learning_rate": 6.94350336489927e-05, + "loss": 1.2121, + "step": 513 + }, + { + "epoch": 0.94, + "grad_norm": 0.2716742985303849, + "learning_rate": 6.937724026421892e-05, + "loss": 1.2444, + "step": 514 + }, + { + "epoch": 0.94, + "grad_norm": 0.2537850139439542, + "learning_rate": 6.931931343035742e-05, + "loss": 1.1327, + "step": 515 + }, + { + "epoch": 0.94, + "grad_norm": 0.28599587967496826, + "learning_rate": 6.926125341054676e-05, + "loss": 1.2236, + "step": 516 + }, + { + "epoch": 0.95, + "grad_norm": 0.26780654378470103, + "learning_rate": 6.920306046853043e-05, + "loss": 1.2295, + "step": 517 + }, + { + "epoch": 0.95, + "grad_norm": 0.23606296888412015, + "learning_rate": 6.914473486865577e-05, + "loss": 1.1543, + "step": 518 + }, + { + "epoch": 0.95, + "grad_norm": 0.34976881174240837, + "learning_rate": 6.90862768758727e-05, + "loss": 1.2067, + "step": 519 + }, + { + "epoch": 0.95, + "grad_norm": 0.2481257873494882, + "learning_rate": 6.902768675573258e-05, + "loss": 1.2188, + "step": 520 + }, + { + "epoch": 0.95, + "grad_norm": 0.2996395778117021, + "learning_rate": 6.896896477438699e-05, + "loss": 1.2326, + "step": 521 + }, + { + "epoch": 0.95, + "grad_norm": 0.8839768816333193, + "learning_rate": 6.891011119858643e-05, + "loss": 1.2435, + "step": 522 + }, + { + "epoch": 0.96, + "grad_norm": 0.2851882482058998, + "learning_rate": 6.885112629567927e-05, + "loss": 1.2644, + "step": 523 + }, + { + "epoch": 0.96, + "grad_norm": 0.2813663482913699, + "learning_rate": 6.879201033361035e-05, + "loss": 1.2309, + "step": 524 + }, + { + "epoch": 0.96, + "grad_norm": 0.3257551560135454, + "learning_rate": 6.873276358091996e-05, + "loss": 1.2755, + "step": 525 + }, + { + "epoch": 0.96, + "grad_norm": 0.28930479952494365, + "learning_rate": 6.867338630674247e-05, + "loss": 1.1962, + "step": 526 + }, + { + "epoch": 0.96, + "grad_norm": 0.3077462996938649, + "learning_rate": 6.861387878080511e-05, + "loss": 1.2402, + "step": 527 + }, + { + "epoch": 0.97, + "grad_norm": 0.2848900193452761, + "learning_rate": 6.855424127342688e-05, + "loss": 1.2748, + "step": 528 + }, + { + "epoch": 0.97, + "grad_norm": 0.4765938812802202, + "learning_rate": 6.849447405551718e-05, + "loss": 1.2226, + "step": 529 + }, + { + "epoch": 0.97, + "grad_norm": 0.53184473292579, + "learning_rate": 6.843457739857467e-05, + "loss": 1.2347, + "step": 530 + }, + { + "epoch": 0.97, + "grad_norm": 0.6416239346492343, + "learning_rate": 6.837455157468596e-05, + "loss": 1.2429, + "step": 531 + }, + { + "epoch": 0.97, + "grad_norm": 0.3188092712502773, + "learning_rate": 6.831439685652442e-05, + "loss": 1.216, + "step": 532 + }, + { + "epoch": 0.97, + "grad_norm": 0.3527495731006385, + "learning_rate": 6.825411351734895e-05, + "loss": 1.1682, + "step": 533 + }, + { + "epoch": 0.98, + "grad_norm": 0.29603753744741856, + "learning_rate": 6.819370183100274e-05, + "loss": 1.1434, + "step": 534 + }, + { + "epoch": 0.98, + "grad_norm": 0.5252450389976622, + "learning_rate": 6.813316207191198e-05, + "loss": 1.1943, + "step": 535 + }, + { + "epoch": 0.98, + "grad_norm": 0.32999419558659937, + "learning_rate": 6.807249451508466e-05, + "loss": 1.192, + "step": 536 + }, + { + "epoch": 0.98, + "grad_norm": 0.3650175469778724, + "learning_rate": 6.801169943610929e-05, + "loss": 1.2141, + "step": 537 + }, + { + "epoch": 0.98, + "grad_norm": 1.0643532150783557, + "learning_rate": 6.795077711115368e-05, + "loss": 1.2253, + "step": 538 + }, + { + "epoch": 0.99, + "grad_norm": 0.5041310609130145, + "learning_rate": 6.788972781696363e-05, + "loss": 1.278, + "step": 539 + }, + { + "epoch": 0.99, + "grad_norm": 0.5123058164360991, + "learning_rate": 6.782855183086177e-05, + "loss": 1.2231, + "step": 540 + }, + { + "epoch": 0.99, + "grad_norm": 0.3533015702394419, + "learning_rate": 6.776724943074619e-05, + "loss": 1.2072, + "step": 541 + }, + { + "epoch": 0.99, + "grad_norm": 0.30253964625417207, + "learning_rate": 6.770582089508927e-05, + "loss": 1.1382, + "step": 542 + }, + { + "epoch": 0.99, + "grad_norm": 0.348991618828202, + "learning_rate": 6.764426650293633e-05, + "loss": 1.2079, + "step": 543 + }, + { + "epoch": 0.99, + "grad_norm": 0.46017440578788743, + "learning_rate": 6.758258653390444e-05, + "loss": 1.1813, + "step": 544 + }, + { + "epoch": 1.0, + "grad_norm": 0.31962101755594885, + "learning_rate": 6.75207812681811e-05, + "loss": 1.1339, + "step": 545 + }, + { + "epoch": 1.0, + "grad_norm": 0.37092024548285923, + "learning_rate": 6.745885098652298e-05, + "loss": 1.2591, + "step": 546 + }, + { + "epoch": 1.0, + "grad_norm": 0.32347106450715835, + "learning_rate": 6.739679597025466e-05, + "loss": 1.2017, + "step": 547 + }, + { + "epoch": 1.0, + "grad_norm": 0.39250187112342494, + "learning_rate": 6.733461650126733e-05, + "loss": 1.0933, + "step": 548 + }, + { + "epoch": 1.0, + "grad_norm": 0.473522452217324, + "learning_rate": 6.727231286201752e-05, + "loss": 1.1124, + "step": 549 + }, + { + "epoch": 1.01, + "grad_norm": 0.4809062179622052, + "learning_rate": 6.720988533552582e-05, + "loss": 1.1585, + "step": 550 + }, + { + "epoch": 1.01, + "grad_norm": 0.3529662801059162, + "learning_rate": 6.714733420537559e-05, + "loss": 1.0501, + "step": 551 + }, + { + "epoch": 1.01, + "grad_norm": 0.5958247214391118, + "learning_rate": 6.708465975571168e-05, + "loss": 1.1086, + "step": 552 + }, + { + "epoch": 1.01, + "grad_norm": 0.5341364205022454, + "learning_rate": 6.70218622712391e-05, + "loss": 1.0518, + "step": 553 + }, + { + "epoch": 1.01, + "grad_norm": 0.3601805724462006, + "learning_rate": 6.695894203722181e-05, + "loss": 1.1779, + "step": 554 + }, + { + "epoch": 1.02, + "grad_norm": 0.43410190338280613, + "learning_rate": 6.68958993394813e-05, + "loss": 1.093, + "step": 555 + }, + { + "epoch": 1.02, + "grad_norm": 0.46217742572873594, + "learning_rate": 6.683273446439546e-05, + "loss": 1.0117, + "step": 556 + }, + { + "epoch": 1.02, + "grad_norm": 0.8591682373623357, + "learning_rate": 6.676944769889708e-05, + "loss": 1.1002, + "step": 557 + }, + { + "epoch": 1.02, + "grad_norm": 0.7383229487622726, + "learning_rate": 6.670603933047272e-05, + "loss": 1.0779, + "step": 558 + }, + { + "epoch": 1.02, + "grad_norm": 0.5965305891207813, + "learning_rate": 6.664250964716131e-05, + "loss": 1.0889, + "step": 559 + }, + { + "epoch": 1.02, + "grad_norm": 0.6030858606684543, + "learning_rate": 6.657885893755288e-05, + "loss": 1.0982, + "step": 560 + }, + { + "epoch": 1.03, + "grad_norm": 0.4644510682398409, + "learning_rate": 6.65150874907872e-05, + "loss": 1.1004, + "step": 561 + }, + { + "epoch": 1.03, + "grad_norm": 0.43943285132452564, + "learning_rate": 6.645119559655254e-05, + "loss": 1.0536, + "step": 562 + }, + { + "epoch": 1.03, + "grad_norm": 0.4456395978600012, + "learning_rate": 6.638718354508427e-05, + "loss": 1.0733, + "step": 563 + }, + { + "epoch": 1.03, + "grad_norm": 0.3303824433217466, + "learning_rate": 6.632305162716365e-05, + "loss": 1.0552, + "step": 564 + }, + { + "epoch": 1.03, + "grad_norm": 0.3617704823170143, + "learning_rate": 6.62588001341164e-05, + "loss": 1.1092, + "step": 565 + }, + { + "epoch": 1.04, + "grad_norm": 0.4465013349903427, + "learning_rate": 6.619442935781141e-05, + "loss": 1.0781, + "step": 566 + }, + { + "epoch": 1.04, + "grad_norm": 0.48516780613791277, + "learning_rate": 6.612993959065947e-05, + "loss": 1.0686, + "step": 567 + }, + { + "epoch": 1.04, + "grad_norm": 0.38867820318536633, + "learning_rate": 6.606533112561186e-05, + "loss": 1.1215, + "step": 568 + }, + { + "epoch": 1.04, + "grad_norm": 0.38566119820378336, + "learning_rate": 6.600060425615907e-05, + "loss": 1.1213, + "step": 569 + }, + { + "epoch": 1.04, + "grad_norm": 0.35534855445058544, + "learning_rate": 6.593575927632947e-05, + "loss": 1.0955, + "step": 570 + }, + { + "epoch": 1.04, + "grad_norm": 0.38124406233349717, + "learning_rate": 6.587079648068795e-05, + "loss": 1.0659, + "step": 571 + }, + { + "epoch": 1.05, + "grad_norm": 0.454750160923548, + "learning_rate": 6.580571616433457e-05, + "loss": 1.1149, + "step": 572 + }, + { + "epoch": 1.05, + "grad_norm": 0.35353190088025255, + "learning_rate": 6.574051862290325e-05, + "loss": 1.0388, + "step": 573 + }, + { + "epoch": 1.05, + "grad_norm": 0.3249395594793626, + "learning_rate": 6.567520415256045e-05, + "loss": 1.0784, + "step": 574 + }, + { + "epoch": 1.05, + "grad_norm": 0.40078898818247227, + "learning_rate": 6.560977305000375e-05, + "loss": 1.0859, + "step": 575 + }, + { + "epoch": 1.05, + "grad_norm": 0.4115264795060035, + "learning_rate": 6.554422561246054e-05, + "loss": 1.1828, + "step": 576 + }, + { + "epoch": 1.06, + "grad_norm": 0.30090229228069215, + "learning_rate": 6.54785621376867e-05, + "loss": 1.0901, + "step": 577 + }, + { + "epoch": 1.06, + "grad_norm": 0.28827860350299206, + "learning_rate": 6.541278292396523e-05, + "loss": 1.0277, + "step": 578 + }, + { + "epoch": 1.06, + "grad_norm": 0.34690404488996757, + "learning_rate": 6.534688827010484e-05, + "loss": 1.048, + "step": 579 + }, + { + "epoch": 1.06, + "grad_norm": 0.29943113556644785, + "learning_rate": 6.528087847543867e-05, + "loss": 1.0646, + "step": 580 + }, + { + "epoch": 1.06, + "grad_norm": 0.37318202575874415, + "learning_rate": 6.521475383982291e-05, + "loss": 1.1091, + "step": 581 + }, + { + "epoch": 1.06, + "grad_norm": 0.3049663659203959, + "learning_rate": 6.51485146636354e-05, + "loss": 1.0552, + "step": 582 + }, + { + "epoch": 1.07, + "grad_norm": 0.3342407867509692, + "learning_rate": 6.508216124777431e-05, + "loss": 1.2227, + "step": 583 + }, + { + "epoch": 1.07, + "grad_norm": 0.3348396047855952, + "learning_rate": 6.501569389365674e-05, + "loss": 1.0861, + "step": 584 + }, + { + "epoch": 1.07, + "grad_norm": 0.30951429367513383, + "learning_rate": 6.494911290321737e-05, + "loss": 1.0461, + "step": 585 + }, + { + "epoch": 1.07, + "grad_norm": 0.33898401361064606, + "learning_rate": 6.488241857890711e-05, + "loss": 1.0854, + "step": 586 + }, + { + "epoch": 1.07, + "grad_norm": 0.4901462068263497, + "learning_rate": 6.481561122369164e-05, + "loss": 1.1012, + "step": 587 + }, + { + "epoch": 1.08, + "grad_norm": 0.3179574879809652, + "learning_rate": 6.474869114105018e-05, + "loss": 1.0451, + "step": 588 + }, + { + "epoch": 1.08, + "grad_norm": 0.32159328915060714, + "learning_rate": 6.468165863497395e-05, + "loss": 1.0458, + "step": 589 + }, + { + "epoch": 1.08, + "grad_norm": 0.36462235008537297, + "learning_rate": 6.461451400996491e-05, + "loss": 1.1247, + "step": 590 + }, + { + "epoch": 1.08, + "grad_norm": 0.5373862753611778, + "learning_rate": 6.454725757103432e-05, + "loss": 1.0542, + "step": 591 + }, + { + "epoch": 1.08, + "grad_norm": 0.3160409270291303, + "learning_rate": 6.447988962370133e-05, + "loss": 1.0829, + "step": 592 + }, + { + "epoch": 1.08, + "grad_norm": 0.390452102978435, + "learning_rate": 6.441241047399169e-05, + "loss": 1.192, + "step": 593 + }, + { + "epoch": 1.09, + "grad_norm": 0.3802122712014928, + "learning_rate": 6.434482042843627e-05, + "loss": 1.1153, + "step": 594 + }, + { + "epoch": 1.09, + "grad_norm": 0.4081584328242501, + "learning_rate": 6.427711979406966e-05, + "loss": 1.1635, + "step": 595 + }, + { + "epoch": 1.09, + "grad_norm": 0.3791962989638633, + "learning_rate": 6.420930887842889e-05, + "loss": 1.1581, + "step": 596 + }, + { + "epoch": 1.09, + "grad_norm": 0.33239440056484193, + "learning_rate": 6.414138798955189e-05, + "loss": 1.0926, + "step": 597 + }, + { + "epoch": 1.09, + "grad_norm": 0.3279881540815014, + "learning_rate": 6.407335743597616e-05, + "loss": 1.1386, + "step": 598 + }, + { + "epoch": 1.1, + "grad_norm": 0.30309644763750837, + "learning_rate": 6.40052175267374e-05, + "loss": 1.0523, + "step": 599 + }, + { + "epoch": 1.1, + "grad_norm": 0.3349097308403333, + "learning_rate": 6.393696857136801e-05, + "loss": 1.0815, + "step": 600 + }, + { + "epoch": 1.1, + "grad_norm": 0.3288227593556618, + "learning_rate": 6.386861087989581e-05, + "loss": 1.015, + "step": 601 + }, + { + "epoch": 1.1, + "grad_norm": 0.36685586740843157, + "learning_rate": 6.380014476284255e-05, + "loss": 1.1232, + "step": 602 + }, + { + "epoch": 1.1, + "grad_norm": 0.3620977714204643, + "learning_rate": 6.373157053122243e-05, + "loss": 1.1138, + "step": 603 + }, + { + "epoch": 1.1, + "grad_norm": 0.3130587018197183, + "learning_rate": 6.366288849654091e-05, + "loss": 1.1255, + "step": 604 + }, + { + "epoch": 1.11, + "grad_norm": 0.3602737087072766, + "learning_rate": 6.359409897079303e-05, + "loss": 1.0282, + "step": 605 + }, + { + "epoch": 1.11, + "grad_norm": 0.31168852571991945, + "learning_rate": 6.352520226646222e-05, + "loss": 1.0779, + "step": 606 + }, + { + "epoch": 1.11, + "grad_norm": 0.3516045580189353, + "learning_rate": 6.345619869651871e-05, + "loss": 1.1028, + "step": 607 + }, + { + "epoch": 1.11, + "grad_norm": 0.3231857927563657, + "learning_rate": 6.33870885744182e-05, + "loss": 1.1202, + "step": 608 + }, + { + "epoch": 1.11, + "grad_norm": 0.30205205129701157, + "learning_rate": 6.331787221410041e-05, + "loss": 1.1369, + "step": 609 + }, + { + "epoch": 1.12, + "grad_norm": 0.3198359813888166, + "learning_rate": 6.32485499299877e-05, + "loss": 1.1763, + "step": 610 + }, + { + "epoch": 1.12, + "grad_norm": 0.3128641370321787, + "learning_rate": 6.31791220369835e-05, + "loss": 1.0223, + "step": 611 + }, + { + "epoch": 1.12, + "grad_norm": 0.2989105616213649, + "learning_rate": 6.31095888504711e-05, + "loss": 1.0358, + "step": 612 + }, + { + "epoch": 1.12, + "grad_norm": 0.3103537906853337, + "learning_rate": 6.303995068631203e-05, + "loss": 1.1261, + "step": 613 + }, + { + "epoch": 1.12, + "grad_norm": 0.28598715532508207, + "learning_rate": 6.297020786084467e-05, + "loss": 1.0629, + "step": 614 + }, + { + "epoch": 1.12, + "grad_norm": 0.29809789918093255, + "learning_rate": 6.290036069088288e-05, + "loss": 1.035, + "step": 615 + }, + { + "epoch": 1.13, + "grad_norm": 0.33765270252261453, + "learning_rate": 6.283040949371451e-05, + "loss": 1.1221, + "step": 616 + }, + { + "epoch": 1.13, + "grad_norm": 0.3424617501293415, + "learning_rate": 6.276035458709993e-05, + "loss": 1.155, + "step": 617 + }, + { + "epoch": 1.13, + "grad_norm": 0.3799189737987811, + "learning_rate": 6.269019628927067e-05, + "loss": 1.0701, + "step": 618 + }, + { + "epoch": 1.13, + "grad_norm": 0.3358898935253196, + "learning_rate": 6.261993491892791e-05, + "loss": 1.1649, + "step": 619 + }, + { + "epoch": 1.13, + "grad_norm": 0.31569979424117356, + "learning_rate": 6.254957079524099e-05, + "loss": 1.0633, + "step": 620 + }, + { + "epoch": 1.14, + "grad_norm": 0.3002168156888237, + "learning_rate": 6.247910423784609e-05, + "loss": 1.0846, + "step": 621 + }, + { + "epoch": 1.14, + "grad_norm": 0.3097238823450595, + "learning_rate": 6.24085355668447e-05, + "loss": 1.0808, + "step": 622 + }, + { + "epoch": 1.14, + "grad_norm": 0.3120312761417578, + "learning_rate": 6.233786510280212e-05, + "loss": 1.0142, + "step": 623 + }, + { + "epoch": 1.14, + "grad_norm": 0.3335343015064923, + "learning_rate": 6.22670931667461e-05, + "loss": 1.0674, + "step": 624 + }, + { + "epoch": 1.14, + "grad_norm": 0.3234062304634526, + "learning_rate": 6.219622008016533e-05, + "loss": 1.0981, + "step": 625 + }, + { + "epoch": 1.14, + "grad_norm": 0.32152678786547273, + "learning_rate": 6.212524616500798e-05, + "loss": 1.0244, + "step": 626 + }, + { + "epoch": 1.15, + "grad_norm": 0.39031977608147594, + "learning_rate": 6.205417174368023e-05, + "loss": 1.1205, + "step": 627 + }, + { + "epoch": 1.15, + "grad_norm": 0.3806189090017157, + "learning_rate": 6.198299713904485e-05, + "loss": 1.1134, + "step": 628 + }, + { + "epoch": 1.15, + "grad_norm": 0.2978349276971668, + "learning_rate": 6.191172267441967e-05, + "loss": 1.0088, + "step": 629 + }, + { + "epoch": 1.15, + "grad_norm": 0.3190354077382501, + "learning_rate": 6.184034867357617e-05, + "loss": 1.108, + "step": 630 + }, + { + "epoch": 1.15, + "grad_norm": 0.32633048665038994, + "learning_rate": 6.176887546073797e-05, + "loss": 1.0825, + "step": 631 + }, + { + "epoch": 1.16, + "grad_norm": 0.3428026413020903, + "learning_rate": 6.169730336057939e-05, + "loss": 1.0765, + "step": 632 + }, + { + "epoch": 1.16, + "grad_norm": 0.3475737151929015, + "learning_rate": 6.162563269822391e-05, + "loss": 1.0693, + "step": 633 + }, + { + "epoch": 1.16, + "grad_norm": 0.3870252154591392, + "learning_rate": 6.15538637992428e-05, + "loss": 1.1081, + "step": 634 + }, + { + "epoch": 1.16, + "grad_norm": 0.33597355193652834, + "learning_rate": 6.148199698965352e-05, + "loss": 1.0893, + "step": 635 + }, + { + "epoch": 1.16, + "grad_norm": 0.30805894179787247, + "learning_rate": 6.141003259591834e-05, + "loss": 1.0995, + "step": 636 + }, + { + "epoch": 1.17, + "grad_norm": 0.3025073882734066, + "learning_rate": 6.133797094494281e-05, + "loss": 1.0388, + "step": 637 + }, + { + "epoch": 1.17, + "grad_norm": 0.3524395196391662, + "learning_rate": 6.126581236407429e-05, + "loss": 1.1196, + "step": 638 + }, + { + "epoch": 1.17, + "grad_norm": 0.3377646188130345, + "learning_rate": 6.119355718110039e-05, + "loss": 1.0382, + "step": 639 + }, + { + "epoch": 1.17, + "grad_norm": 0.35508400659785483, + "learning_rate": 6.112120572424763e-05, + "loss": 1.1402, + "step": 640 + }, + { + "epoch": 1.17, + "grad_norm": 0.3454418793700457, + "learning_rate": 6.104875832217982e-05, + "loss": 1.1032, + "step": 641 + }, + { + "epoch": 1.17, + "grad_norm": 0.32629806837059866, + "learning_rate": 6.097621530399661e-05, + "loss": 1.0959, + "step": 642 + }, + { + "epoch": 1.18, + "grad_norm": 0.3329536837751315, + "learning_rate": 6.090357699923202e-05, + "loss": 1.0467, + "step": 643 + }, + { + "epoch": 1.18, + "grad_norm": 0.32302233828349475, + "learning_rate": 6.083084373785287e-05, + "loss": 1.0858, + "step": 644 + }, + { + "epoch": 1.18, + "grad_norm": 0.3310358826507611, + "learning_rate": 6.075801585025739e-05, + "loss": 1.0715, + "step": 645 + }, + { + "epoch": 1.18, + "grad_norm": 0.319322035854079, + "learning_rate": 6.068509366727362e-05, + "loss": 1.177, + "step": 646 + }, + { + "epoch": 1.18, + "grad_norm": 0.3065230667302707, + "learning_rate": 6.061207752015797e-05, + "loss": 1.0649, + "step": 647 + }, + { + "epoch": 1.19, + "grad_norm": 0.29926795565748227, + "learning_rate": 6.053896774059368e-05, + "loss": 1.1325, + "step": 648 + }, + { + "epoch": 1.19, + "grad_norm": 0.3556069634279046, + "learning_rate": 6.046576466068931e-05, + "loss": 1.1366, + "step": 649 + }, + { + "epoch": 1.19, + "grad_norm": 0.3189191131461966, + "learning_rate": 6.039246861297727e-05, + "loss": 1.0693, + "step": 650 + }, + { + "epoch": 1.19, + "grad_norm": 0.3347197156648834, + "learning_rate": 6.031907993041227e-05, + "loss": 1.1009, + "step": 651 + }, + { + "epoch": 1.19, + "grad_norm": 0.32274156348185445, + "learning_rate": 6.0245598946369826e-05, + "loss": 1.1675, + "step": 652 + }, + { + "epoch": 1.19, + "grad_norm": 0.35534089035455224, + "learning_rate": 6.017202599464476e-05, + "loss": 1.1723, + "step": 653 + }, + { + "epoch": 1.2, + "grad_norm": 0.3106026578570133, + "learning_rate": 6.009836140944965e-05, + "loss": 1.0954, + "step": 654 + }, + { + "epoch": 1.2, + "grad_norm": 0.3309144454564729, + "learning_rate": 6.002460552541331e-05, + "loss": 1.0209, + "step": 655 + }, + { + "epoch": 1.2, + "grad_norm": 0.3023619281400003, + "learning_rate": 5.9950758677579345e-05, + "loss": 1.0363, + "step": 656 + }, + { + "epoch": 1.2, + "grad_norm": 0.3311182880219704, + "learning_rate": 5.987682120140451e-05, + "loss": 1.0515, + "step": 657 + }, + { + "epoch": 1.2, + "grad_norm": 0.33396486010030413, + "learning_rate": 5.980279343275729e-05, + "loss": 1.1251, + "step": 658 + }, + { + "epoch": 1.21, + "grad_norm": 0.3465764556678002, + "learning_rate": 5.97286757079163e-05, + "loss": 1.165, + "step": 659 + }, + { + "epoch": 1.21, + "grad_norm": 0.304193441363374, + "learning_rate": 5.965446836356882e-05, + "loss": 1.0228, + "step": 660 + }, + { + "epoch": 1.21, + "grad_norm": 0.3415149030413082, + "learning_rate": 5.9580171736809224e-05, + "loss": 1.0742, + "step": 661 + }, + { + "epoch": 1.21, + "grad_norm": 0.33138658321132064, + "learning_rate": 5.950578616513746e-05, + "loss": 1.0843, + "step": 662 + }, + { + "epoch": 1.21, + "grad_norm": 0.30774403421162994, + "learning_rate": 5.943131198645752e-05, + "loss": 1.065, + "step": 663 + }, + { + "epoch": 1.21, + "grad_norm": 0.3428877492183819, + "learning_rate": 5.9356749539075885e-05, + "loss": 1.1101, + "step": 664 + }, + { + "epoch": 1.22, + "grad_norm": 0.3621290546130101, + "learning_rate": 5.928209916170003e-05, + "loss": 1.1372, + "step": 665 + }, + { + "epoch": 1.22, + "grad_norm": 0.3482375945469884, + "learning_rate": 5.9207361193436865e-05, + "loss": 1.132, + "step": 666 + }, + { + "epoch": 1.22, + "grad_norm": 0.31754384974068384, + "learning_rate": 5.9132535973791156e-05, + "loss": 1.148, + "step": 667 + }, + { + "epoch": 1.22, + "grad_norm": 0.36003834782050365, + "learning_rate": 5.9057623842664044e-05, + "loss": 1.1099, + "step": 668 + }, + { + "epoch": 1.22, + "grad_norm": 0.2963701622969662, + "learning_rate": 5.8982625140351464e-05, + "loss": 1.0755, + "step": 669 + }, + { + "epoch": 1.23, + "grad_norm": 0.32579569606066516, + "learning_rate": 5.8907540207542616e-05, + "loss": 1.0809, + "step": 670 + }, + { + "epoch": 1.23, + "grad_norm": 0.4247563451753457, + "learning_rate": 5.8832369385318416e-05, + "loss": 1.097, + "step": 671 + }, + { + "epoch": 1.23, + "grad_norm": 0.33076932102169776, + "learning_rate": 5.875711301514992e-05, + "loss": 1.1078, + "step": 672 + }, + { + "epoch": 1.23, + "grad_norm": 0.3609238032332309, + "learning_rate": 5.8681771438896815e-05, + "loss": 1.1031, + "step": 673 + }, + { + "epoch": 1.23, + "grad_norm": 0.325159585649425, + "learning_rate": 5.860634499880583e-05, + "loss": 1.0707, + "step": 674 + }, + { + "epoch": 1.23, + "grad_norm": 0.4620687271068983, + "learning_rate": 5.853083403750922e-05, + "loss": 1.1017, + "step": 675 + }, + { + "epoch": 1.24, + "grad_norm": 0.33485279064365936, + "learning_rate": 5.845523889802316e-05, + "loss": 1.0989, + "step": 676 + }, + { + "epoch": 1.24, + "grad_norm": 0.30952573170841513, + "learning_rate": 5.8379559923746214e-05, + "loss": 1.0393, + "step": 677 + }, + { + "epoch": 1.24, + "grad_norm": 0.33498605810588283, + "learning_rate": 5.830379745845781e-05, + "loss": 1.1259, + "step": 678 + }, + { + "epoch": 1.24, + "grad_norm": 0.35771921163037307, + "learning_rate": 5.822795184631659e-05, + "loss": 1.0815, + "step": 679 + }, + { + "epoch": 1.24, + "grad_norm": 0.3329650192347647, + "learning_rate": 5.815202343185894e-05, + "loss": 1.1344, + "step": 680 + }, + { + "epoch": 1.25, + "grad_norm": 0.3356634465845771, + "learning_rate": 5.807601255999736e-05, + "loss": 1.1297, + "step": 681 + }, + { + "epoch": 1.25, + "grad_norm": 0.3289442034151235, + "learning_rate": 5.7999919576018934e-05, + "loss": 1.022, + "step": 682 + }, + { + "epoch": 1.25, + "grad_norm": 0.3207007334784113, + "learning_rate": 5.7923744825583745e-05, + "loss": 1.0571, + "step": 683 + }, + { + "epoch": 1.25, + "grad_norm": 0.3582460325329284, + "learning_rate": 5.7847488654723304e-05, + "loss": 1.0778, + "step": 684 + }, + { + "epoch": 1.25, + "grad_norm": 0.3563317666176927, + "learning_rate": 5.777115140983899e-05, + "loss": 1.1003, + "step": 685 + }, + { + "epoch": 1.25, + "grad_norm": 3.4694912945702105, + "learning_rate": 5.769473343770047e-05, + "loss": 1.121, + "step": 686 + }, + { + "epoch": 1.26, + "grad_norm": 0.43002349520483113, + "learning_rate": 5.761823508544411e-05, + "loss": 1.0765, + "step": 687 + }, + { + "epoch": 1.26, + "grad_norm": 0.39467783104839754, + "learning_rate": 5.754165670057142e-05, + "loss": 1.0788, + "step": 688 + }, + { + "epoch": 1.26, + "grad_norm": 0.39629029674867916, + "learning_rate": 5.7464998630947464e-05, + "loss": 1.0812, + "step": 689 + }, + { + "epoch": 1.26, + "grad_norm": 0.3880152093965208, + "learning_rate": 5.738826122479929e-05, + "loss": 1.1228, + "step": 690 + }, + { + "epoch": 1.26, + "grad_norm": 0.3777874121959188, + "learning_rate": 5.7311444830714324e-05, + "loss": 1.0907, + "step": 691 + }, + { + "epoch": 1.27, + "grad_norm": 0.38004041653523696, + "learning_rate": 5.723454979763882e-05, + "loss": 1.1263, + "step": 692 + }, + { + "epoch": 1.27, + "grad_norm": 0.37049672627797636, + "learning_rate": 5.7157576474876246e-05, + "loss": 1.1438, + "step": 693 + }, + { + "epoch": 1.27, + "grad_norm": 0.32973606103437614, + "learning_rate": 5.7080525212085725e-05, + "loss": 1.0553, + "step": 694 + }, + { + "epoch": 1.27, + "grad_norm": 0.31674639252070325, + "learning_rate": 5.700339635928038e-05, + "loss": 1.06, + "step": 695 + }, + { + "epoch": 1.27, + "grad_norm": 0.32282199426553837, + "learning_rate": 5.692619026682588e-05, + "loss": 1.0841, + "step": 696 + }, + { + "epoch": 1.27, + "grad_norm": 0.4810882958061859, + "learning_rate": 5.684890728543869e-05, + "loss": 1.0803, + "step": 697 + }, + { + "epoch": 1.28, + "grad_norm": 0.3995638550178378, + "learning_rate": 5.6771547766184566e-05, + "loss": 1.1187, + "step": 698 + }, + { + "epoch": 1.28, + "grad_norm": 0.35264932960583484, + "learning_rate": 5.669411206047699e-05, + "loss": 1.0641, + "step": 699 + }, + { + "epoch": 1.28, + "grad_norm": 0.35240640524733, + "learning_rate": 5.661660052007547e-05, + "loss": 1.076, + "step": 700 + }, + { + "epoch": 1.28, + "grad_norm": 0.3540694609860389, + "learning_rate": 5.653901349708401e-05, + "loss": 1.1369, + "step": 701 + }, + { + "epoch": 1.28, + "grad_norm": 0.3196055112925304, + "learning_rate": 5.646135134394955e-05, + "loss": 1.0677, + "step": 702 + }, + { + "epoch": 1.29, + "grad_norm": 0.4214141007955914, + "learning_rate": 5.6383614413460266e-05, + "loss": 1.1139, + "step": 703 + }, + { + "epoch": 1.29, + "grad_norm": 0.3625611311798579, + "learning_rate": 5.630580305874402e-05, + "loss": 1.1845, + "step": 704 + }, + { + "epoch": 1.29, + "grad_norm": 0.3425208672181188, + "learning_rate": 5.62279176332668e-05, + "loss": 1.174, + "step": 705 + }, + { + "epoch": 1.29, + "grad_norm": 0.3108419862818321, + "learning_rate": 5.6149958490830996e-05, + "loss": 1.0331, + "step": 706 + }, + { + "epoch": 1.29, + "grad_norm": 0.3274644181571904, + "learning_rate": 5.607192598557394e-05, + "loss": 1.0664, + "step": 707 + }, + { + "epoch": 1.29, + "grad_norm": 0.346218197215145, + "learning_rate": 5.599382047196617e-05, + "loss": 1.2088, + "step": 708 + }, + { + "epoch": 1.3, + "grad_norm": 0.328497632267458, + "learning_rate": 5.591564230480989e-05, + "loss": 1.0287, + "step": 709 + }, + { + "epoch": 1.3, + "grad_norm": 0.3708173720611468, + "learning_rate": 5.583739183923732e-05, + "loss": 1.0883, + "step": 710 + }, + { + "epoch": 1.3, + "grad_norm": 0.3631427403535479, + "learning_rate": 5.575906943070915e-05, + "loss": 1.1155, + "step": 711 + }, + { + "epoch": 1.3, + "grad_norm": 0.3305201458598695, + "learning_rate": 5.5680675435012834e-05, + "loss": 1.0958, + "step": 712 + }, + { + "epoch": 1.3, + "grad_norm": 0.34978833532083714, + "learning_rate": 5.5602210208261036e-05, + "loss": 1.1437, + "step": 713 + }, + { + "epoch": 1.31, + "grad_norm": 0.3510553882510229, + "learning_rate": 5.552367410688999e-05, + "loss": 1.0941, + "step": 714 + }, + { + "epoch": 1.31, + "grad_norm": 0.3523747462465078, + "learning_rate": 5.544506748765789e-05, + "loss": 1.1289, + "step": 715 + }, + { + "epoch": 1.31, + "grad_norm": 0.38262637783927445, + "learning_rate": 5.5366390707643266e-05, + "loss": 1.099, + "step": 716 + }, + { + "epoch": 1.31, + "grad_norm": 0.38620065989073454, + "learning_rate": 5.528764412424334e-05, + "loss": 1.083, + "step": 717 + }, + { + "epoch": 1.31, + "grad_norm": 0.3401355276121096, + "learning_rate": 5.520882809517245e-05, + "loss": 1.028, + "step": 718 + }, + { + "epoch": 1.32, + "grad_norm": 0.3392061008943934, + "learning_rate": 5.512994297846039e-05, + "loss": 1.1083, + "step": 719 + }, + { + "epoch": 1.32, + "grad_norm": 0.34219480421015414, + "learning_rate": 5.505098913245077e-05, + "loss": 1.1108, + "step": 720 + }, + { + "epoch": 1.32, + "grad_norm": 0.3275058061553761, + "learning_rate": 5.497196691579945e-05, + "loss": 1.111, + "step": 721 + }, + { + "epoch": 1.32, + "grad_norm": 0.36800249746509384, + "learning_rate": 5.489287668747283e-05, + "loss": 1.1221, + "step": 722 + }, + { + "epoch": 1.32, + "grad_norm": 0.4129005533101575, + "learning_rate": 5.481371880674628e-05, + "loss": 1.0966, + "step": 723 + }, + { + "epoch": 1.32, + "grad_norm": 0.36563906596251655, + "learning_rate": 5.4734493633202505e-05, + "loss": 1.0927, + "step": 724 + }, + { + "epoch": 1.33, + "grad_norm": 0.3614650536839971, + "learning_rate": 5.465520152672986e-05, + "loss": 1.13, + "step": 725 + }, + { + "epoch": 1.33, + "grad_norm": 0.36419665098633497, + "learning_rate": 5.4575842847520765e-05, + "loss": 1.1183, + "step": 726 + }, + { + "epoch": 1.33, + "grad_norm": 0.34490689807258995, + "learning_rate": 5.449641795607005e-05, + "loss": 1.0919, + "step": 727 + }, + { + "epoch": 1.33, + "grad_norm": 0.3627643746876298, + "learning_rate": 5.441692721317334e-05, + "loss": 1.0411, + "step": 728 + }, + { + "epoch": 1.33, + "grad_norm": 0.323620411949565, + "learning_rate": 5.433737097992537e-05, + "loss": 1.0725, + "step": 729 + }, + { + "epoch": 1.34, + "grad_norm": 0.3521599501824965, + "learning_rate": 5.425774961771838e-05, + "loss": 1.0926, + "step": 730 + }, + { + "epoch": 1.34, + "grad_norm": 0.3302390546764222, + "learning_rate": 5.417806348824047e-05, + "loss": 1.0468, + "step": 731 + }, + { + "epoch": 1.34, + "grad_norm": 0.3833325802616019, + "learning_rate": 5.4098312953473956e-05, + "loss": 1.1291, + "step": 732 + }, + { + "epoch": 1.34, + "grad_norm": 0.3708621126835512, + "learning_rate": 5.401849837569372e-05, + "loss": 1.0887, + "step": 733 + }, + { + "epoch": 1.34, + "grad_norm": 0.3625834373416278, + "learning_rate": 5.393862011746555e-05, + "loss": 1.0981, + "step": 734 + }, + { + "epoch": 1.34, + "grad_norm": 0.3583343965080617, + "learning_rate": 5.385867854164451e-05, + "loss": 1.1021, + "step": 735 + }, + { + "epoch": 1.35, + "grad_norm": 0.34598320594096066, + "learning_rate": 5.377867401137332e-05, + "loss": 1.1376, + "step": 736 + }, + { + "epoch": 1.35, + "grad_norm": 0.3046382791315433, + "learning_rate": 5.369860689008066e-05, + "loss": 1.0206, + "step": 737 + }, + { + "epoch": 1.35, + "grad_norm": 0.34464948380043725, + "learning_rate": 5.3618477541479505e-05, + "loss": 1.1084, + "step": 738 + }, + { + "epoch": 1.35, + "grad_norm": 0.3203242519627101, + "learning_rate": 5.353828632956557e-05, + "loss": 1.0731, + "step": 739 + }, + { + "epoch": 1.35, + "grad_norm": 0.3431169960355163, + "learning_rate": 5.3458033618615516e-05, + "loss": 1.091, + "step": 740 + }, + { + "epoch": 1.36, + "grad_norm": 0.33492074521678705, + "learning_rate": 5.337771977318543e-05, + "loss": 1.1112, + "step": 741 + }, + { + "epoch": 1.36, + "grad_norm": 0.32576546585541344, + "learning_rate": 5.3297345158109086e-05, + "loss": 1.0993, + "step": 742 + }, + { + "epoch": 1.36, + "grad_norm": 0.3410007245037574, + "learning_rate": 5.3216910138496286e-05, + "loss": 1.094, + "step": 743 + }, + { + "epoch": 1.36, + "grad_norm": 0.34891180680896833, + "learning_rate": 5.313641507973128e-05, + "loss": 1.1331, + "step": 744 + }, + { + "epoch": 1.36, + "grad_norm": 0.37135766946717214, + "learning_rate": 5.3055860347471006e-05, + "loss": 1.1, + "step": 745 + }, + { + "epoch": 1.36, + "grad_norm": 0.3465019415478411, + "learning_rate": 5.297524630764349e-05, + "loss": 1.1256, + "step": 746 + }, + { + "epoch": 1.37, + "grad_norm": 0.37035388481626563, + "learning_rate": 5.289457332644615e-05, + "loss": 1.0366, + "step": 747 + }, + { + "epoch": 1.37, + "grad_norm": 0.33853883270759155, + "learning_rate": 5.281384177034421e-05, + "loss": 1.0547, + "step": 748 + }, + { + "epoch": 1.37, + "grad_norm": 0.364306618627317, + "learning_rate": 5.2733052006068897e-05, + "loss": 1.0768, + "step": 749 + }, + { + "epoch": 1.37, + "grad_norm": 0.4021754315731627, + "learning_rate": 5.2652204400615916e-05, + "loss": 1.1382, + "step": 750 + }, + { + "epoch": 1.37, + "grad_norm": 0.3332185389039008, + "learning_rate": 5.257129932124368e-05, + "loss": 1.0815, + "step": 751 + }, + { + "epoch": 1.38, + "grad_norm": 0.3453105709879854, + "learning_rate": 5.249033713547173e-05, + "loss": 1.1109, + "step": 752 + }, + { + "epoch": 1.38, + "grad_norm": 0.3385397539717797, + "learning_rate": 5.2409318211078966e-05, + "loss": 1.0529, + "step": 753 + }, + { + "epoch": 1.38, + "grad_norm": 0.33197994450130447, + "learning_rate": 5.232824291610206e-05, + "loss": 1.0721, + "step": 754 + }, + { + "epoch": 1.38, + "grad_norm": 0.32836289576124167, + "learning_rate": 5.224711161883375e-05, + "loss": 1.0459, + "step": 755 + }, + { + "epoch": 1.38, + "grad_norm": 0.32491620058831744, + "learning_rate": 5.216592468782117e-05, + "loss": 1.0897, + "step": 756 + }, + { + "epoch": 1.38, + "grad_norm": 0.3137879047811153, + "learning_rate": 5.2084682491864155e-05, + "loss": 1.096, + "step": 757 + }, + { + "epoch": 1.39, + "grad_norm": 0.3356938043023012, + "learning_rate": 5.200338540001364e-05, + "loss": 1.0827, + "step": 758 + }, + { + "epoch": 1.39, + "grad_norm": 0.36044340490819055, + "learning_rate": 5.192203378156984e-05, + "loss": 1.0617, + "step": 759 + }, + { + "epoch": 1.39, + "grad_norm": 0.34674262047888293, + "learning_rate": 5.184062800608077e-05, + "loss": 1.1267, + "step": 760 + }, + { + "epoch": 1.39, + "grad_norm": 0.32469442322149333, + "learning_rate": 5.1759168443340375e-05, + "loss": 1.1483, + "step": 761 + }, + { + "epoch": 1.39, + "grad_norm": 0.3290384307774216, + "learning_rate": 5.167765546338698e-05, + "loss": 1.047, + "step": 762 + }, + { + "epoch": 1.4, + "grad_norm": 0.31637612188770403, + "learning_rate": 5.1596089436501525e-05, + "loss": 1.0311, + "step": 763 + }, + { + "epoch": 1.4, + "grad_norm": 0.3168693829641207, + "learning_rate": 5.151447073320597e-05, + "loss": 1.1405, + "step": 764 + }, + { + "epoch": 1.4, + "grad_norm": 0.34322421571238926, + "learning_rate": 5.143279972426153e-05, + "loss": 1.1428, + "step": 765 + }, + { + "epoch": 1.4, + "grad_norm": 0.3291030435830325, + "learning_rate": 5.1351076780667026e-05, + "loss": 1.0473, + "step": 766 + }, + { + "epoch": 1.4, + "grad_norm": 0.33772039158758044, + "learning_rate": 5.1269302273657195e-05, + "loss": 1.0909, + "step": 767 + }, + { + "epoch": 1.4, + "grad_norm": 0.3802031736890876, + "learning_rate": 5.118747657470102e-05, + "loss": 1.1482, + "step": 768 + }, + { + "epoch": 1.41, + "grad_norm": 0.3296067628997962, + "learning_rate": 5.1105600055500025e-05, + "loss": 1.0085, + "step": 769 + }, + { + "epoch": 1.41, + "grad_norm": 0.3707139982828035, + "learning_rate": 5.102367308798658e-05, + "loss": 1.0746, + "step": 770 + }, + { + "epoch": 1.41, + "grad_norm": 0.3378537316757011, + "learning_rate": 5.094169604432225e-05, + "loss": 1.0482, + "step": 771 + }, + { + "epoch": 1.41, + "grad_norm": 0.4008417246255145, + "learning_rate": 5.085966929689601e-05, + "loss": 1.1065, + "step": 772 + }, + { + "epoch": 1.41, + "grad_norm": 0.3244385106988064, + "learning_rate": 5.077759321832271e-05, + "loss": 1.0827, + "step": 773 + }, + { + "epoch": 1.42, + "grad_norm": 0.37228575732812336, + "learning_rate": 5.0695468181441215e-05, + "loss": 1.1146, + "step": 774 + }, + { + "epoch": 1.42, + "grad_norm": 0.33761714797540276, + "learning_rate": 5.061329455931283e-05, + "loss": 1.092, + "step": 775 + }, + { + "epoch": 1.42, + "grad_norm": 0.3158158390913494, + "learning_rate": 5.053107272521955e-05, + "loss": 1.1058, + "step": 776 + }, + { + "epoch": 1.42, + "grad_norm": 0.3691501929738938, + "learning_rate": 5.044880305266239e-05, + "loss": 1.1599, + "step": 777 + }, + { + "epoch": 1.42, + "grad_norm": 0.33730914019805525, + "learning_rate": 5.0366485915359645e-05, + "loss": 1.0615, + "step": 778 + }, + { + "epoch": 1.42, + "grad_norm": 0.34970059240017, + "learning_rate": 5.0284121687245257e-05, + "loss": 1.1475, + "step": 779 + }, + { + "epoch": 1.43, + "grad_norm": 0.3374028029407197, + "learning_rate": 5.020171074246707e-05, + "loss": 1.0926, + "step": 780 + }, + { + "epoch": 1.43, + "grad_norm": 0.3350020681123992, + "learning_rate": 5.011925345538514e-05, + "loss": 1.1276, + "step": 781 + }, + { + "epoch": 1.43, + "grad_norm": 0.3224228965786606, + "learning_rate": 5.003675020057003e-05, + "loss": 1.0183, + "step": 782 + }, + { + "epoch": 1.43, + "grad_norm": 0.3357310714740298, + "learning_rate": 4.995420135280114e-05, + "loss": 1.1114, + "step": 783 + }, + { + "epoch": 1.43, + "grad_norm": 0.3590203255363759, + "learning_rate": 4.9871607287064966e-05, + "loss": 1.1504, + "step": 784 + }, + { + "epoch": 1.44, + "grad_norm": 0.33011195419611655, + "learning_rate": 4.9788968378553396e-05, + "loss": 1.0826, + "step": 785 + }, + { + "epoch": 1.44, + "grad_norm": 0.31088868195439445, + "learning_rate": 4.970628500266207e-05, + "loss": 1.0704, + "step": 786 + }, + { + "epoch": 1.44, + "grad_norm": 0.3144996103179409, + "learning_rate": 4.962355753498858e-05, + "loss": 1.1403, + "step": 787 + }, + { + "epoch": 1.44, + "grad_norm": 0.3147269555419068, + "learning_rate": 4.954078635133081e-05, + "loss": 1.0898, + "step": 788 + }, + { + "epoch": 1.44, + "grad_norm": 0.3280151747783868, + "learning_rate": 4.945797182768524e-05, + "loss": 1.1115, + "step": 789 + }, + { + "epoch": 1.44, + "grad_norm": 0.3551996569232493, + "learning_rate": 4.937511434024524e-05, + "loss": 1.1731, + "step": 790 + }, + { + "epoch": 1.45, + "grad_norm": 0.343863208057807, + "learning_rate": 4.9292214265399336e-05, + "loss": 1.0866, + "step": 791 + }, + { + "epoch": 1.45, + "grad_norm": 0.37316699385322466, + "learning_rate": 4.920927197972949e-05, + "loss": 1.1083, + "step": 792 + }, + { + "epoch": 1.45, + "grad_norm": 0.3635739774067832, + "learning_rate": 4.9126287860009453e-05, + "loss": 1.1393, + "step": 793 + }, + { + "epoch": 1.45, + "grad_norm": 0.3755910554972886, + "learning_rate": 4.9043262283202974e-05, + "loss": 1.1624, + "step": 794 + }, + { + "epoch": 1.45, + "grad_norm": 0.3635899120146823, + "learning_rate": 4.8960195626462145e-05, + "loss": 1.2095, + "step": 795 + }, + { + "epoch": 1.46, + "grad_norm": 0.3642202684342816, + "learning_rate": 4.8877088267125664e-05, + "loss": 1.1099, + "step": 796 + }, + { + "epoch": 1.46, + "grad_norm": 0.3339946548799316, + "learning_rate": 4.879394058271712e-05, + "loss": 1.1157, + "step": 797 + }, + { + "epoch": 1.46, + "grad_norm": 0.3457189703100475, + "learning_rate": 4.871075295094329e-05, + "loss": 1.129, + "step": 798 + }, + { + "epoch": 1.46, + "grad_norm": 0.3550931839691424, + "learning_rate": 4.862752574969241e-05, + "loss": 1.076, + "step": 799 + }, + { + "epoch": 1.46, + "grad_norm": 0.36139108917966734, + "learning_rate": 4.8544259357032475e-05, + "loss": 1.1577, + "step": 800 + } + ], + "logging_steps": 1.0, + "max_steps": 1638, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 50, + "total_flos": 829528372936704.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-800/training_args.bin b/checkpoint-800/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..c5d2416a3b70bb5260978ec9996f00154a724ba7 --- /dev/null +++ b/checkpoint-800/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b22e8f9d51a16d03a2c506fa3d1eafa8f4b1ae992992c2086a4d435ffd97387e +size 6712 diff --git a/checkpoint-800/zero_to_fp32.py b/checkpoint-800/zero_to_fp32.py new file mode 100755 index 0000000000000000000000000000000000000000..24cc342e78d1a006c782b3a4cd68d9ce786d8fd8 --- /dev/null +++ b/checkpoint-800/zero_to_fp32.py @@ -0,0 +1,604 @@ +#!/usr/bin/env python + +# Copyright (c) Microsoft Corporation. +# SPDX-License-Identifier: Apache-2.0 + +# DeepSpeed Team + +# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets +# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in +# the future. Once extracted, the weights don't require DeepSpeed and can be used in any +# application. +# +# example: python zero_to_fp32.py . pytorch_model.bin + +import argparse +import torch +import glob +import math +import os +import re +from collections import OrderedDict +from dataclasses import dataclass + +# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with +# DeepSpeed data structures it has to be available in the current python environment. +from deepspeed.utils import logger +from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS, + FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES, + FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS) + + +@dataclass +class zero_model_state: + buffers: dict() + param_shapes: dict() + shared_params: list + ds_version: int + frozen_param_shapes: dict() + frozen_param_fragments: dict() + + +debug = 0 + +# load to cpu +device = torch.device('cpu') + + +def atoi(text): + return int(text) if text.isdigit() else text + + +def natural_keys(text): + ''' + alist.sort(key=natural_keys) sorts in human order + http://nedbatchelder.com/blog/200712/human_sorting.html + (See Toothy's implementation in the comments) + ''' + return [atoi(c) for c in re.split(r'(\d+)', text)] + + +def get_model_state_file(checkpoint_dir, zero_stage): + if not os.path.isdir(checkpoint_dir): + raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist") + + # there should be only one file + if zero_stage <= 2: + file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt") + elif zero_stage == 3: + file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt") + + if not os.path.exists(file): + raise FileNotFoundError(f"can't find model states file at '{file}'") + + return file + + +def get_checkpoint_files(checkpoint_dir, glob_pattern): + # XXX: need to test that this simple glob rule works for multi-node setup too + ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys) + + if len(ckpt_files) == 0: + raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'") + + return ckpt_files + + +def get_optim_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt") + + +def get_model_state_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_model_states.pt") + + +def parse_model_states(files): + zero_model_states = [] + for file in files: + state_dict = torch.load(file, map_location=device) + + if BUFFER_NAMES not in state_dict: + raise ValueError(f"{file} is not a model state checkpoint") + buffer_names = state_dict[BUFFER_NAMES] + if debug: + print("Found buffers:", buffer_names) + + # recover just the buffers while restoring them to fp32 if they were saved in fp16 + buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names} + param_shapes = state_dict[PARAM_SHAPES] + + # collect parameters that are included in param_shapes + param_names = [] + for s in param_shapes: + for name in s.keys(): + param_names.append(name) + + # update with frozen parameters + frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None) + if frozen_param_shapes is not None: + if debug: + print(f"Found frozen_param_shapes: {frozen_param_shapes}") + param_names += list(frozen_param_shapes.keys()) + + # handle shared params + shared_params = [[k, v] for k, v in state_dict["shared_params"].items()] + + ds_version = state_dict.get(DS_VERSION, None) + + frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None) + + z_model_state = zero_model_state(buffers=buffers, + param_shapes=param_shapes, + shared_params=shared_params, + ds_version=ds_version, + frozen_param_shapes=frozen_param_shapes, + frozen_param_fragments=frozen_param_fragments) + zero_model_states.append(z_model_state) + + return zero_model_states + + +def parse_optim_states(files, ds_checkpoint_dir): + + total_files = len(files) + state_dicts = [] + for f in files: + state_dict = torch.load(f, map_location=device) + # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights + # and also handle the case where it was already removed by another helper script + state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None) + state_dicts.append(state_dict) + + if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]: + raise ValueError(f"{files[0]} is not a zero checkpoint") + zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE] + world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT] + + # For ZeRO-2 each param group can have different partition_count as data parallelism for expert + # parameters can be different from data parallelism for non-expert parameters. So we can just + # use the max of the partition_count to get the dp world_size. + + if type(world_size) is list: + world_size = max(world_size) + + if world_size != total_files: + raise ValueError( + f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. " + "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes." + ) + + # the groups are named differently in each stage + if zero_stage <= 2: + fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS + elif zero_stage == 3: + fp32_groups_key = FP32_FLAT_GROUPS + else: + raise ValueError(f"unknown zero stage {zero_stage}") + + if zero_stage <= 2: + fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))] + elif zero_stage == 3: + # if there is more than one param group, there will be multiple flattened tensors - one + # flattened tensor per group - for simplicity merge them into a single tensor + # + # XXX: could make the script more memory efficient for when there are multiple groups - it + # will require matching the sub-lists of param_shapes for each param group flattened tensor + + fp32_flat_groups = [ + torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts)) + ] + + return zero_stage, world_size, fp32_flat_groups + + +def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters): + """ + Returns fp32 state_dict reconstructed from ds checkpoint + + Args: + - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are) + + """ + print(f"Processing zero checkpoint '{ds_checkpoint_dir}'") + + optim_files = get_optim_files(ds_checkpoint_dir) + zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir) + print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}") + + model_files = get_model_state_files(ds_checkpoint_dir) + + zero_model_states = parse_model_states(model_files) + print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}') + + if zero_stage <= 2: + return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + elif zero_stage == 3: + return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + + +def _zero2_merge_frozen_params(state_dict, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + frozen_param_fragments = zero_model_states[0].frozen_param_fragments + + if debug: + num_elem = sum(s.numel() for s in frozen_param_shapes.values()) + print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in frozen_param_fragments.values()]) + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + state_dict[name] = frozen_param_fragments[name] + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _has_callable(obj, fn): + attr = getattr(obj, fn, None) + return callable(attr) + + +def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + + # Reconstruction protocol: + # + # XXX: document this + + if debug: + for i in range(world_size): + for j in range(len(fp32_flat_groups[0])): + print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}") + + # XXX: memory usage doubles here (zero2) + num_param_groups = len(fp32_flat_groups[0]) + merged_single_partition_of_fp32_groups = [] + for i in range(num_param_groups): + merged_partitions = [sd[i] for sd in fp32_flat_groups] + full_single_fp32_vector = torch.cat(merged_partitions, 0) + merged_single_partition_of_fp32_groups.append(full_single_fp32_vector) + avail_numel = sum( + [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups]) + + if debug: + wanted_params = sum([len(shapes) for shapes in param_shapes]) + wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes]) + # not asserting if there is a mismatch due to possible padding + print(f"Have {avail_numel} numels to process.") + print(f"Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + total_numel = 0 + total_params = 0 + for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups): + offset = 0 + avail_numel = full_single_fp32_vector.numel() + for name, shape in shapes.items(): + + unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape) + total_numel += unpartitioned_numel + total_params += 1 + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape) + offset += unpartitioned_numel + + # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and + # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex + # paddings performed in the code it's almost impossible to predict the exact numbers w/o the + # live optimizer object, so we are checking that the numbers are within the right range + align_to = 2 * world_size + + def zero2_align(x): + return align_to * math.ceil(x / align_to) + + if debug: + print(f"original offset={offset}, avail_numel={avail_numel}") + + offset = zero2_align(offset) + avail_numel = zero2_align(avail_numel) + + if debug: + print(f"aligned offset={offset}, avail_numel={avail_numel}") + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero2_merge_frozen_params(state_dict, zero_model_states) + + _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def zero3_partitioned_param_info(unpartitioned_numel, world_size): + remainder = unpartitioned_numel % world_size + padding_numel = (world_size - remainder) if remainder else 0 + partitioned_numel = math.ceil(unpartitioned_numel / world_size) + return partitioned_numel, padding_numel + + +def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + if debug: + for i in range(world_size): + num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values()) + print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in zero_model_states[0].frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states) + state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape) + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + avail_numel = fp32_flat_groups[0].numel() * world_size + # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each + # param, re-consolidating each param, while dealing with padding if any + + # merge list of dicts, preserving order + param_shapes = {k: v for d in param_shapes for k, v in d.items()} + + if debug: + for i in range(world_size): + print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}") + + wanted_params = len(param_shapes) + wanted_numel = sum(shape.numel() for shape in param_shapes.values()) + # not asserting if there is a mismatch due to possible padding + avail_numel = fp32_flat_groups[0].numel() * world_size + print(f"Trainable params: Have {avail_numel} numels to process.") + print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + offset = 0 + total_numel = 0 + total_params = 0 + for name, shape in param_shapes.items(): + + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + total_params += 1 + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + # XXX: memory usage doubles here + state_dict[name] = torch.cat( + tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)), + 0).narrow(0, 0, unpartitioned_numel).view(shape) + offset += partitioned_numel + + offset *= world_size + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero3_merge_frozen_params(state_dict, world_size, zero_model_states) + + _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with + ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example + via a model hub. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + + Returns: + - pytorch ``state_dict`` + + Note: this approach may not work if your application doesn't have sufficient free CPU memory and + you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with + the checkpoint. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + # do the training and checkpoint saving + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu + model = model.cpu() # move to cpu + model.load_state_dict(state_dict) + # submit to model hub or save the model to share with others + + In this example the ``model`` will no longer be usable in the deepspeed context of the same + application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead. + + """ + if tag is None: + latest_path = os.path.join(checkpoint_dir, 'latest') + if os.path.isfile(latest_path): + with open(latest_path, 'r') as fd: + tag = fd.read().strip() + else: + raise ValueError(f"Unable to find 'latest' file at {latest_path}") + + ds_checkpoint_dir = os.path.join(checkpoint_dir, tag) + + if not os.path.isdir(ds_checkpoint_dir): + raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist") + + return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters) + + +def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be + loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + """ + + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters) + print(f"Saving fp32 state dict to {output_file}") + torch.save(state_dict, output_file) + + +def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None): + """ + 1. Put the provided model to cpu + 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` + 3. Load it into the provided model + + Args: + - ``model``: the model object to update + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + + Returns: + - ``model`: modified model + + Make sure you have plenty of CPU memory available before you call this function. If you don't + have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it + conveniently placed for you in the checkpoint folder. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint + model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir) + # submit to model hub or save the model to share with others + + Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context + of the same application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + """ + logger.info(f"Extracting fp32 weights") + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag) + + logger.info(f"Overwriting model with fp32 weights") + model = model.cpu() + model.load_state_dict(state_dict, strict=False) + + return model + + +if __name__ == "__main__": + + parser = argparse.ArgumentParser() + parser.add_argument("checkpoint_dir", + type=str, + help="path to the desired checkpoint folder, e.g., path/checkpoint-12") + parser.add_argument( + "output_file", + type=str, + help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)") + parser.add_argument("-t", + "--tag", + type=str, + default=None, + help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1") + parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters") + parser.add_argument("-d", "--debug", action='store_true', help="enable debug") + args = parser.parse_args() + + debug = args.debug + + convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, + args.output_file, + tag=args.tag, + exclude_frozen_parameters=args.exclude_frozen_parameters) diff --git a/checkpoint-850/README.md b/checkpoint-850/README.md new file mode 100644 index 0000000000000000000000000000000000000000..16b1eacdd9353dec380a08ee77ce6ed5ab50f12e --- /dev/null +++ b/checkpoint-850/README.md @@ -0,0 +1,202 @@ +--- +library_name: peft +base_model: gotzmann/uni +--- + +# Model Card for Model ID + + + + + +## Model Details + +### Model Description + + + + + +- **Developed by:** [More Information Needed] +- **Funded by [optional]:** [More Information Needed] +- **Shared by [optional]:** [More Information Needed] +- **Model type:** [More Information Needed] +- **Language(s) (NLP):** [More Information Needed] +- **License:** [More Information Needed] +- **Finetuned from model [optional]:** [More Information Needed] + +### Model Sources [optional] + + + +- **Repository:** [More Information Needed] +- **Paper [optional]:** [More Information Needed] +- **Demo [optional]:** [More Information Needed] + +## Uses + + + +### Direct Use + + + +[More Information Needed] + +### Downstream Use [optional] + + + +[More Information Needed] + +### Out-of-Scope Use + + + +[More Information Needed] + +## Bias, Risks, and Limitations + + + +[More Information Needed] + +### Recommendations + + + +Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations. + +## How to Get Started with the Model + +Use the code below to get started with the model. + +[More Information Needed] + +## Training Details + +### Training Data + + + +[More Information Needed] + +### Training Procedure + + + +#### Preprocessing [optional] + +[More Information Needed] + + +#### Training Hyperparameters + +- **Training regime:** [More Information Needed] + +#### Speeds, Sizes, Times [optional] + + + +[More Information Needed] + +## Evaluation + + + +### Testing Data, Factors & Metrics + +#### Testing Data + + + +[More Information Needed] + +#### Factors + + + +[More Information Needed] + +#### Metrics + + + +[More Information Needed] + +### Results + +[More Information Needed] + +#### Summary + + + +## Model Examination [optional] + + + +[More Information Needed] + +## Environmental Impact + + + +Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700). + +- **Hardware Type:** [More Information Needed] +- **Hours used:** [More Information Needed] +- **Cloud Provider:** [More Information Needed] +- **Compute Region:** [More Information Needed] +- **Carbon Emitted:** [More Information Needed] + +## Technical Specifications [optional] + +### Model Architecture and Objective + +[More Information Needed] + +### Compute Infrastructure + +[More Information Needed] + +#### Hardware + +[More Information Needed] + +#### Software + +[More Information Needed] + +## Citation [optional] + + + +**BibTeX:** + +[More Information Needed] + +**APA:** + +[More Information Needed] + +## Glossary [optional] + + + +[More Information Needed] + +## More Information [optional] + +[More Information Needed] + +## Model Card Authors [optional] + +[More Information Needed] + +## Model Card Contact + +[More Information Needed] +### Framework versions + +- PEFT 0.10.0 \ No newline at end of file diff --git a/checkpoint-850/adapter_config.json b/checkpoint-850/adapter_config.json new file mode 100644 index 0000000000000000000000000000000000000000..832188d72d81e59dd2b5259e86f371199b441aca --- /dev/null +++ b/checkpoint-850/adapter_config.json @@ -0,0 +1,31 @@ +{ + "alpha_pattern": {}, + "auto_mapping": null, + "base_model_name_or_path": "gotzmann/uni", + "bias": "none", + "fan_in_fan_out": false, + "inference_mode": true, + "init_lora_weights": true, + "layer_replication": null, + "layers_pattern": null, + "layers_to_transform": null, + "loftq_config": {}, + "lora_alpha": 128, + "lora_dropout": 0.0, + "megatron_config": null, + "megatron_core": "megatron.core", + "modules_to_save": null, + "peft_type": "LORA", + "r": 128, + "rank_pattern": {}, + "revision": null, + "target_modules": [ + "o_proj", + "k_proj", + "q_proj", + "v_proj" + ], + "task_type": "CAUSAL_LM", + "use_dora": false, + "use_rslora": true +} \ No newline at end of file diff --git a/checkpoint-850/adapter_model.safetensors b/checkpoint-850/adapter_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f4477cadfe1a6896201937acdd0279c467421ba7 --- /dev/null +++ b/checkpoint-850/adapter_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0a24db9b562bcecc0a16772cf5beb9986aa18dcec3d9e061d2564a990a2defff +size 1048664848 diff --git a/checkpoint-850/global_step850/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt b/checkpoint-850/global_step850/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..5821fc3be3ff89dc4437b7fdebfcddc2203888e9 --- /dev/null +++ b/checkpoint-850/global_step850/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:b094a344753b2dd9453905dee7ba19ab5579b0fd97f0736748401cdcb725c03f +size 787270042 diff --git a/checkpoint-850/global_step850/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt b/checkpoint-850/global_step850/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..ef6882de0e60be4e0d25f14c77bdd62b034c0cee --- /dev/null +++ b/checkpoint-850/global_step850/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:15512858904ed80ac96751a57ab6d1389b6fa1cf23f083e510effe23f73449d7 +size 787270042 diff --git a/checkpoint-850/global_step850/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt b/checkpoint-850/global_step850/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..5d044be17aaeec22100ef3c56a0b0a0779a1dc2b --- /dev/null +++ b/checkpoint-850/global_step850/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:ce1d8368380f8f11bbb04be5eba4f6fd44e9113dd75241a10f1ea2b9bb769d87 +size 787270042 diff --git a/checkpoint-850/global_step850/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt b/checkpoint-850/global_step850/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..f1d1d4f607353077270163839626297786db503e --- /dev/null +++ b/checkpoint-850/global_step850/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:96f3d36c71dbaa47b45b1414fb351580f28c4a259b4aad5b89d16b2869aa49cd +size 787270042 diff --git a/checkpoint-850/global_step850/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt b/checkpoint-850/global_step850/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..406e99433fa65969d7852fee72b071b1f4061198 --- /dev/null +++ b/checkpoint-850/global_step850/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:996fd2d3d1da74eddc9a3e03696b24aaa98ca32e74df017291e45418e74e2a1d +size 787270042 diff --git a/checkpoint-850/global_step850/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt b/checkpoint-850/global_step850/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..9ead6f116901a184f51c2e45c35538045d1676e0 --- /dev/null +++ b/checkpoint-850/global_step850/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:770f29434ee8dc9d35ad14a8f1dcb202a82b34a3ef164b74d721c539b4bfc572 +size 787270042 diff --git a/checkpoint-850/global_step850/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt b/checkpoint-850/global_step850/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..9c9f2303c2f5d2e2dfe75d08276c8dfcf725bf98 --- /dev/null +++ b/checkpoint-850/global_step850/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4f796c99ad999667f8709b536277d74fb2dfc4b29aec43974e3f82d358707106 +size 787270042 diff --git a/checkpoint-850/global_step850/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt b/checkpoint-850/global_step850/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..66cb5089fc15b2de83ca2f0850468aec1806e3fe --- /dev/null +++ b/checkpoint-850/global_step850/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:b5b595656aec082e53ab6112580cda2d0864a32ef3a04dc92e011ce170ffc134 +size 787270042 diff --git a/checkpoint-850/global_step850/zero_pp_rank_0_mp_rank_00_model_states.pt b/checkpoint-850/global_step850/zero_pp_rank_0_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..4266102da870765feb8c6f4931dd09173cc4a635 --- /dev/null +++ b/checkpoint-850/global_step850/zero_pp_rank_0_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3c4636383f7f0687a09f235b47de5324f37846c740b1f4d010d36adad52ee512 +size 653742 diff --git a/checkpoint-850/global_step850/zero_pp_rank_1_mp_rank_00_model_states.pt b/checkpoint-850/global_step850/zero_pp_rank_1_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..11a7e399cbaeea9b1d0221d94e9a5aeef34ef063 --- /dev/null +++ b/checkpoint-850/global_step850/zero_pp_rank_1_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3059c7b370b27cc5e627a87a08c4b19e7cd3639b7a839379299724a3ab0d9361 +size 653742 diff --git a/checkpoint-850/global_step850/zero_pp_rank_2_mp_rank_00_model_states.pt b/checkpoint-850/global_step850/zero_pp_rank_2_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..eeb138a36fcfd340418b4d32cb974f13b7697694 --- /dev/null +++ b/checkpoint-850/global_step850/zero_pp_rank_2_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c711e98fb08c163203dc87b78751464a86ead4b2fd198f8789cc169707d4777d +size 653742 diff --git a/checkpoint-850/global_step850/zero_pp_rank_3_mp_rank_00_model_states.pt b/checkpoint-850/global_step850/zero_pp_rank_3_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..25032ea96c6490ce3d796172a32126864cdf416d --- /dev/null +++ b/checkpoint-850/global_step850/zero_pp_rank_3_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f780e605edce29c8a4c8cabf71af90b8ab02bc1a9ff4d499a57f7bc8bddf73ae +size 653742 diff --git a/checkpoint-850/global_step850/zero_pp_rank_4_mp_rank_00_model_states.pt b/checkpoint-850/global_step850/zero_pp_rank_4_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..21daac9e3b0da385f2b5d5c010418cf2ac2c471b --- /dev/null +++ b/checkpoint-850/global_step850/zero_pp_rank_4_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7a80187552ff311ab5f3292c94b850ebeab169f3a8accf72fb9d556dd3895e7a +size 653742 diff --git a/checkpoint-850/global_step850/zero_pp_rank_5_mp_rank_00_model_states.pt b/checkpoint-850/global_step850/zero_pp_rank_5_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..11c0875a0ca935cb9c9f02b372af124165e9bbfe --- /dev/null +++ b/checkpoint-850/global_step850/zero_pp_rank_5_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2879bde5625128d1ce9d91bb79a25eb3ea12549772c9df8f8c4c89025fffd9dc +size 653742 diff --git a/checkpoint-850/global_step850/zero_pp_rank_6_mp_rank_00_model_states.pt b/checkpoint-850/global_step850/zero_pp_rank_6_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..7e9d7004a780769cfd567a5bf98afc7f08c661ea --- /dev/null +++ b/checkpoint-850/global_step850/zero_pp_rank_6_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2ad50c5a7d9b587b16f3ae5dfcd43af013ddb103d983b5c5638fc3329d553c91 +size 653742 diff --git a/checkpoint-850/global_step850/zero_pp_rank_7_mp_rank_00_model_states.pt b/checkpoint-850/global_step850/zero_pp_rank_7_mp_rank_00_model_states.pt new file mode 100644 index 0000000000000000000000000000000000000000..e65614fbda6c94141248f18d46b6f71f66a72809 --- /dev/null +++ b/checkpoint-850/global_step850/zero_pp_rank_7_mp_rank_00_model_states.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0fb0b6e0853703d42cf555fd9864d4db4c2b6d04f08be900fe24218069f955a9 +size 653742 diff --git a/checkpoint-850/latest b/checkpoint-850/latest new file mode 100644 index 0000000000000000000000000000000000000000..3691022819d6dd7dd441445e8bf742e36ca808cd --- /dev/null +++ b/checkpoint-850/latest @@ -0,0 +1 @@ +global_step850 \ No newline at end of file diff --git a/checkpoint-850/rng_state_0.pth b/checkpoint-850/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..9dd2a62da4ca83b3b986d96dbf0eaeb82207ca93 --- /dev/null +++ b/checkpoint-850/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a0628a9017696045a3a29e9eaffc71e9262d855716e773c0c3be760a1fe85bc8 +size 15984 diff --git a/checkpoint-850/rng_state_1.pth b/checkpoint-850/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..1ba5f3aba4388a582cd47f7f9e57cd5879b1cbd2 --- /dev/null +++ b/checkpoint-850/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:df342004a4d8e3626bf2a9f689fde7c8bfd6d995e14931f5496eda1f456cb6f2 +size 15984 diff --git a/checkpoint-850/rng_state_2.pth b/checkpoint-850/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..27b0f7845c2b9530c3e6ed3ce232ff4e86b86122 --- /dev/null +++ b/checkpoint-850/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f02096eb4e8850b91490e80e4a042e2e60f71bd2abc6a269d62c271649cb77d2 +size 15984 diff --git a/checkpoint-850/rng_state_3.pth b/checkpoint-850/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..fcfb583fc43c6dd4395671708744cfd18c419970 --- /dev/null +++ b/checkpoint-850/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:326c778d3d0e7e3d5665fa0a9ecd92986609c430da08b41611d6c05dc19815a8 +size 15984 diff --git a/checkpoint-850/rng_state_4.pth b/checkpoint-850/rng_state_4.pth new file mode 100644 index 0000000000000000000000000000000000000000..7a8c64b1f15ac655b2be2a42fe61cabe2a877704 --- /dev/null +++ b/checkpoint-850/rng_state_4.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d978dcb0c34e022ee6750e9d86814b8c82e4965d7e07662f35f06eeac12938f3 +size 15984 diff --git a/checkpoint-850/rng_state_5.pth b/checkpoint-850/rng_state_5.pth new file mode 100644 index 0000000000000000000000000000000000000000..262e8187e6caeca12ef3b0aa923b12afd697e03d --- /dev/null +++ b/checkpoint-850/rng_state_5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:01e83399aed1d9d173c3e07b2efa8530c956b62b2b68394c2ed0d43bd8bba9d1 +size 15984 diff --git a/checkpoint-850/rng_state_6.pth b/checkpoint-850/rng_state_6.pth new file mode 100644 index 0000000000000000000000000000000000000000..72f794e31f8d3e0c63972e5076e1ed90c52087ba --- /dev/null +++ b/checkpoint-850/rng_state_6.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:606ab3ca92e3d20c327c69fdcce7f7e39bec2f2c3538b036088b255f917e3ba4 +size 15984 diff --git a/checkpoint-850/rng_state_7.pth b/checkpoint-850/rng_state_7.pth new file mode 100644 index 0000000000000000000000000000000000000000..244e7fdaa1cef2e82bd4e16afb10f32f68318bcc --- /dev/null +++ b/checkpoint-850/rng_state_7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1276a987dd22c9093fec58921ba19f340a28f18bff635cc01324e09a3c37ac3a +size 15984 diff --git a/checkpoint-850/scheduler.pt b/checkpoint-850/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..58ac1ab9bfba8e8a2e6e6e316e2f5c7c070cb178 --- /dev/null +++ b/checkpoint-850/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3c4b4462e080a5c39faf9317093c20bf1f40a2d57c50836da1f781d634a5c527 +size 1064 diff --git a/checkpoint-850/special_tokens_map.json b/checkpoint-850/special_tokens_map.json new file mode 100644 index 0000000000000000000000000000000000000000..72ecfeeb7e14d244c936169d2ed139eeae235ef1 --- /dev/null +++ b/checkpoint-850/special_tokens_map.json @@ -0,0 +1,24 @@ +{ + "bos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "eos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "pad_token": "", + "unk_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + } +} diff --git a/checkpoint-850/tokenizer.model b/checkpoint-850/tokenizer.model new file mode 100644 index 0000000000000000000000000000000000000000..6c00c742ce03c627d6cd5b795984876fa49fa899 --- /dev/null +++ b/checkpoint-850/tokenizer.model @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347 +size 499723 diff --git a/checkpoint-850/tokenizer_config.json b/checkpoint-850/tokenizer_config.json new file mode 100644 index 0000000000000000000000000000000000000000..bb5a9f09d8c0f3c32c66fc6118fe5c76c5c6fd90 --- /dev/null +++ b/checkpoint-850/tokenizer_config.json @@ -0,0 +1,45 @@ +{ + "add_bos_token": true, + "add_eos_token": false, + "add_prefix_space": false, + "added_tokens_decoder": { + "0": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "1": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "2": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + } + }, + "bos_token": "", + "chat_template": "{% if messages[0]['role'] == 'system' %}{% set system_message = messages[0]['content'] %}{% endif %}{% if system_message is defined %}{{ '' + '### System:\\n\\n' + system_message }}{% endif %}{% for message in messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ '\\n\\n### Human:\\n\\n' + content }}{% elif message['role'] == 'assistant' %}{{ '\\n\\n### Assistant:\\n\\n' + content + '' }}{% endif %}{% endfor %}", + "clean_up_tokenization_spaces": false, + "eos_token": "", + "legacy": false, + "model_max_length": 1000000000000000019884624838656, + "pad_token": "", + "padding_side": "right", + "sp_model_kwargs": {}, + "spaces_between_special_tokens": false, + "split_special_tokens": false, + "tokenizer_class": "LlamaTokenizer", + "unk_token": "", + "use_default_system_prompt": false +} diff --git a/checkpoint-850/trainer_state.json b/checkpoint-850/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..00833dcaff44143877e57dd222998795b26c9cd7 --- /dev/null +++ b/checkpoint-850/trainer_state.json @@ -0,0 +1,5971 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 1.0914494741655236, + "eval_steps": 500, + "global_step": 850, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "grad_norm": 0.849355824164473, + "learning_rate": 4.878048780487805e-07, + "loss": 1.3655, + "step": 1 + }, + { + "epoch": 0.0, + "grad_norm": 10.01567518957158, + "learning_rate": 9.75609756097561e-07, + "loss": 1.5767, + "step": 2 + }, + { + "epoch": 0.01, + "grad_norm": 0.6466000875559635, + "learning_rate": 1.4634146341463414e-06, + "loss": 1.3913, + "step": 3 + }, + { + "epoch": 0.01, + "grad_norm": 0.6644565932010504, + "learning_rate": 1.951219512195122e-06, + "loss": 1.3218, + "step": 4 + }, + { + "epoch": 0.01, + "grad_norm": 0.571354207588475, + "learning_rate": 2.4390243902439027e-06, + "loss": 1.3597, + "step": 5 + }, + { + "epoch": 0.01, + "grad_norm": 0.31036262839244955, + "learning_rate": 2.926829268292683e-06, + "loss": 1.2832, + "step": 6 + }, + { + "epoch": 0.01, + "grad_norm": 0.2622135027188184, + "learning_rate": 3.414634146341464e-06, + "loss": 1.2161, + "step": 7 + }, + { + "epoch": 0.01, + "grad_norm": 0.296824630261661, + "learning_rate": 3.902439024390244e-06, + "loss": 1.2985, + "step": 8 + }, + { + "epoch": 0.02, + "grad_norm": 0.2557267467361569, + "learning_rate": 4.390243902439025e-06, + "loss": 1.3175, + "step": 9 + }, + { + "epoch": 0.02, + "grad_norm": 0.23418939513890769, + "learning_rate": 4.8780487804878055e-06, + "loss": 1.2617, + "step": 10 + }, + { + "epoch": 0.02, + "grad_norm": 0.2364760983285843, + "learning_rate": 5.365853658536586e-06, + "loss": 1.3103, + "step": 11 + }, + { + "epoch": 0.02, + "grad_norm": 0.23893034721889, + "learning_rate": 5.853658536585366e-06, + "loss": 1.2405, + "step": 12 + }, + { + "epoch": 0.02, + "grad_norm": 0.25563593295485887, + "learning_rate": 6.341463414634147e-06, + "loss": 1.2831, + "step": 13 + }, + { + "epoch": 0.03, + "grad_norm": 0.23239975352661665, + "learning_rate": 6.829268292682928e-06, + "loss": 1.3125, + "step": 14 + }, + { + "epoch": 0.03, + "grad_norm": 0.3092813858209507, + "learning_rate": 7.317073170731707e-06, + "loss": 1.2422, + "step": 15 + }, + { + "epoch": 0.03, + "grad_norm": 0.282563380367434, + "learning_rate": 7.804878048780489e-06, + "loss": 1.2453, + "step": 16 + }, + { + "epoch": 0.03, + "grad_norm": 0.22065680088315018, + "learning_rate": 8.292682926829268e-06, + "loss": 1.2491, + "step": 17 + }, + { + "epoch": 0.03, + "grad_norm": 0.22777800877980184, + "learning_rate": 8.78048780487805e-06, + "loss": 1.2655, + "step": 18 + }, + { + "epoch": 0.03, + "grad_norm": 0.22145212540177928, + "learning_rate": 9.268292682926831e-06, + "loss": 1.2413, + "step": 19 + }, + { + "epoch": 0.04, + "grad_norm": 0.22482351883112714, + "learning_rate": 9.756097560975611e-06, + "loss": 1.2653, + "step": 20 + }, + { + "epoch": 0.04, + "grad_norm": 0.20823080508385733, + "learning_rate": 1.024390243902439e-05, + "loss": 1.2374, + "step": 21 + }, + { + "epoch": 0.04, + "grad_norm": 0.26025492562935737, + "learning_rate": 1.0731707317073172e-05, + "loss": 1.2065, + "step": 22 + }, + { + "epoch": 0.04, + "grad_norm": 0.2150252124176173, + "learning_rate": 1.1219512195121953e-05, + "loss": 1.2782, + "step": 23 + }, + { + "epoch": 0.04, + "grad_norm": 0.2505915177425618, + "learning_rate": 1.1707317073170731e-05, + "loss": 1.2742, + "step": 24 + }, + { + "epoch": 0.05, + "grad_norm": 0.20129223044786942, + "learning_rate": 1.2195121951219513e-05, + "loss": 1.3366, + "step": 25 + }, + { + "epoch": 0.05, + "grad_norm": 0.1973508510397107, + "learning_rate": 1.2682926829268294e-05, + "loss": 1.2476, + "step": 26 + }, + { + "epoch": 0.05, + "grad_norm": 0.27103325392437194, + "learning_rate": 1.3170731707317076e-05, + "loss": 1.2325, + "step": 27 + }, + { + "epoch": 0.05, + "grad_norm": 0.17954976411006285, + "learning_rate": 1.3658536585365855e-05, + "loss": 1.2523, + "step": 28 + }, + { + "epoch": 0.05, + "grad_norm": 0.22216997851088888, + "learning_rate": 1.4146341463414635e-05, + "loss": 1.3297, + "step": 29 + }, + { + "epoch": 0.05, + "grad_norm": 0.2071458864548587, + "learning_rate": 1.4634146341463415e-05, + "loss": 1.2127, + "step": 30 + }, + { + "epoch": 0.06, + "grad_norm": 0.18039422081622164, + "learning_rate": 1.5121951219512196e-05, + "loss": 1.2509, + "step": 31 + }, + { + "epoch": 0.06, + "grad_norm": 0.18631254372974412, + "learning_rate": 1.5609756097560978e-05, + "loss": 1.2247, + "step": 32 + }, + { + "epoch": 0.06, + "grad_norm": 0.18843872523649827, + "learning_rate": 1.6097560975609757e-05, + "loss": 1.195, + "step": 33 + }, + { + "epoch": 0.06, + "grad_norm": 0.2163847267778325, + "learning_rate": 1.6585365853658537e-05, + "loss": 1.2179, + "step": 34 + }, + { + "epoch": 0.06, + "grad_norm": 0.19687688475496104, + "learning_rate": 1.7073170731707317e-05, + "loss": 1.2763, + "step": 35 + }, + { + "epoch": 0.07, + "grad_norm": 0.20409643064887947, + "learning_rate": 1.75609756097561e-05, + "loss": 1.253, + "step": 36 + }, + { + "epoch": 0.07, + "grad_norm": 0.1879182661759335, + "learning_rate": 1.804878048780488e-05, + "loss": 1.2586, + "step": 37 + }, + { + "epoch": 0.07, + "grad_norm": 0.19400648948514373, + "learning_rate": 1.8536585365853663e-05, + "loss": 1.2154, + "step": 38 + }, + { + "epoch": 0.07, + "grad_norm": 0.1878879343148452, + "learning_rate": 1.902439024390244e-05, + "loss": 1.2304, + "step": 39 + }, + { + "epoch": 0.07, + "grad_norm": 0.17687475469924052, + "learning_rate": 1.9512195121951222e-05, + "loss": 1.2351, + "step": 40 + }, + { + "epoch": 0.07, + "grad_norm": 0.18223935625384885, + "learning_rate": 2e-05, + "loss": 1.2222, + "step": 41 + }, + { + "epoch": 0.08, + "grad_norm": 0.1943061629408338, + "learning_rate": 2.048780487804878e-05, + "loss": 1.2044, + "step": 42 + }, + { + "epoch": 0.08, + "grad_norm": 0.17027514338700078, + "learning_rate": 2.0975609756097564e-05, + "loss": 1.1548, + "step": 43 + }, + { + "epoch": 0.08, + "grad_norm": 0.18553769630586192, + "learning_rate": 2.1463414634146344e-05, + "loss": 1.2721, + "step": 44 + }, + { + "epoch": 0.08, + "grad_norm": 0.19732826914228765, + "learning_rate": 2.1951219512195124e-05, + "loss": 1.3097, + "step": 45 + }, + { + "epoch": 0.08, + "grad_norm": 0.18714230986631472, + "learning_rate": 2.2439024390243907e-05, + "loss": 1.2662, + "step": 46 + }, + { + "epoch": 0.09, + "grad_norm": 0.19988987568002223, + "learning_rate": 2.2926829268292683e-05, + "loss": 1.2904, + "step": 47 + }, + { + "epoch": 0.09, + "grad_norm": 0.17744650133390918, + "learning_rate": 2.3414634146341463e-05, + "loss": 1.1825, + "step": 48 + }, + { + "epoch": 0.09, + "grad_norm": 0.16576734763834533, + "learning_rate": 2.3902439024390246e-05, + "loss": 1.1858, + "step": 49 + }, + { + "epoch": 0.09, + "grad_norm": 0.179591794065527, + "learning_rate": 2.4390243902439026e-05, + "loss": 1.2711, + "step": 50 + }, + { + "epoch": 0.09, + "grad_norm": 0.17923464471176911, + "learning_rate": 2.4878048780487805e-05, + "loss": 1.2289, + "step": 51 + }, + { + "epoch": 0.1, + "grad_norm": 0.18991742907836837, + "learning_rate": 2.536585365853659e-05, + "loss": 1.3097, + "step": 52 + }, + { + "epoch": 0.1, + "grad_norm": 0.19849796137254636, + "learning_rate": 2.5853658536585368e-05, + "loss": 1.2489, + "step": 53 + }, + { + "epoch": 0.1, + "grad_norm": 0.17452371110976383, + "learning_rate": 2.634146341463415e-05, + "loss": 1.2461, + "step": 54 + }, + { + "epoch": 0.1, + "grad_norm": 0.17671022353085036, + "learning_rate": 2.682926829268293e-05, + "loss": 1.153, + "step": 55 + }, + { + "epoch": 0.1, + "grad_norm": 0.36820559192096686, + "learning_rate": 2.731707317073171e-05, + "loss": 1.2431, + "step": 56 + }, + { + "epoch": 0.1, + "grad_norm": 0.20331468526494198, + "learning_rate": 2.7804878048780487e-05, + "loss": 1.2575, + "step": 57 + }, + { + "epoch": 0.11, + "grad_norm": 0.2402486598118377, + "learning_rate": 2.829268292682927e-05, + "loss": 1.2538, + "step": 58 + }, + { + "epoch": 0.11, + "grad_norm": 0.2549409484173144, + "learning_rate": 2.878048780487805e-05, + "loss": 1.2065, + "step": 59 + }, + { + "epoch": 0.11, + "grad_norm": 0.2053105349872685, + "learning_rate": 2.926829268292683e-05, + "loss": 1.2094, + "step": 60 + }, + { + "epoch": 0.11, + "grad_norm": 0.17971910872957886, + "learning_rate": 2.9756097560975613e-05, + "loss": 1.228, + "step": 61 + }, + { + "epoch": 0.11, + "grad_norm": 0.1885853654992973, + "learning_rate": 3.0243902439024392e-05, + "loss": 1.2286, + "step": 62 + }, + { + "epoch": 0.12, + "grad_norm": 0.1848524571968613, + "learning_rate": 3.073170731707317e-05, + "loss": 1.2718, + "step": 63 + }, + { + "epoch": 0.12, + "grad_norm": 0.18734105883548513, + "learning_rate": 3.1219512195121955e-05, + "loss": 1.2357, + "step": 64 + }, + { + "epoch": 0.12, + "grad_norm": 0.17774668052121825, + "learning_rate": 3.170731707317074e-05, + "loss": 1.1509, + "step": 65 + }, + { + "epoch": 0.12, + "grad_norm": 0.17890968008080646, + "learning_rate": 3.2195121951219514e-05, + "loss": 1.1924, + "step": 66 + }, + { + "epoch": 0.12, + "grad_norm": 0.18249273371332375, + "learning_rate": 3.268292682926829e-05, + "loss": 1.2545, + "step": 67 + }, + { + "epoch": 0.12, + "grad_norm": 0.21064122671902577, + "learning_rate": 3.3170731707317074e-05, + "loss": 1.2832, + "step": 68 + }, + { + "epoch": 0.13, + "grad_norm": 0.1820064171955093, + "learning_rate": 3.365853658536586e-05, + "loss": 1.2071, + "step": 69 + }, + { + "epoch": 0.13, + "grad_norm": 0.16996662800553433, + "learning_rate": 3.414634146341463e-05, + "loss": 1.2073, + "step": 70 + }, + { + "epoch": 0.13, + "grad_norm": 0.1618669302922445, + "learning_rate": 3.4634146341463416e-05, + "loss": 1.1289, + "step": 71 + }, + { + "epoch": 0.13, + "grad_norm": 0.18948744950985544, + "learning_rate": 3.51219512195122e-05, + "loss": 1.2915, + "step": 72 + }, + { + "epoch": 0.13, + "grad_norm": 0.18326143691603383, + "learning_rate": 3.5609756097560976e-05, + "loss": 1.2238, + "step": 73 + }, + { + "epoch": 0.14, + "grad_norm": 0.17410704510700503, + "learning_rate": 3.609756097560976e-05, + "loss": 1.1784, + "step": 74 + }, + { + "epoch": 0.14, + "grad_norm": 0.1983667344995625, + "learning_rate": 3.658536585365854e-05, + "loss": 1.2452, + "step": 75 + }, + { + "epoch": 0.14, + "grad_norm": 0.3416310763369357, + "learning_rate": 3.7073170731707325e-05, + "loss": 1.1972, + "step": 76 + }, + { + "epoch": 0.14, + "grad_norm": 0.2776466983511955, + "learning_rate": 3.75609756097561e-05, + "loss": 1.3121, + "step": 77 + }, + { + "epoch": 0.14, + "grad_norm": 0.20026129636576834, + "learning_rate": 3.804878048780488e-05, + "loss": 1.2436, + "step": 78 + }, + { + "epoch": 0.14, + "grad_norm": 0.21064549243917835, + "learning_rate": 3.853658536585366e-05, + "loss": 1.2064, + "step": 79 + }, + { + "epoch": 0.15, + "grad_norm": 0.22119482175714267, + "learning_rate": 3.9024390243902444e-05, + "loss": 1.2715, + "step": 80 + }, + { + "epoch": 0.15, + "grad_norm": 0.23047133748844142, + "learning_rate": 3.951219512195122e-05, + "loss": 1.2888, + "step": 81 + }, + { + "epoch": 0.15, + "grad_norm": 0.18741863156973176, + "learning_rate": 4e-05, + "loss": 1.248, + "step": 82 + }, + { + "epoch": 0.15, + "grad_norm": 0.1747859810629604, + "learning_rate": 4.0487804878048786e-05, + "loss": 1.1683, + "step": 83 + }, + { + "epoch": 0.15, + "grad_norm": 0.1896944798413341, + "learning_rate": 4.097560975609756e-05, + "loss": 1.2155, + "step": 84 + }, + { + "epoch": 0.16, + "grad_norm": 0.18724128114363303, + "learning_rate": 4.1463414634146346e-05, + "loss": 1.2273, + "step": 85 + }, + { + "epoch": 0.16, + "grad_norm": 0.17368125504855478, + "learning_rate": 4.195121951219513e-05, + "loss": 1.224, + "step": 86 + }, + { + "epoch": 0.16, + "grad_norm": 0.18371141013625703, + "learning_rate": 4.2439024390243905e-05, + "loss": 1.2294, + "step": 87 + }, + { + "epoch": 0.16, + "grad_norm": 0.1791029365673714, + "learning_rate": 4.292682926829269e-05, + "loss": 1.2895, + "step": 88 + }, + { + "epoch": 0.16, + "grad_norm": 0.20259974283859655, + "learning_rate": 4.341463414634147e-05, + "loss": 1.1841, + "step": 89 + }, + { + "epoch": 0.16, + "grad_norm": 0.17457456183272174, + "learning_rate": 4.390243902439025e-05, + "loss": 1.2357, + "step": 90 + }, + { + "epoch": 0.17, + "grad_norm": 0.1815824380789748, + "learning_rate": 4.439024390243903e-05, + "loss": 1.2304, + "step": 91 + }, + { + "epoch": 0.17, + "grad_norm": 0.17566480599583392, + "learning_rate": 4.4878048780487814e-05, + "loss": 1.242, + "step": 92 + }, + { + "epoch": 0.17, + "grad_norm": 0.18422975005984474, + "learning_rate": 4.536585365853658e-05, + "loss": 1.2177, + "step": 93 + }, + { + "epoch": 0.17, + "grad_norm": 0.16796781877940678, + "learning_rate": 4.5853658536585366e-05, + "loss": 1.1482, + "step": 94 + }, + { + "epoch": 0.17, + "grad_norm": 0.18636131653783305, + "learning_rate": 4.634146341463415e-05, + "loss": 1.1758, + "step": 95 + }, + { + "epoch": 0.18, + "grad_norm": 0.1823665700289814, + "learning_rate": 4.6829268292682926e-05, + "loss": 1.289, + "step": 96 + }, + { + "epoch": 0.18, + "grad_norm": 0.1719900691262439, + "learning_rate": 4.731707317073171e-05, + "loss": 1.1626, + "step": 97 + }, + { + "epoch": 0.18, + "grad_norm": 0.17937994168039778, + "learning_rate": 4.780487804878049e-05, + "loss": 1.175, + "step": 98 + }, + { + "epoch": 0.18, + "grad_norm": 0.16631851422106986, + "learning_rate": 4.829268292682927e-05, + "loss": 1.2177, + "step": 99 + }, + { + "epoch": 0.18, + "grad_norm": 0.19143696232800309, + "learning_rate": 4.878048780487805e-05, + "loss": 1.3071, + "step": 100 + }, + { + "epoch": 0.18, + "grad_norm": 0.17859506638780318, + "learning_rate": 4.9268292682926835e-05, + "loss": 1.2351, + "step": 101 + }, + { + "epoch": 0.19, + "grad_norm": 0.18381520321248196, + "learning_rate": 4.975609756097561e-05, + "loss": 1.2342, + "step": 102 + }, + { + "epoch": 0.19, + "grad_norm": 0.17968218683773912, + "learning_rate": 5.0243902439024394e-05, + "loss": 1.2074, + "step": 103 + }, + { + "epoch": 0.19, + "grad_norm": 0.18139489969339018, + "learning_rate": 5.073170731707318e-05, + "loss": 1.1558, + "step": 104 + }, + { + "epoch": 0.19, + "grad_norm": 0.17366624842514394, + "learning_rate": 5.121951219512195e-05, + "loss": 1.1897, + "step": 105 + }, + { + "epoch": 0.19, + "grad_norm": 0.16034845455223745, + "learning_rate": 5.1707317073170736e-05, + "loss": 1.179, + "step": 106 + }, + { + "epoch": 0.2, + "grad_norm": 0.17583069577827776, + "learning_rate": 5.219512195121952e-05, + "loss": 1.1856, + "step": 107 + }, + { + "epoch": 0.2, + "grad_norm": 0.1853758076989552, + "learning_rate": 5.26829268292683e-05, + "loss": 1.2072, + "step": 108 + }, + { + "epoch": 0.2, + "grad_norm": 0.19597443965936462, + "learning_rate": 5.317073170731708e-05, + "loss": 1.2271, + "step": 109 + }, + { + "epoch": 0.2, + "grad_norm": 0.1899206334098331, + "learning_rate": 5.365853658536586e-05, + "loss": 1.1961, + "step": 110 + }, + { + "epoch": 0.2, + "grad_norm": 0.17463763837757018, + "learning_rate": 5.4146341463414645e-05, + "loss": 1.2049, + "step": 111 + }, + { + "epoch": 0.2, + "grad_norm": 0.20431371701229986, + "learning_rate": 5.463414634146342e-05, + "loss": 1.2891, + "step": 112 + }, + { + "epoch": 0.21, + "grad_norm": 0.1814475107638498, + "learning_rate": 5.51219512195122e-05, + "loss": 1.2346, + "step": 113 + }, + { + "epoch": 0.21, + "grad_norm": 0.1883849423207823, + "learning_rate": 5.5609756097560974e-05, + "loss": 1.244, + "step": 114 + }, + { + "epoch": 0.21, + "grad_norm": 0.1857258128640568, + "learning_rate": 5.609756097560976e-05, + "loss": 1.2669, + "step": 115 + }, + { + "epoch": 0.21, + "grad_norm": 0.1740768514118401, + "learning_rate": 5.658536585365854e-05, + "loss": 1.2414, + "step": 116 + }, + { + "epoch": 0.21, + "grad_norm": 0.1919320335584178, + "learning_rate": 5.7073170731707317e-05, + "loss": 1.2886, + "step": 117 + }, + { + "epoch": 0.22, + "grad_norm": 0.18288775167828136, + "learning_rate": 5.75609756097561e-05, + "loss": 1.1875, + "step": 118 + }, + { + "epoch": 0.22, + "grad_norm": 0.18208588867750863, + "learning_rate": 5.804878048780488e-05, + "loss": 1.2388, + "step": 119 + }, + { + "epoch": 0.22, + "grad_norm": 0.1743260015658331, + "learning_rate": 5.853658536585366e-05, + "loss": 1.1762, + "step": 120 + }, + { + "epoch": 0.22, + "grad_norm": 0.17856046291517946, + "learning_rate": 5.902439024390244e-05, + "loss": 1.2888, + "step": 121 + }, + { + "epoch": 0.22, + "grad_norm": 0.17493794870966536, + "learning_rate": 5.9512195121951225e-05, + "loss": 1.2222, + "step": 122 + }, + { + "epoch": 0.22, + "grad_norm": 0.1909202655203384, + "learning_rate": 6.000000000000001e-05, + "loss": 1.2414, + "step": 123 + }, + { + "epoch": 0.23, + "grad_norm": 0.18345819482834988, + "learning_rate": 6.0487804878048785e-05, + "loss": 1.2756, + "step": 124 + }, + { + "epoch": 0.23, + "grad_norm": 0.2057069352956621, + "learning_rate": 6.097560975609757e-05, + "loss": 1.261, + "step": 125 + }, + { + "epoch": 0.23, + "grad_norm": 0.299775882469108, + "learning_rate": 6.146341463414634e-05, + "loss": 1.2566, + "step": 126 + }, + { + "epoch": 0.23, + "grad_norm": 0.1869687633018095, + "learning_rate": 6.195121951219513e-05, + "loss": 1.3039, + "step": 127 + }, + { + "epoch": 0.23, + "grad_norm": 0.17747149926197442, + "learning_rate": 6.243902439024391e-05, + "loss": 1.2524, + "step": 128 + }, + { + "epoch": 0.24, + "grad_norm": 0.17885157788044242, + "learning_rate": 6.29268292682927e-05, + "loss": 1.2455, + "step": 129 + }, + { + "epoch": 0.24, + "grad_norm": 0.17617298187845123, + "learning_rate": 6.341463414634148e-05, + "loss": 1.2009, + "step": 130 + }, + { + "epoch": 0.24, + "grad_norm": 0.20164176323497066, + "learning_rate": 6.390243902439025e-05, + "loss": 1.2634, + "step": 131 + }, + { + "epoch": 0.24, + "grad_norm": 0.20459903417307612, + "learning_rate": 6.439024390243903e-05, + "loss": 1.1963, + "step": 132 + }, + { + "epoch": 0.24, + "grad_norm": 0.1863755486334296, + "learning_rate": 6.487804878048781e-05, + "loss": 1.2387, + "step": 133 + }, + { + "epoch": 0.25, + "grad_norm": 0.19265866140295207, + "learning_rate": 6.536585365853658e-05, + "loss": 1.2688, + "step": 134 + }, + { + "epoch": 0.25, + "grad_norm": 0.1823425868969493, + "learning_rate": 6.585365853658536e-05, + "loss": 1.2041, + "step": 135 + }, + { + "epoch": 0.25, + "grad_norm": 0.2016853266472781, + "learning_rate": 6.634146341463415e-05, + "loss": 1.1223, + "step": 136 + }, + { + "epoch": 0.25, + "grad_norm": 0.17282675192463448, + "learning_rate": 6.682926829268293e-05, + "loss": 1.1879, + "step": 137 + }, + { + "epoch": 0.25, + "grad_norm": 0.17398811693399288, + "learning_rate": 6.731707317073171e-05, + "loss": 1.2682, + "step": 138 + }, + { + "epoch": 0.25, + "grad_norm": 0.18516916965434696, + "learning_rate": 6.78048780487805e-05, + "loss": 1.1666, + "step": 139 + }, + { + "epoch": 0.26, + "grad_norm": 0.1852213129647933, + "learning_rate": 6.829268292682927e-05, + "loss": 1.2501, + "step": 140 + }, + { + "epoch": 0.26, + "grad_norm": 0.17915948766591883, + "learning_rate": 6.878048780487805e-05, + "loss": 1.2264, + "step": 141 + }, + { + "epoch": 0.26, + "grad_norm": 0.21599939417233183, + "learning_rate": 6.926829268292683e-05, + "loss": 1.2376, + "step": 142 + }, + { + "epoch": 0.26, + "grad_norm": 0.17839304459521851, + "learning_rate": 6.975609756097562e-05, + "loss": 1.2353, + "step": 143 + }, + { + "epoch": 0.26, + "grad_norm": 0.20826913231380875, + "learning_rate": 7.02439024390244e-05, + "loss": 1.1901, + "step": 144 + }, + { + "epoch": 0.27, + "grad_norm": 0.20788894913361589, + "learning_rate": 7.073170731707318e-05, + "loss": 1.2577, + "step": 145 + }, + { + "epoch": 0.27, + "grad_norm": 0.18420055842301297, + "learning_rate": 7.121951219512195e-05, + "loss": 1.1393, + "step": 146 + }, + { + "epoch": 0.27, + "grad_norm": 0.19903048468685589, + "learning_rate": 7.170731707317073e-05, + "loss": 1.2321, + "step": 147 + }, + { + "epoch": 0.27, + "grad_norm": 0.19074116314985748, + "learning_rate": 7.219512195121952e-05, + "loss": 1.1912, + "step": 148 + }, + { + "epoch": 0.27, + "grad_norm": 0.2353816469403903, + "learning_rate": 7.26829268292683e-05, + "loss": 1.28, + "step": 149 + }, + { + "epoch": 0.27, + "grad_norm": 0.21634875684769345, + "learning_rate": 7.317073170731708e-05, + "loss": 1.3312, + "step": 150 + }, + { + "epoch": 0.28, + "grad_norm": 0.18290969006743918, + "learning_rate": 7.365853658536587e-05, + "loss": 1.2214, + "step": 151 + }, + { + "epoch": 0.28, + "grad_norm": 0.18484243897545208, + "learning_rate": 7.414634146341465e-05, + "loss": 1.1895, + "step": 152 + }, + { + "epoch": 0.28, + "grad_norm": 0.21882343112978872, + "learning_rate": 7.463414634146342e-05, + "loss": 1.2219, + "step": 153 + }, + { + "epoch": 0.28, + "grad_norm": 0.19868284379241205, + "learning_rate": 7.51219512195122e-05, + "loss": 1.2176, + "step": 154 + }, + { + "epoch": 0.28, + "grad_norm": 0.20912516312950613, + "learning_rate": 7.560975609756097e-05, + "loss": 1.242, + "step": 155 + }, + { + "epoch": 0.29, + "grad_norm": 0.23811880045549916, + "learning_rate": 7.609756097560976e-05, + "loss": 1.2838, + "step": 156 + }, + { + "epoch": 0.29, + "grad_norm": 0.19511077122033713, + "learning_rate": 7.658536585365854e-05, + "loss": 1.1594, + "step": 157 + }, + { + "epoch": 0.29, + "grad_norm": 0.20094129399534238, + "learning_rate": 7.707317073170732e-05, + "loss": 1.2966, + "step": 158 + }, + { + "epoch": 0.29, + "grad_norm": 0.19366245038292418, + "learning_rate": 7.75609756097561e-05, + "loss": 1.2246, + "step": 159 + }, + { + "epoch": 0.29, + "grad_norm": 0.19409570223867306, + "learning_rate": 7.804878048780489e-05, + "loss": 1.2312, + "step": 160 + }, + { + "epoch": 0.29, + "grad_norm": 0.2087258457033805, + "learning_rate": 7.853658536585366e-05, + "loss": 1.2169, + "step": 161 + }, + { + "epoch": 0.3, + "grad_norm": 0.18765223996270428, + "learning_rate": 7.902439024390244e-05, + "loss": 1.2383, + "step": 162 + }, + { + "epoch": 0.3, + "grad_norm": 0.20734180224147242, + "learning_rate": 7.951219512195122e-05, + "loss": 1.2587, + "step": 163 + }, + { + "epoch": 0.3, + "grad_norm": 0.24690929540287834, + "learning_rate": 8e-05, + "loss": 1.1951, + "step": 164 + }, + { + "epoch": 0.3, + "grad_norm": 0.2003538797619543, + "learning_rate": 7.999990914797545e-05, + "loss": 1.1982, + "step": 165 + }, + { + "epoch": 0.3, + "grad_norm": 0.22469075613510484, + "learning_rate": 7.99996365923145e-05, + "loss": 1.2355, + "step": 166 + }, + { + "epoch": 0.31, + "grad_norm": 0.21870100788336058, + "learning_rate": 7.999918233425526e-05, + "loss": 1.1103, + "step": 167 + }, + { + "epoch": 0.31, + "grad_norm": 0.20939989594131886, + "learning_rate": 7.999854637586122e-05, + "loss": 1.1966, + "step": 168 + }, + { + "epoch": 0.31, + "grad_norm": 0.43108211416237796, + "learning_rate": 7.999772872002132e-05, + "loss": 1.2882, + "step": 169 + }, + { + "epoch": 0.31, + "grad_norm": 0.27045413432174487, + "learning_rate": 7.999672937044984e-05, + "loss": 1.2399, + "step": 170 + }, + { + "epoch": 0.31, + "grad_norm": 0.19700483036740515, + "learning_rate": 7.999554833168642e-05, + "loss": 1.202, + "step": 171 + }, + { + "epoch": 0.31, + "grad_norm": 0.3335979493370708, + "learning_rate": 7.999418560909604e-05, + "loss": 1.1995, + "step": 172 + }, + { + "epoch": 0.32, + "grad_norm": 0.3165803974474567, + "learning_rate": 7.999264120886902e-05, + "loss": 1.1569, + "step": 173 + }, + { + "epoch": 0.32, + "grad_norm": 0.1951699080346223, + "learning_rate": 7.999091513802093e-05, + "loss": 1.1778, + "step": 174 + }, + { + "epoch": 0.32, + "grad_norm": 0.2087559121749787, + "learning_rate": 7.998900740439265e-05, + "loss": 1.1736, + "step": 175 + }, + { + "epoch": 0.32, + "grad_norm": 0.20345180977460478, + "learning_rate": 7.998691801665024e-05, + "loss": 1.2281, + "step": 176 + }, + { + "epoch": 0.32, + "grad_norm": 0.24617644827252333, + "learning_rate": 7.998464698428495e-05, + "loss": 1.2072, + "step": 177 + }, + { + "epoch": 0.33, + "grad_norm": 0.2469050959356265, + "learning_rate": 7.998219431761318e-05, + "loss": 1.2242, + "step": 178 + }, + { + "epoch": 0.33, + "grad_norm": 0.19529317748460623, + "learning_rate": 7.997956002777642e-05, + "loss": 1.2567, + "step": 179 + }, + { + "epoch": 0.33, + "grad_norm": 0.19048389491381376, + "learning_rate": 7.99767441267412e-05, + "loss": 1.2982, + "step": 180 + }, + { + "epoch": 0.33, + "grad_norm": 0.2085799116493225, + "learning_rate": 7.997374662729904e-05, + "loss": 1.1254, + "step": 181 + }, + { + "epoch": 0.33, + "grad_norm": 0.20636853256378995, + "learning_rate": 7.997056754306636e-05, + "loss": 1.2435, + "step": 182 + }, + { + "epoch": 0.33, + "grad_norm": 0.20590016382290252, + "learning_rate": 7.99672068884845e-05, + "loss": 1.2658, + "step": 183 + }, + { + "epoch": 0.34, + "grad_norm": 0.1931166169764433, + "learning_rate": 7.996366467881955e-05, + "loss": 1.1637, + "step": 184 + }, + { + "epoch": 0.34, + "grad_norm": 0.18873318157988098, + "learning_rate": 7.995994093016237e-05, + "loss": 1.1335, + "step": 185 + }, + { + "epoch": 0.34, + "grad_norm": 0.19210254625199108, + "learning_rate": 7.995603565942846e-05, + "loss": 1.1928, + "step": 186 + }, + { + "epoch": 0.34, + "grad_norm": 0.2130986479765664, + "learning_rate": 7.995194888435792e-05, + "loss": 1.2158, + "step": 187 + }, + { + "epoch": 0.34, + "grad_norm": 0.22003854501814088, + "learning_rate": 7.994768062351532e-05, + "loss": 1.2288, + "step": 188 + }, + { + "epoch": 0.35, + "grad_norm": 0.20330803191993058, + "learning_rate": 7.994323089628968e-05, + "loss": 1.2426, + "step": 189 + }, + { + "epoch": 0.35, + "grad_norm": 0.20567314642208634, + "learning_rate": 7.993859972289434e-05, + "loss": 1.2649, + "step": 190 + }, + { + "epoch": 0.35, + "grad_norm": 0.21556663727342962, + "learning_rate": 7.993378712436686e-05, + "loss": 1.2545, + "step": 191 + }, + { + "epoch": 0.35, + "grad_norm": 0.20309165469109888, + "learning_rate": 7.992879312256897e-05, + "loss": 1.3338, + "step": 192 + }, + { + "epoch": 0.35, + "grad_norm": 0.19574356669421325, + "learning_rate": 7.992361774018641e-05, + "loss": 1.278, + "step": 193 + }, + { + "epoch": 0.35, + "grad_norm": 0.2763613746722313, + "learning_rate": 7.991826100072891e-05, + "loss": 1.2571, + "step": 194 + }, + { + "epoch": 0.36, + "grad_norm": 0.19346552479915102, + "learning_rate": 7.991272292852996e-05, + "loss": 1.2027, + "step": 195 + }, + { + "epoch": 0.36, + "grad_norm": 0.2281167812123908, + "learning_rate": 7.990700354874683e-05, + "loss": 1.2586, + "step": 196 + }, + { + "epoch": 0.36, + "grad_norm": 0.19699013712137542, + "learning_rate": 7.990110288736042e-05, + "loss": 1.1371, + "step": 197 + }, + { + "epoch": 0.36, + "grad_norm": 0.21768209981475933, + "learning_rate": 7.989502097117503e-05, + "loss": 1.2522, + "step": 198 + }, + { + "epoch": 0.36, + "grad_norm": 0.21335427847754582, + "learning_rate": 7.988875782781838e-05, + "loss": 1.2437, + "step": 199 + }, + { + "epoch": 0.37, + "grad_norm": 0.21856710629066897, + "learning_rate": 7.988231348574147e-05, + "loss": 1.2135, + "step": 200 + }, + { + "epoch": 0.37, + "grad_norm": 0.20482062658774797, + "learning_rate": 7.987568797421836e-05, + "loss": 1.1755, + "step": 201 + }, + { + "epoch": 0.37, + "grad_norm": 0.2017756813960897, + "learning_rate": 7.986888132334608e-05, + "loss": 1.1699, + "step": 202 + }, + { + "epoch": 0.37, + "grad_norm": 0.20496443848153809, + "learning_rate": 7.986189356404458e-05, + "loss": 1.2125, + "step": 203 + }, + { + "epoch": 0.37, + "grad_norm": 0.2134603800558358, + "learning_rate": 7.985472472805643e-05, + "loss": 1.2391, + "step": 204 + }, + { + "epoch": 0.37, + "grad_norm": 0.2364175573420861, + "learning_rate": 7.98473748479468e-05, + "loss": 1.2384, + "step": 205 + }, + { + "epoch": 0.38, + "grad_norm": 0.1872419861598724, + "learning_rate": 7.983984395710326e-05, + "loss": 1.1457, + "step": 206 + }, + { + "epoch": 0.38, + "grad_norm": 0.28222194007095774, + "learning_rate": 7.983213208973566e-05, + "loss": 1.2952, + "step": 207 + }, + { + "epoch": 0.38, + "grad_norm": 0.1916094851162064, + "learning_rate": 7.982423928087593e-05, + "loss": 1.1763, + "step": 208 + }, + { + "epoch": 0.38, + "grad_norm": 0.18446245256166657, + "learning_rate": 7.981616556637795e-05, + "loss": 1.1863, + "step": 209 + }, + { + "epoch": 0.38, + "grad_norm": 0.195191961022491, + "learning_rate": 7.980791098291737e-05, + "loss": 1.2036, + "step": 210 + }, + { + "epoch": 0.39, + "grad_norm": 0.2652439657825496, + "learning_rate": 7.979947556799151e-05, + "loss": 1.2834, + "step": 211 + }, + { + "epoch": 0.39, + "grad_norm": 0.24308438957843412, + "learning_rate": 7.979085935991906e-05, + "loss": 1.234, + "step": 212 + }, + { + "epoch": 0.39, + "grad_norm": 0.21294701043622016, + "learning_rate": 7.978206239784004e-05, + "loss": 1.3006, + "step": 213 + }, + { + "epoch": 0.39, + "grad_norm": 0.25809277041859524, + "learning_rate": 7.977308472171553e-05, + "loss": 1.2272, + "step": 214 + }, + { + "epoch": 0.39, + "grad_norm": 0.193463860107294, + "learning_rate": 7.976392637232754e-05, + "loss": 1.2295, + "step": 215 + }, + { + "epoch": 0.4, + "grad_norm": 0.2150023760609626, + "learning_rate": 7.975458739127877e-05, + "loss": 1.2135, + "step": 216 + }, + { + "epoch": 0.4, + "grad_norm": 0.22590495955605894, + "learning_rate": 7.974506782099253e-05, + "loss": 1.2532, + "step": 217 + }, + { + "epoch": 0.4, + "grad_norm": 0.21023744668403702, + "learning_rate": 7.973536770471242e-05, + "loss": 1.2472, + "step": 218 + }, + { + "epoch": 0.4, + "grad_norm": 0.2345749799511543, + "learning_rate": 7.972548708650218e-05, + "loss": 1.1791, + "step": 219 + }, + { + "epoch": 0.4, + "grad_norm": 0.2158876734005217, + "learning_rate": 7.971542601124553e-05, + "loss": 1.2483, + "step": 220 + }, + { + "epoch": 0.4, + "grad_norm": 0.29455339949432446, + "learning_rate": 7.970518452464593e-05, + "loss": 1.2894, + "step": 221 + }, + { + "epoch": 0.41, + "grad_norm": 0.23983708730626851, + "learning_rate": 7.969476267322636e-05, + "loss": 1.271, + "step": 222 + }, + { + "epoch": 0.41, + "grad_norm": 0.1922400905426158, + "learning_rate": 7.968416050432912e-05, + "loss": 1.2139, + "step": 223 + }, + { + "epoch": 0.41, + "grad_norm": 0.2238136844422931, + "learning_rate": 7.967337806611568e-05, + "loss": 1.2655, + "step": 224 + }, + { + "epoch": 0.41, + "grad_norm": 0.21230292828267672, + "learning_rate": 7.966241540756631e-05, + "loss": 1.2406, + "step": 225 + }, + { + "epoch": 0.41, + "grad_norm": 0.26656119419070456, + "learning_rate": 7.965127257848004e-05, + "loss": 1.2595, + "step": 226 + }, + { + "epoch": 0.42, + "grad_norm": 0.22381385502992684, + "learning_rate": 7.963994962947426e-05, + "loss": 1.1737, + "step": 227 + }, + { + "epoch": 0.42, + "grad_norm": 0.20056702203994298, + "learning_rate": 7.962844661198462e-05, + "loss": 1.1969, + "step": 228 + }, + { + "epoch": 0.42, + "grad_norm": 0.20148701321526885, + "learning_rate": 7.961676357826478e-05, + "loss": 1.2151, + "step": 229 + }, + { + "epoch": 0.42, + "grad_norm": 0.20034834807028637, + "learning_rate": 7.960490058138604e-05, + "loss": 1.1455, + "step": 230 + }, + { + "epoch": 0.42, + "grad_norm": 0.21050838521846033, + "learning_rate": 7.959285767523732e-05, + "loss": 1.2223, + "step": 231 + }, + { + "epoch": 0.42, + "grad_norm": 0.20904772138969777, + "learning_rate": 7.95806349145247e-05, + "loss": 1.2534, + "step": 232 + }, + { + "epoch": 0.43, + "grad_norm": 0.20307877304792957, + "learning_rate": 7.956823235477134e-05, + "loss": 1.1352, + "step": 233 + }, + { + "epoch": 0.43, + "grad_norm": 0.20501105270897094, + "learning_rate": 7.95556500523171e-05, + "loss": 1.2031, + "step": 234 + }, + { + "epoch": 0.43, + "grad_norm": 0.19800586972038586, + "learning_rate": 7.954288806431838e-05, + "loss": 1.2567, + "step": 235 + }, + { + "epoch": 0.43, + "grad_norm": 0.2175102450594135, + "learning_rate": 7.952994644874777e-05, + "loss": 1.2538, + "step": 236 + }, + { + "epoch": 0.43, + "grad_norm": 0.22698189300067595, + "learning_rate": 7.951682526439391e-05, + "loss": 1.3088, + "step": 237 + }, + { + "epoch": 0.44, + "grad_norm": 0.19208392014975315, + "learning_rate": 7.950352457086109e-05, + "loss": 1.2336, + "step": 238 + }, + { + "epoch": 0.44, + "grad_norm": 0.27004086334319655, + "learning_rate": 7.949004442856905e-05, + "loss": 1.2012, + "step": 239 + }, + { + "epoch": 0.44, + "grad_norm": 0.23420974954538043, + "learning_rate": 7.947638489875272e-05, + "loss": 1.2244, + "step": 240 + }, + { + "epoch": 0.44, + "grad_norm": 0.20514399124802024, + "learning_rate": 7.946254604346186e-05, + "loss": 1.2548, + "step": 241 + }, + { + "epoch": 0.44, + "grad_norm": 0.19334973602372896, + "learning_rate": 7.944852792556092e-05, + "loss": 1.2104, + "step": 242 + }, + { + "epoch": 0.44, + "grad_norm": 0.1992640714537956, + "learning_rate": 7.943433060872858e-05, + "loss": 1.2628, + "step": 243 + }, + { + "epoch": 0.45, + "grad_norm": 0.203284617090413, + "learning_rate": 7.941995415745761e-05, + "loss": 1.2002, + "step": 244 + }, + { + "epoch": 0.45, + "grad_norm": 0.22795306969682058, + "learning_rate": 7.94053986370545e-05, + "loss": 1.2215, + "step": 245 + }, + { + "epoch": 0.45, + "grad_norm": 0.20789041346838505, + "learning_rate": 7.939066411363915e-05, + "loss": 1.0998, + "step": 246 + }, + { + "epoch": 0.45, + "grad_norm": 0.22354868884742066, + "learning_rate": 7.937575065414464e-05, + "loss": 1.2564, + "step": 247 + }, + { + "epoch": 0.45, + "grad_norm": 0.21176392726647736, + "learning_rate": 7.936065832631687e-05, + "loss": 1.2816, + "step": 248 + }, + { + "epoch": 0.46, + "grad_norm": 0.19967179557235587, + "learning_rate": 7.934538719871427e-05, + "loss": 1.1961, + "step": 249 + }, + { + "epoch": 0.46, + "grad_norm": 0.210819577350627, + "learning_rate": 7.932993734070747e-05, + "loss": 1.2167, + "step": 250 + }, + { + "epoch": 0.46, + "grad_norm": 0.21537794551756187, + "learning_rate": 7.931430882247903e-05, + "loss": 1.2341, + "step": 251 + }, + { + "epoch": 0.46, + "grad_norm": 0.22850872387256574, + "learning_rate": 7.929850171502304e-05, + "loss": 1.1686, + "step": 252 + }, + { + "epoch": 0.46, + "grad_norm": 0.22380366415076383, + "learning_rate": 7.928251609014493e-05, + "loss": 1.1462, + "step": 253 + }, + { + "epoch": 0.46, + "grad_norm": 0.22426923149036065, + "learning_rate": 7.926635202046102e-05, + "loss": 1.1792, + "step": 254 + }, + { + "epoch": 0.47, + "grad_norm": 0.42082703321103965, + "learning_rate": 7.925000957939822e-05, + "loss": 1.2718, + "step": 255 + }, + { + "epoch": 0.47, + "grad_norm": 0.2235432774854074, + "learning_rate": 7.92334888411937e-05, + "loss": 1.2598, + "step": 256 + }, + { + "epoch": 0.47, + "grad_norm": 0.281644028934108, + "learning_rate": 7.92167898808946e-05, + "loss": 1.2205, + "step": 257 + }, + { + "epoch": 0.47, + "grad_norm": 0.2037705143888748, + "learning_rate": 7.919991277435763e-05, + "loss": 1.1737, + "step": 258 + }, + { + "epoch": 0.47, + "grad_norm": 0.20917419230028977, + "learning_rate": 7.918285759824879e-05, + "loss": 1.2035, + "step": 259 + }, + { + "epoch": 0.48, + "grad_norm": 0.20510847570635518, + "learning_rate": 7.916562443004292e-05, + "loss": 1.2135, + "step": 260 + }, + { + "epoch": 0.48, + "grad_norm": 0.25172483071092466, + "learning_rate": 7.914821334802342e-05, + "loss": 1.2218, + "step": 261 + }, + { + "epoch": 0.48, + "grad_norm": 0.21102706700634313, + "learning_rate": 7.91306244312819e-05, + "loss": 1.1738, + "step": 262 + }, + { + "epoch": 0.48, + "grad_norm": 0.22626060872645815, + "learning_rate": 7.911285775971781e-05, + "loss": 1.238, + "step": 263 + }, + { + "epoch": 0.48, + "grad_norm": 0.22448567539778486, + "learning_rate": 7.909491341403805e-05, + "loss": 1.2404, + "step": 264 + }, + { + "epoch": 0.48, + "grad_norm": 0.2019099786139193, + "learning_rate": 7.907679147575661e-05, + "loss": 1.213, + "step": 265 + }, + { + "epoch": 0.49, + "grad_norm": 0.24307234839096267, + "learning_rate": 7.905849202719422e-05, + "loss": 1.2322, + "step": 266 + }, + { + "epoch": 0.49, + "grad_norm": 0.19801890521743487, + "learning_rate": 7.904001515147802e-05, + "loss": 1.2448, + "step": 267 + }, + { + "epoch": 0.49, + "grad_norm": 0.2102742273575385, + "learning_rate": 7.902136093254106e-05, + "loss": 1.1657, + "step": 268 + }, + { + "epoch": 0.49, + "grad_norm": 0.2173464476815016, + "learning_rate": 7.900252945512201e-05, + "loss": 1.2549, + "step": 269 + }, + { + "epoch": 0.49, + "grad_norm": 0.20957275458699595, + "learning_rate": 7.898352080476479e-05, + "loss": 1.2536, + "step": 270 + }, + { + "epoch": 0.5, + "grad_norm": 0.20691966388952363, + "learning_rate": 7.896433506781811e-05, + "loss": 1.2661, + "step": 271 + }, + { + "epoch": 0.5, + "grad_norm": 0.2276662275112648, + "learning_rate": 7.894497233143509e-05, + "loss": 1.2409, + "step": 272 + }, + { + "epoch": 0.5, + "grad_norm": 0.23854109569301263, + "learning_rate": 7.892543268357297e-05, + "loss": 1.2681, + "step": 273 + }, + { + "epoch": 0.5, + "grad_norm": 0.2233864156677627, + "learning_rate": 7.890571621299252e-05, + "loss": 1.1687, + "step": 274 + }, + { + "epoch": 0.5, + "grad_norm": 0.20114129147925475, + "learning_rate": 7.888582300925787e-05, + "loss": 1.2184, + "step": 275 + }, + { + "epoch": 0.5, + "grad_norm": 0.2154654670569462, + "learning_rate": 7.886575316273586e-05, + "loss": 1.1982, + "step": 276 + }, + { + "epoch": 0.51, + "grad_norm": 0.2292982209343639, + "learning_rate": 7.884550676459583e-05, + "loss": 1.2129, + "step": 277 + }, + { + "epoch": 0.51, + "grad_norm": 0.21302713135229548, + "learning_rate": 7.882508390680908e-05, + "loss": 1.1605, + "step": 278 + }, + { + "epoch": 0.51, + "grad_norm": 0.2123661020671048, + "learning_rate": 7.88044846821485e-05, + "loss": 1.2308, + "step": 279 + }, + { + "epoch": 0.51, + "grad_norm": 0.2080577410800404, + "learning_rate": 7.878370918418818e-05, + "loss": 1.2195, + "step": 280 + }, + { + "epoch": 0.51, + "grad_norm": 0.19663901881127385, + "learning_rate": 7.876275750730289e-05, + "loss": 1.1591, + "step": 281 + }, + { + "epoch": 0.52, + "grad_norm": 0.20534502031312163, + "learning_rate": 7.874162974666776e-05, + "loss": 1.2664, + "step": 282 + }, + { + "epoch": 0.52, + "grad_norm": 0.23240445399513837, + "learning_rate": 7.872032599825779e-05, + "loss": 1.2151, + "step": 283 + }, + { + "epoch": 0.52, + "grad_norm": 0.2672527316717507, + "learning_rate": 7.86988463588474e-05, + "loss": 1.2406, + "step": 284 + }, + { + "epoch": 0.52, + "grad_norm": 0.19893903058743695, + "learning_rate": 7.867719092601003e-05, + "loss": 1.1291, + "step": 285 + }, + { + "epoch": 0.52, + "grad_norm": 0.33275268109930917, + "learning_rate": 7.865535979811768e-05, + "loss": 1.1406, + "step": 286 + }, + { + "epoch": 0.52, + "grad_norm": 0.2373619455690358, + "learning_rate": 7.863335307434045e-05, + "loss": 1.2799, + "step": 287 + }, + { + "epoch": 0.53, + "grad_norm": 0.263235735390858, + "learning_rate": 7.861117085464612e-05, + "loss": 1.2415, + "step": 288 + }, + { + "epoch": 0.53, + "grad_norm": 0.25884281780784324, + "learning_rate": 7.858881323979965e-05, + "loss": 1.3919, + "step": 289 + }, + { + "epoch": 0.53, + "grad_norm": 0.25426288332255736, + "learning_rate": 7.85662803313628e-05, + "loss": 1.174, + "step": 290 + }, + { + "epoch": 0.53, + "grad_norm": 0.26655405527881243, + "learning_rate": 7.854357223169356e-05, + "loss": 1.2806, + "step": 291 + }, + { + "epoch": 0.53, + "grad_norm": 0.20909844432349833, + "learning_rate": 7.852068904394579e-05, + "loss": 1.2627, + "step": 292 + }, + { + "epoch": 0.54, + "grad_norm": 0.21307115068935759, + "learning_rate": 7.849763087206866e-05, + "loss": 1.1879, + "step": 293 + }, + { + "epoch": 0.54, + "grad_norm": 0.25009949471398946, + "learning_rate": 7.847439782080628e-05, + "loss": 1.2881, + "step": 294 + }, + { + "epoch": 0.54, + "grad_norm": 0.20960783418679174, + "learning_rate": 7.845098999569712e-05, + "loss": 1.2723, + "step": 295 + }, + { + "epoch": 0.54, + "grad_norm": 0.24968832437925104, + "learning_rate": 7.842740750307362e-05, + "loss": 1.2029, + "step": 296 + }, + { + "epoch": 0.54, + "grad_norm": 0.22981196585125677, + "learning_rate": 7.84036504500616e-05, + "loss": 1.1695, + "step": 297 + }, + { + "epoch": 0.55, + "grad_norm": 0.2320606844751365, + "learning_rate": 7.837971894457991e-05, + "loss": 1.2317, + "step": 298 + }, + { + "epoch": 0.55, + "grad_norm": 0.23051459673906124, + "learning_rate": 7.835561309533981e-05, + "loss": 1.2046, + "step": 299 + }, + { + "epoch": 0.55, + "grad_norm": 0.2510027231060586, + "learning_rate": 7.833133301184457e-05, + "loss": 1.199, + "step": 300 + }, + { + "epoch": 0.55, + "grad_norm": 0.23601180466018787, + "learning_rate": 7.830687880438895e-05, + "loss": 1.1755, + "step": 301 + }, + { + "epoch": 0.55, + "grad_norm": 0.24740820934385369, + "learning_rate": 7.828225058405864e-05, + "loss": 1.2054, + "step": 302 + }, + { + "epoch": 0.55, + "grad_norm": 0.23065372979111173, + "learning_rate": 7.825744846272984e-05, + "loss": 1.2066, + "step": 303 + }, + { + "epoch": 0.56, + "grad_norm": 0.22385077334838213, + "learning_rate": 7.823247255306866e-05, + "loss": 1.2147, + "step": 304 + }, + { + "epoch": 0.56, + "grad_norm": 0.42981213948386104, + "learning_rate": 7.820732296853074e-05, + "loss": 1.2314, + "step": 305 + }, + { + "epoch": 0.56, + "grad_norm": 0.21122844902751076, + "learning_rate": 7.818199982336058e-05, + "loss": 1.1462, + "step": 306 + }, + { + "epoch": 0.56, + "grad_norm": 0.23374869692118933, + "learning_rate": 7.815650323259117e-05, + "loss": 1.2051, + "step": 307 + }, + { + "epoch": 0.56, + "grad_norm": 0.21662363795962128, + "learning_rate": 7.813083331204332e-05, + "loss": 1.1575, + "step": 308 + }, + { + "epoch": 0.57, + "grad_norm": 0.2088315773384112, + "learning_rate": 7.810499017832526e-05, + "loss": 1.1316, + "step": 309 + }, + { + "epoch": 0.57, + "grad_norm": 0.2095238410730976, + "learning_rate": 7.807897394883203e-05, + "loss": 1.2087, + "step": 310 + }, + { + "epoch": 0.57, + "grad_norm": 0.22672932127256515, + "learning_rate": 7.805278474174499e-05, + "loss": 1.2512, + "step": 311 + }, + { + "epoch": 0.57, + "grad_norm": 0.21873052340922736, + "learning_rate": 7.802642267603126e-05, + "loss": 1.1909, + "step": 312 + }, + { + "epoch": 0.57, + "grad_norm": 0.219814521916342, + "learning_rate": 7.79998878714432e-05, + "loss": 1.1669, + "step": 313 + }, + { + "epoch": 0.57, + "grad_norm": 0.3049426027257317, + "learning_rate": 7.797318044851786e-05, + "loss": 1.1797, + "step": 314 + }, + { + "epoch": 0.58, + "grad_norm": 0.22309435690065985, + "learning_rate": 7.794630052857638e-05, + "loss": 1.1417, + "step": 315 + }, + { + "epoch": 0.58, + "grad_norm": 0.3891885169154885, + "learning_rate": 7.791924823372354e-05, + "loss": 1.2369, + "step": 316 + }, + { + "epoch": 0.58, + "grad_norm": 0.24780269452456372, + "learning_rate": 7.789202368684711e-05, + "loss": 1.2521, + "step": 317 + }, + { + "epoch": 0.58, + "grad_norm": 0.21660460720269362, + "learning_rate": 7.786462701161738e-05, + "loss": 1.2151, + "step": 318 + }, + { + "epoch": 0.58, + "grad_norm": 0.23635409466561857, + "learning_rate": 7.783705833248649e-05, + "loss": 1.2363, + "step": 319 + }, + { + "epoch": 0.59, + "grad_norm": 0.2616135839903218, + "learning_rate": 7.780931777468797e-05, + "loss": 1.2428, + "step": 320 + }, + { + "epoch": 0.59, + "grad_norm": 0.21461059159245083, + "learning_rate": 7.77814054642361e-05, + "loss": 1.1434, + "step": 321 + }, + { + "epoch": 0.59, + "grad_norm": 0.25348824286656163, + "learning_rate": 7.775332152792539e-05, + "loss": 1.2368, + "step": 322 + }, + { + "epoch": 0.59, + "grad_norm": 0.22275034726331247, + "learning_rate": 7.772506609332995e-05, + "loss": 1.1827, + "step": 323 + }, + { + "epoch": 0.59, + "grad_norm": 0.25030821228147526, + "learning_rate": 7.769663928880298e-05, + "loss": 1.2428, + "step": 324 + }, + { + "epoch": 0.59, + "grad_norm": 0.22251804398745534, + "learning_rate": 7.766804124347608e-05, + "loss": 1.1889, + "step": 325 + }, + { + "epoch": 0.6, + "grad_norm": 0.23381455520411995, + "learning_rate": 7.763927208725879e-05, + "loss": 1.2115, + "step": 326 + }, + { + "epoch": 0.6, + "grad_norm": 0.27341902651946226, + "learning_rate": 7.761033195083791e-05, + "loss": 1.2535, + "step": 327 + }, + { + "epoch": 0.6, + "grad_norm": 0.24862471659814522, + "learning_rate": 7.758122096567694e-05, + "loss": 1.2128, + "step": 328 + }, + { + "epoch": 0.6, + "grad_norm": 0.2251357082045494, + "learning_rate": 7.755193926401547e-05, + "loss": 1.2334, + "step": 329 + }, + { + "epoch": 0.6, + "grad_norm": 0.3173274941622932, + "learning_rate": 7.752248697886857e-05, + "loss": 1.226, + "step": 330 + }, + { + "epoch": 0.61, + "grad_norm": 0.23056440717672175, + "learning_rate": 7.74928642440263e-05, + "loss": 1.2339, + "step": 331 + }, + { + "epoch": 0.61, + "grad_norm": 0.2801507500859342, + "learning_rate": 7.746307119405286e-05, + "loss": 1.287, + "step": 332 + }, + { + "epoch": 0.61, + "grad_norm": 0.2267818430426272, + "learning_rate": 7.743310796428622e-05, + "loss": 1.1916, + "step": 333 + }, + { + "epoch": 0.61, + "grad_norm": 0.2777329160365585, + "learning_rate": 7.74029746908374e-05, + "loss": 1.252, + "step": 334 + }, + { + "epoch": 0.61, + "grad_norm": 0.25289169762353, + "learning_rate": 7.737267151058983e-05, + "loss": 1.2153, + "step": 335 + }, + { + "epoch": 0.61, + "grad_norm": 0.2424670686901653, + "learning_rate": 7.734219856119875e-05, + "loss": 1.2227, + "step": 336 + }, + { + "epoch": 0.62, + "grad_norm": 0.22747092217441645, + "learning_rate": 7.731155598109067e-05, + "loss": 1.19, + "step": 337 + }, + { + "epoch": 0.62, + "grad_norm": 0.2307810940100189, + "learning_rate": 7.728074390946257e-05, + "loss": 1.1818, + "step": 338 + }, + { + "epoch": 0.62, + "grad_norm": 0.2583402574655623, + "learning_rate": 7.724976248628142e-05, + "loss": 1.1608, + "step": 339 + }, + { + "epoch": 0.62, + "grad_norm": 0.22140209760890694, + "learning_rate": 7.721861185228347e-05, + "loss": 1.1245, + "step": 340 + }, + { + "epoch": 0.62, + "grad_norm": 0.25859310758244686, + "learning_rate": 7.718729214897362e-05, + "loss": 1.2247, + "step": 341 + }, + { + "epoch": 0.63, + "grad_norm": 0.26371179531372124, + "learning_rate": 7.715580351862482e-05, + "loss": 1.2128, + "step": 342 + }, + { + "epoch": 0.63, + "grad_norm": 0.26575541302851047, + "learning_rate": 7.712414610427733e-05, + "loss": 1.2443, + "step": 343 + }, + { + "epoch": 0.63, + "grad_norm": 0.269978305197599, + "learning_rate": 7.709232004973816e-05, + "loss": 1.2231, + "step": 344 + }, + { + "epoch": 0.63, + "grad_norm": 0.26583998705977047, + "learning_rate": 7.70603254995804e-05, + "loss": 1.2476, + "step": 345 + }, + { + "epoch": 0.63, + "grad_norm": 0.24256062164066097, + "learning_rate": 7.702816259914253e-05, + "loss": 1.2901, + "step": 346 + }, + { + "epoch": 0.63, + "grad_norm": 0.3463123472658915, + "learning_rate": 7.699583149452779e-05, + "loss": 1.3277, + "step": 347 + }, + { + "epoch": 0.64, + "grad_norm": 0.2269096590531878, + "learning_rate": 7.696333233260345e-05, + "loss": 1.2047, + "step": 348 + }, + { + "epoch": 0.64, + "grad_norm": 0.25136883001050025, + "learning_rate": 7.693066526100031e-05, + "loss": 1.1619, + "step": 349 + }, + { + "epoch": 0.64, + "grad_norm": 0.2565112571116145, + "learning_rate": 7.68978304281118e-05, + "loss": 1.2389, + "step": 350 + }, + { + "epoch": 0.64, + "grad_norm": 0.22175779550828703, + "learning_rate": 7.686482798309349e-05, + "loss": 1.2238, + "step": 351 + }, + { + "epoch": 0.64, + "grad_norm": 0.22588304332216555, + "learning_rate": 7.683165807586234e-05, + "loss": 1.174, + "step": 352 + }, + { + "epoch": 0.65, + "grad_norm": 0.24889474296529737, + "learning_rate": 7.6798320857096e-05, + "loss": 1.2366, + "step": 353 + }, + { + "epoch": 0.65, + "grad_norm": 0.27339703806525034, + "learning_rate": 7.676481647823214e-05, + "loss": 1.2356, + "step": 354 + }, + { + "epoch": 0.65, + "grad_norm": 0.23424666722888365, + "learning_rate": 7.673114509146782e-05, + "loss": 1.2089, + "step": 355 + }, + { + "epoch": 0.65, + "grad_norm": 0.27978285392461766, + "learning_rate": 7.66973068497587e-05, + "loss": 1.2609, + "step": 356 + }, + { + "epoch": 0.65, + "grad_norm": 0.2509423350138824, + "learning_rate": 7.666330190681844e-05, + "loss": 1.1777, + "step": 357 + }, + { + "epoch": 0.65, + "grad_norm": 0.23007730927468031, + "learning_rate": 7.662913041711793e-05, + "loss": 1.154, + "step": 358 + }, + { + "epoch": 0.66, + "grad_norm": 0.2438648674953112, + "learning_rate": 7.659479253588462e-05, + "loss": 1.2257, + "step": 359 + }, + { + "epoch": 0.66, + "grad_norm": 0.28816093242092233, + "learning_rate": 7.65602884191018e-05, + "loss": 1.2558, + "step": 360 + }, + { + "epoch": 0.66, + "grad_norm": 0.24972815300596035, + "learning_rate": 7.652561822350793e-05, + "loss": 1.2837, + "step": 361 + }, + { + "epoch": 0.66, + "grad_norm": 0.2543189139697063, + "learning_rate": 7.649078210659587e-05, + "loss": 1.2193, + "step": 362 + }, + { + "epoch": 0.66, + "grad_norm": 0.2237937956718952, + "learning_rate": 7.645578022661224e-05, + "loss": 1.2237, + "step": 363 + }, + { + "epoch": 0.67, + "grad_norm": 0.29742029408787396, + "learning_rate": 7.642061274255657e-05, + "loss": 1.2116, + "step": 364 + }, + { + "epoch": 0.67, + "grad_norm": 0.2462883147335493, + "learning_rate": 7.638527981418075e-05, + "loss": 1.1827, + "step": 365 + }, + { + "epoch": 0.67, + "grad_norm": 0.2647802498907096, + "learning_rate": 7.634978160198817e-05, + "loss": 1.2739, + "step": 366 + }, + { + "epoch": 0.67, + "grad_norm": 0.22360398779217264, + "learning_rate": 7.631411826723306e-05, + "loss": 1.2185, + "step": 367 + }, + { + "epoch": 0.67, + "grad_norm": 0.2635048004593543, + "learning_rate": 7.627828997191973e-05, + "loss": 1.2317, + "step": 368 + }, + { + "epoch": 0.67, + "grad_norm": 0.2764803449917684, + "learning_rate": 7.624229687880184e-05, + "loss": 1.1923, + "step": 369 + }, + { + "epoch": 0.68, + "grad_norm": 0.25724943233414527, + "learning_rate": 7.620613915138166e-05, + "loss": 1.2218, + "step": 370 + }, + { + "epoch": 0.68, + "grad_norm": 0.2858318045794755, + "learning_rate": 7.61698169539093e-05, + "loss": 1.1496, + "step": 371 + }, + { + "epoch": 0.68, + "grad_norm": 0.23547216647460364, + "learning_rate": 7.613333045138206e-05, + "loss": 1.1905, + "step": 372 + }, + { + "epoch": 0.68, + "grad_norm": 0.22984814903684375, + "learning_rate": 7.609667980954355e-05, + "loss": 1.2009, + "step": 373 + }, + { + "epoch": 0.68, + "grad_norm": 0.2551903754079084, + "learning_rate": 7.605986519488301e-05, + "loss": 1.2042, + "step": 374 + }, + { + "epoch": 0.69, + "grad_norm": 0.2508257410125616, + "learning_rate": 7.602288677463457e-05, + "loss": 1.2468, + "step": 375 + }, + { + "epoch": 0.69, + "grad_norm": 0.25324577774935964, + "learning_rate": 7.598574471677644e-05, + "loss": 1.2603, + "step": 376 + }, + { + "epoch": 0.69, + "grad_norm": 0.35888776531769967, + "learning_rate": 7.59484391900302e-05, + "loss": 1.1929, + "step": 377 + }, + { + "epoch": 0.69, + "grad_norm": 0.22048517191014724, + "learning_rate": 7.591097036385994e-05, + "loss": 1.1783, + "step": 378 + }, + { + "epoch": 0.69, + "grad_norm": 0.2781160412746083, + "learning_rate": 7.587333840847162e-05, + "loss": 1.3397, + "step": 379 + }, + { + "epoch": 0.7, + "grad_norm": 0.24033046830332258, + "learning_rate": 7.583554349481222e-05, + "loss": 1.2436, + "step": 380 + }, + { + "epoch": 0.7, + "grad_norm": 0.26413762380260003, + "learning_rate": 7.579758579456893e-05, + "loss": 1.1917, + "step": 381 + }, + { + "epoch": 0.7, + "grad_norm": 0.2390937887338632, + "learning_rate": 7.575946548016847e-05, + "loss": 1.2186, + "step": 382 + }, + { + "epoch": 0.7, + "grad_norm": 0.25131263043429275, + "learning_rate": 7.572118272477622e-05, + "loss": 1.2538, + "step": 383 + }, + { + "epoch": 0.7, + "grad_norm": 0.223974104870702, + "learning_rate": 7.568273770229546e-05, + "loss": 1.2165, + "step": 384 + }, + { + "epoch": 0.7, + "grad_norm": 0.25840356830252875, + "learning_rate": 7.564413058736663e-05, + "loss": 1.1848, + "step": 385 + }, + { + "epoch": 0.71, + "grad_norm": 0.2723156683076603, + "learning_rate": 7.560536155536641e-05, + "loss": 1.1982, + "step": 386 + }, + { + "epoch": 0.71, + "grad_norm": 0.265687427976889, + "learning_rate": 7.556643078240708e-05, + "loss": 1.231, + "step": 387 + }, + { + "epoch": 0.71, + "grad_norm": 0.25152762080976077, + "learning_rate": 7.552733844533562e-05, + "loss": 1.1974, + "step": 388 + }, + { + "epoch": 0.71, + "grad_norm": 0.2366049485053541, + "learning_rate": 7.548808472173292e-05, + "loss": 1.3119, + "step": 389 + }, + { + "epoch": 0.71, + "grad_norm": 0.22092196577077122, + "learning_rate": 7.5448669789913e-05, + "loss": 1.195, + "step": 390 + }, + { + "epoch": 0.72, + "grad_norm": 0.22667521540462374, + "learning_rate": 7.540909382892217e-05, + "loss": 1.1431, + "step": 391 + }, + { + "epoch": 0.72, + "grad_norm": 0.25432207282646513, + "learning_rate": 7.536935701853823e-05, + "loss": 1.2173, + "step": 392 + }, + { + "epoch": 0.72, + "grad_norm": 0.29950506457923864, + "learning_rate": 7.53294595392697e-05, + "loss": 1.1962, + "step": 393 + }, + { + "epoch": 0.72, + "grad_norm": 0.24735689607229913, + "learning_rate": 7.528940157235487e-05, + "loss": 1.2053, + "step": 394 + }, + { + "epoch": 0.72, + "grad_norm": 0.24394198607459663, + "learning_rate": 7.524918329976114e-05, + "loss": 1.1979, + "step": 395 + }, + { + "epoch": 0.72, + "grad_norm": 0.2630369372689188, + "learning_rate": 7.520880490418409e-05, + "loss": 1.2111, + "step": 396 + }, + { + "epoch": 0.73, + "grad_norm": 0.26275028416291457, + "learning_rate": 7.516826656904664e-05, + "loss": 1.2133, + "step": 397 + }, + { + "epoch": 0.73, + "grad_norm": 0.23938074620956928, + "learning_rate": 7.512756847849831e-05, + "loss": 1.1355, + "step": 398 + }, + { + "epoch": 0.73, + "grad_norm": 0.3724960610098138, + "learning_rate": 7.508671081741428e-05, + "loss": 1.2572, + "step": 399 + }, + { + "epoch": 0.73, + "grad_norm": 0.24161685847894723, + "learning_rate": 7.504569377139462e-05, + "loss": 1.1706, + "step": 400 + }, + { + "epoch": 0.73, + "grad_norm": 0.26121591322670523, + "learning_rate": 7.50045175267634e-05, + "loss": 1.2135, + "step": 401 + }, + { + "epoch": 0.74, + "grad_norm": 0.2465579498164775, + "learning_rate": 7.496318227056788e-05, + "loss": 1.1641, + "step": 402 + }, + { + "epoch": 0.74, + "grad_norm": 0.2556288696122787, + "learning_rate": 7.492168819057767e-05, + "loss": 1.2939, + "step": 403 + }, + { + "epoch": 0.74, + "grad_norm": 0.261481216336303, + "learning_rate": 7.488003547528382e-05, + "loss": 1.2026, + "step": 404 + }, + { + "epoch": 0.74, + "grad_norm": 0.2389415135676362, + "learning_rate": 7.483822431389799e-05, + "loss": 1.2131, + "step": 405 + }, + { + "epoch": 0.74, + "grad_norm": 0.2559201956627192, + "learning_rate": 7.479625489635162e-05, + "loss": 1.1246, + "step": 406 + }, + { + "epoch": 0.74, + "grad_norm": 0.27127932491822604, + "learning_rate": 7.475412741329504e-05, + "loss": 1.2429, + "step": 407 + }, + { + "epoch": 0.75, + "grad_norm": 0.27006004008695594, + "learning_rate": 7.47118420560966e-05, + "loss": 1.2388, + "step": 408 + }, + { + "epoch": 0.75, + "grad_norm": 0.23716823297200537, + "learning_rate": 7.466939901684182e-05, + "loss": 1.1264, + "step": 409 + }, + { + "epoch": 0.75, + "grad_norm": 0.2885373898669248, + "learning_rate": 7.462679848833252e-05, + "loss": 1.2786, + "step": 410 + }, + { + "epoch": 0.75, + "grad_norm": 0.49215227598639927, + "learning_rate": 7.458404066408588e-05, + "loss": 1.2386, + "step": 411 + }, + { + "epoch": 0.75, + "grad_norm": 0.24235735604947403, + "learning_rate": 7.454112573833368e-05, + "loss": 1.1423, + "step": 412 + }, + { + "epoch": 0.76, + "grad_norm": 0.2584614748054343, + "learning_rate": 7.449805390602127e-05, + "loss": 1.2669, + "step": 413 + }, + { + "epoch": 0.76, + "grad_norm": 0.23806123085998873, + "learning_rate": 7.445482536280684e-05, + "loss": 1.1763, + "step": 414 + }, + { + "epoch": 0.76, + "grad_norm": 0.24459517607786851, + "learning_rate": 7.441144030506043e-05, + "loss": 1.198, + "step": 415 + }, + { + "epoch": 0.76, + "grad_norm": 0.25801616402700395, + "learning_rate": 7.436789892986304e-05, + "loss": 1.2136, + "step": 416 + }, + { + "epoch": 0.76, + "grad_norm": 0.2814819942392514, + "learning_rate": 7.432420143500578e-05, + "loss": 1.2398, + "step": 417 + }, + { + "epoch": 0.76, + "grad_norm": 0.22134709322606153, + "learning_rate": 7.428034801898893e-05, + "loss": 1.1592, + "step": 418 + }, + { + "epoch": 0.77, + "grad_norm": 0.2899677536995633, + "learning_rate": 7.42363388810211e-05, + "loss": 1.2296, + "step": 419 + }, + { + "epoch": 0.77, + "grad_norm": 0.24005943230262294, + "learning_rate": 7.419217422101822e-05, + "loss": 1.2223, + "step": 420 + }, + { + "epoch": 0.77, + "grad_norm": 0.26417562369496167, + "learning_rate": 7.414785423960275e-05, + "loss": 1.2261, + "step": 421 + }, + { + "epoch": 0.77, + "grad_norm": 0.2580815883535521, + "learning_rate": 7.410337913810271e-05, + "loss": 1.2021, + "step": 422 + }, + { + "epoch": 0.77, + "grad_norm": 0.25242217589496435, + "learning_rate": 7.405874911855071e-05, + "loss": 1.239, + "step": 423 + }, + { + "epoch": 0.78, + "grad_norm": 0.21991733999839932, + "learning_rate": 7.401396438368315e-05, + "loss": 1.1716, + "step": 424 + }, + { + "epoch": 0.78, + "grad_norm": 0.40116538322720213, + "learning_rate": 7.396902513693924e-05, + "loss": 1.2773, + "step": 425 + }, + { + "epoch": 0.78, + "grad_norm": 0.277333939455099, + "learning_rate": 7.392393158246002e-05, + "loss": 1.2574, + "step": 426 + }, + { + "epoch": 0.78, + "grad_norm": 0.27146087746385755, + "learning_rate": 7.387868392508756e-05, + "loss": 1.2243, + "step": 427 + }, + { + "epoch": 0.78, + "grad_norm": 0.255881055620786, + "learning_rate": 7.38332823703639e-05, + "loss": 1.223, + "step": 428 + }, + { + "epoch": 0.78, + "grad_norm": 0.24807364856677255, + "learning_rate": 7.378772712453021e-05, + "loss": 1.1985, + "step": 429 + }, + { + "epoch": 0.79, + "grad_norm": 0.25746257617764423, + "learning_rate": 7.37420183945258e-05, + "loss": 1.2502, + "step": 430 + }, + { + "epoch": 0.79, + "grad_norm": 0.28851991049982234, + "learning_rate": 7.369615638798722e-05, + "loss": 1.2535, + "step": 431 + }, + { + "epoch": 0.79, + "grad_norm": 0.24113389811604363, + "learning_rate": 7.365014131324725e-05, + "loss": 1.2227, + "step": 432 + }, + { + "epoch": 0.79, + "grad_norm": 0.2414465151257969, + "learning_rate": 7.360397337933405e-05, + "loss": 1.1884, + "step": 433 + }, + { + "epoch": 0.79, + "grad_norm": 0.2735463134699831, + "learning_rate": 7.355765279597011e-05, + "loss": 1.2756, + "step": 434 + }, + { + "epoch": 0.8, + "grad_norm": 0.2588437452987293, + "learning_rate": 7.351117977357139e-05, + "loss": 1.2108, + "step": 435 + }, + { + "epoch": 0.8, + "grad_norm": 0.26573294117796553, + "learning_rate": 7.346455452324629e-05, + "loss": 1.1821, + "step": 436 + }, + { + "epoch": 0.8, + "grad_norm": 0.2555476577827304, + "learning_rate": 7.341777725679473e-05, + "loss": 1.1937, + "step": 437 + }, + { + "epoch": 0.8, + "grad_norm": 0.2867704132108098, + "learning_rate": 7.337084818670716e-05, + "loss": 1.2272, + "step": 438 + }, + { + "epoch": 0.8, + "grad_norm": 0.27726678115981157, + "learning_rate": 7.332376752616367e-05, + "loss": 1.2331, + "step": 439 + }, + { + "epoch": 0.8, + "grad_norm": 0.26955338021079955, + "learning_rate": 7.32765354890329e-05, + "loss": 1.1731, + "step": 440 + }, + { + "epoch": 0.81, + "grad_norm": 0.25250321202536524, + "learning_rate": 7.322915228987116e-05, + "loss": 1.2653, + "step": 441 + }, + { + "epoch": 0.81, + "grad_norm": 0.24748844179765395, + "learning_rate": 7.318161814392143e-05, + "loss": 1.24, + "step": 442 + }, + { + "epoch": 0.81, + "grad_norm": 0.28177805247356325, + "learning_rate": 7.313393326711239e-05, + "loss": 1.185, + "step": 443 + }, + { + "epoch": 0.81, + "grad_norm": 0.24093242000396312, + "learning_rate": 7.30860978760574e-05, + "loss": 1.1994, + "step": 444 + }, + { + "epoch": 0.81, + "grad_norm": 0.26277803901457075, + "learning_rate": 7.30381121880536e-05, + "loss": 1.212, + "step": 445 + }, + { + "epoch": 0.82, + "grad_norm": 0.2506524258682433, + "learning_rate": 7.298997642108079e-05, + "loss": 1.2421, + "step": 446 + }, + { + "epoch": 0.82, + "grad_norm": 0.2840599700015824, + "learning_rate": 7.294169079380061e-05, + "loss": 1.1818, + "step": 447 + }, + { + "epoch": 0.82, + "grad_norm": 0.24892184038117549, + "learning_rate": 7.289325552555538e-05, + "loss": 1.1916, + "step": 448 + }, + { + "epoch": 0.82, + "grad_norm": 0.2700898428541357, + "learning_rate": 7.284467083636722e-05, + "loss": 1.2517, + "step": 449 + }, + { + "epoch": 0.82, + "grad_norm": 0.2617848546539419, + "learning_rate": 7.279593694693698e-05, + "loss": 1.2063, + "step": 450 + }, + { + "epoch": 0.82, + "grad_norm": 0.2698278585334131, + "learning_rate": 7.274705407864332e-05, + "loss": 1.194, + "step": 451 + }, + { + "epoch": 0.83, + "grad_norm": 0.23678313024953834, + "learning_rate": 7.26980224535416e-05, + "loss": 1.2349, + "step": 452 + }, + { + "epoch": 0.83, + "grad_norm": 0.24851875792002978, + "learning_rate": 7.264884229436293e-05, + "loss": 1.1758, + "step": 453 + }, + { + "epoch": 0.83, + "grad_norm": 0.24122080121681125, + "learning_rate": 7.259951382451318e-05, + "loss": 1.1962, + "step": 454 + }, + { + "epoch": 0.83, + "grad_norm": 0.22741322959884405, + "learning_rate": 7.25500372680719e-05, + "loss": 1.1702, + "step": 455 + }, + { + "epoch": 0.83, + "grad_norm": 0.2297475610861458, + "learning_rate": 7.250041284979137e-05, + "loss": 1.1466, + "step": 456 + }, + { + "epoch": 0.84, + "grad_norm": 0.3057605989721467, + "learning_rate": 7.245064079509553e-05, + "loss": 1.246, + "step": 457 + }, + { + "epoch": 0.84, + "grad_norm": 0.2719638501597136, + "learning_rate": 7.240072133007899e-05, + "loss": 1.2184, + "step": 458 + }, + { + "epoch": 0.84, + "grad_norm": 0.2436807816414479, + "learning_rate": 7.235065468150593e-05, + "loss": 1.2324, + "step": 459 + }, + { + "epoch": 0.84, + "grad_norm": 0.23436349430255515, + "learning_rate": 7.23004410768092e-05, + "loss": 1.1813, + "step": 460 + }, + { + "epoch": 0.84, + "grad_norm": 0.2398940990211377, + "learning_rate": 7.22500807440892e-05, + "loss": 1.1924, + "step": 461 + }, + { + "epoch": 0.84, + "grad_norm": 0.2605716625062531, + "learning_rate": 7.219957391211281e-05, + "loss": 1.182, + "step": 462 + }, + { + "epoch": 0.85, + "grad_norm": 0.260462524570941, + "learning_rate": 7.214892081031244e-05, + "loss": 1.2136, + "step": 463 + }, + { + "epoch": 0.85, + "grad_norm": 0.21979766512306334, + "learning_rate": 7.209812166878491e-05, + "loss": 1.2066, + "step": 464 + }, + { + "epoch": 0.85, + "grad_norm": 0.23324453647530663, + "learning_rate": 7.204717671829051e-05, + "loss": 1.1657, + "step": 465 + }, + { + "epoch": 0.85, + "grad_norm": 0.2529434935507481, + "learning_rate": 7.199608619025177e-05, + "loss": 1.2093, + "step": 466 + }, + { + "epoch": 0.85, + "grad_norm": 0.25371701891720116, + "learning_rate": 7.194485031675265e-05, + "loss": 1.2225, + "step": 467 + }, + { + "epoch": 0.86, + "grad_norm": 0.23272423066292103, + "learning_rate": 7.189346933053725e-05, + "loss": 1.1721, + "step": 468 + }, + { + "epoch": 0.86, + "grad_norm": 0.25122928735587546, + "learning_rate": 7.184194346500892e-05, + "loss": 1.2537, + "step": 469 + }, + { + "epoch": 0.86, + "grad_norm": 0.2159270875490409, + "learning_rate": 7.179027295422913e-05, + "loss": 1.197, + "step": 470 + }, + { + "epoch": 0.86, + "grad_norm": 0.2633111059076544, + "learning_rate": 7.173845803291636e-05, + "loss": 1.1721, + "step": 471 + }, + { + "epoch": 0.86, + "grad_norm": 0.30555936322098703, + "learning_rate": 7.168649893644517e-05, + "loss": 1.3011, + "step": 472 + }, + { + "epoch": 0.87, + "grad_norm": 0.23492670111453726, + "learning_rate": 7.163439590084502e-05, + "loss": 1.1601, + "step": 473 + }, + { + "epoch": 0.87, + "grad_norm": 0.26602734263721806, + "learning_rate": 7.158214916279923e-05, + "loss": 1.2808, + "step": 474 + }, + { + "epoch": 0.87, + "grad_norm": 0.3182695007856262, + "learning_rate": 7.152975895964386e-05, + "loss": 1.2967, + "step": 475 + }, + { + "epoch": 0.87, + "grad_norm": 0.2785021674736721, + "learning_rate": 7.147722552936673e-05, + "loss": 1.1789, + "step": 476 + }, + { + "epoch": 0.87, + "grad_norm": 0.279474303138652, + "learning_rate": 7.142454911060627e-05, + "loss": 1.2596, + "step": 477 + }, + { + "epoch": 0.87, + "grad_norm": 0.2556980144910755, + "learning_rate": 7.137172994265044e-05, + "loss": 1.2426, + "step": 478 + }, + { + "epoch": 0.88, + "grad_norm": 0.3311256331993533, + "learning_rate": 7.131876826543565e-05, + "loss": 1.2059, + "step": 479 + }, + { + "epoch": 0.88, + "grad_norm": 0.26467296197775253, + "learning_rate": 7.12656643195457e-05, + "loss": 1.2482, + "step": 480 + }, + { + "epoch": 0.88, + "grad_norm": 0.27444885274652553, + "learning_rate": 7.121241834621064e-05, + "loss": 1.2528, + "step": 481 + }, + { + "epoch": 0.88, + "grad_norm": 0.2572283861115396, + "learning_rate": 7.115903058730567e-05, + "loss": 1.1849, + "step": 482 + }, + { + "epoch": 0.88, + "grad_norm": 0.2677065778235683, + "learning_rate": 7.11055012853501e-05, + "loss": 1.2011, + "step": 483 + }, + { + "epoch": 0.89, + "grad_norm": 0.29470622036742816, + "learning_rate": 7.105183068350619e-05, + "loss": 1.2398, + "step": 484 + }, + { + "epoch": 0.89, + "grad_norm": 0.27609230248969197, + "learning_rate": 7.099801902557811e-05, + "loss": 1.2259, + "step": 485 + }, + { + "epoch": 0.89, + "grad_norm": 0.24248634168099284, + "learning_rate": 7.094406655601073e-05, + "loss": 1.2282, + "step": 486 + }, + { + "epoch": 0.89, + "grad_norm": 0.2765941767688746, + "learning_rate": 7.088997351988865e-05, + "loss": 1.2319, + "step": 487 + }, + { + "epoch": 0.89, + "grad_norm": 0.29347776909858947, + "learning_rate": 7.083574016293493e-05, + "loss": 1.1765, + "step": 488 + }, + { + "epoch": 0.89, + "grad_norm": 0.285370295424537, + "learning_rate": 7.078136673151008e-05, + "loss": 1.26, + "step": 489 + }, + { + "epoch": 0.9, + "grad_norm": 0.29408734903836536, + "learning_rate": 7.072685347261093e-05, + "loss": 1.226, + "step": 490 + }, + { + "epoch": 0.9, + "grad_norm": 0.27437470239205813, + "learning_rate": 7.067220063386947e-05, + "loss": 1.1976, + "step": 491 + }, + { + "epoch": 0.9, + "grad_norm": 0.2680770258777871, + "learning_rate": 7.061740846355176e-05, + "loss": 1.1915, + "step": 492 + }, + { + "epoch": 0.9, + "grad_norm": 0.27200362879502954, + "learning_rate": 7.056247721055678e-05, + "loss": 1.2002, + "step": 493 + }, + { + "epoch": 0.9, + "grad_norm": 0.2637811092577037, + "learning_rate": 7.050740712441528e-05, + "loss": 1.287, + "step": 494 + }, + { + "epoch": 0.91, + "grad_norm": 0.24657959209271266, + "learning_rate": 7.045219845528875e-05, + "loss": 1.2284, + "step": 495 + }, + { + "epoch": 0.91, + "grad_norm": 0.25311992110358666, + "learning_rate": 7.039685145396812e-05, + "loss": 1.1616, + "step": 496 + }, + { + "epoch": 0.91, + "grad_norm": 0.2564633694193358, + "learning_rate": 7.034136637187275e-05, + "loss": 1.2067, + "step": 497 + }, + { + "epoch": 0.91, + "grad_norm": 0.2446797651174144, + "learning_rate": 7.028574346104926e-05, + "loss": 1.2284, + "step": 498 + }, + { + "epoch": 0.91, + "grad_norm": 0.2592751463399255, + "learning_rate": 7.022998297417034e-05, + "loss": 1.2371, + "step": 499 + }, + { + "epoch": 0.91, + "grad_norm": 0.2500713943206808, + "learning_rate": 7.017408516453365e-05, + "loss": 1.1061, + "step": 500 + }, + { + "epoch": 0.92, + "grad_norm": 0.2812266276040743, + "learning_rate": 7.011805028606064e-05, + "loss": 1.1949, + "step": 501 + }, + { + "epoch": 0.92, + "grad_norm": 0.298829667668083, + "learning_rate": 7.006187859329544e-05, + "loss": 1.2313, + "step": 502 + }, + { + "epoch": 0.92, + "grad_norm": 0.26518768159745104, + "learning_rate": 7.000557034140361e-05, + "loss": 1.2246, + "step": 503 + }, + { + "epoch": 0.92, + "grad_norm": 0.3037280360760458, + "learning_rate": 6.994912578617113e-05, + "loss": 1.1617, + "step": 504 + }, + { + "epoch": 0.92, + "grad_norm": 0.2726903109255714, + "learning_rate": 6.989254518400309e-05, + "loss": 1.2415, + "step": 505 + }, + { + "epoch": 0.93, + "grad_norm": 0.25568082003046966, + "learning_rate": 6.98358287919226e-05, + "loss": 1.1817, + "step": 506 + }, + { + "epoch": 0.93, + "grad_norm": 0.25633294893705044, + "learning_rate": 6.97789768675696e-05, + "loss": 1.2149, + "step": 507 + }, + { + "epoch": 0.93, + "grad_norm": 0.28291439435087123, + "learning_rate": 6.972198966919972e-05, + "loss": 1.1578, + "step": 508 + }, + { + "epoch": 0.93, + "grad_norm": 0.27195184756655516, + "learning_rate": 6.966486745568308e-05, + "loss": 1.2355, + "step": 509 + }, + { + "epoch": 0.93, + "grad_norm": 0.239159568376005, + "learning_rate": 6.960761048650312e-05, + "loss": 1.1688, + "step": 510 + }, + { + "epoch": 0.93, + "grad_norm": 0.22961475425949177, + "learning_rate": 6.955021902175543e-05, + "loss": 1.2094, + "step": 511 + }, + { + "epoch": 0.94, + "grad_norm": 0.27443773600741117, + "learning_rate": 6.949269332214651e-05, + "loss": 1.2559, + "step": 512 + }, + { + "epoch": 0.94, + "grad_norm": 0.26230551832002097, + "learning_rate": 6.94350336489927e-05, + "loss": 1.2121, + "step": 513 + }, + { + "epoch": 0.94, + "grad_norm": 0.2716742985303849, + "learning_rate": 6.937724026421892e-05, + "loss": 1.2444, + "step": 514 + }, + { + "epoch": 0.94, + "grad_norm": 0.2537850139439542, + "learning_rate": 6.931931343035742e-05, + "loss": 1.1327, + "step": 515 + }, + { + "epoch": 0.94, + "grad_norm": 0.28599587967496826, + "learning_rate": 6.926125341054676e-05, + "loss": 1.2236, + "step": 516 + }, + { + "epoch": 0.95, + "grad_norm": 0.26780654378470103, + "learning_rate": 6.920306046853043e-05, + "loss": 1.2295, + "step": 517 + }, + { + "epoch": 0.95, + "grad_norm": 0.23606296888412015, + "learning_rate": 6.914473486865577e-05, + "loss": 1.1543, + "step": 518 + }, + { + "epoch": 0.95, + "grad_norm": 0.34976881174240837, + "learning_rate": 6.90862768758727e-05, + "loss": 1.2067, + "step": 519 + }, + { + "epoch": 0.95, + "grad_norm": 0.2481257873494882, + "learning_rate": 6.902768675573258e-05, + "loss": 1.2188, + "step": 520 + }, + { + "epoch": 0.95, + "grad_norm": 0.2996395778117021, + "learning_rate": 6.896896477438699e-05, + "loss": 1.2326, + "step": 521 + }, + { + "epoch": 0.95, + "grad_norm": 0.8839768816333193, + "learning_rate": 6.891011119858643e-05, + "loss": 1.2435, + "step": 522 + }, + { + "epoch": 0.96, + "grad_norm": 0.2851882482058998, + "learning_rate": 6.885112629567927e-05, + "loss": 1.2644, + "step": 523 + }, + { + "epoch": 0.96, + "grad_norm": 0.2813663482913699, + "learning_rate": 6.879201033361035e-05, + "loss": 1.2309, + "step": 524 + }, + { + "epoch": 0.96, + "grad_norm": 0.3257551560135454, + "learning_rate": 6.873276358091996e-05, + "loss": 1.2755, + "step": 525 + }, + { + "epoch": 0.96, + "grad_norm": 0.28930479952494365, + "learning_rate": 6.867338630674247e-05, + "loss": 1.1962, + "step": 526 + }, + { + "epoch": 0.96, + "grad_norm": 0.3077462996938649, + "learning_rate": 6.861387878080511e-05, + "loss": 1.2402, + "step": 527 + }, + { + "epoch": 0.97, + "grad_norm": 0.2848900193452761, + "learning_rate": 6.855424127342688e-05, + "loss": 1.2748, + "step": 528 + }, + { + "epoch": 0.97, + "grad_norm": 0.4765938812802202, + "learning_rate": 6.849447405551718e-05, + "loss": 1.2226, + "step": 529 + }, + { + "epoch": 0.97, + "grad_norm": 0.53184473292579, + "learning_rate": 6.843457739857467e-05, + "loss": 1.2347, + "step": 530 + }, + { + "epoch": 0.97, + "grad_norm": 0.6416239346492343, + "learning_rate": 6.837455157468596e-05, + "loss": 1.2429, + "step": 531 + }, + { + "epoch": 0.97, + "grad_norm": 0.3188092712502773, + "learning_rate": 6.831439685652442e-05, + "loss": 1.216, + "step": 532 + }, + { + "epoch": 0.97, + "grad_norm": 0.3527495731006385, + "learning_rate": 6.825411351734895e-05, + "loss": 1.1682, + "step": 533 + }, + { + "epoch": 0.98, + "grad_norm": 0.29603753744741856, + "learning_rate": 6.819370183100274e-05, + "loss": 1.1434, + "step": 534 + }, + { + "epoch": 0.98, + "grad_norm": 0.5252450389976622, + "learning_rate": 6.813316207191198e-05, + "loss": 1.1943, + "step": 535 + }, + { + "epoch": 0.98, + "grad_norm": 0.32999419558659937, + "learning_rate": 6.807249451508466e-05, + "loss": 1.192, + "step": 536 + }, + { + "epoch": 0.98, + "grad_norm": 0.3650175469778724, + "learning_rate": 6.801169943610929e-05, + "loss": 1.2141, + "step": 537 + }, + { + "epoch": 0.98, + "grad_norm": 1.0643532150783557, + "learning_rate": 6.795077711115368e-05, + "loss": 1.2253, + "step": 538 + }, + { + "epoch": 0.99, + "grad_norm": 0.5041310609130145, + "learning_rate": 6.788972781696363e-05, + "loss": 1.278, + "step": 539 + }, + { + "epoch": 0.99, + "grad_norm": 0.5123058164360991, + "learning_rate": 6.782855183086177e-05, + "loss": 1.2231, + "step": 540 + }, + { + "epoch": 0.99, + "grad_norm": 0.3533015702394419, + "learning_rate": 6.776724943074619e-05, + "loss": 1.2072, + "step": 541 + }, + { + "epoch": 0.99, + "grad_norm": 0.30253964625417207, + "learning_rate": 6.770582089508927e-05, + "loss": 1.1382, + "step": 542 + }, + { + "epoch": 0.99, + "grad_norm": 0.348991618828202, + "learning_rate": 6.764426650293633e-05, + "loss": 1.2079, + "step": 543 + }, + { + "epoch": 0.99, + "grad_norm": 0.46017440578788743, + "learning_rate": 6.758258653390444e-05, + "loss": 1.1813, + "step": 544 + }, + { + "epoch": 1.0, + "grad_norm": 0.31962101755594885, + "learning_rate": 6.75207812681811e-05, + "loss": 1.1339, + "step": 545 + }, + { + "epoch": 1.0, + "grad_norm": 0.37092024548285923, + "learning_rate": 6.745885098652298e-05, + "loss": 1.2591, + "step": 546 + }, + { + "epoch": 1.0, + "grad_norm": 0.32347106450715835, + "learning_rate": 6.739679597025466e-05, + "loss": 1.2017, + "step": 547 + }, + { + "epoch": 1.0, + "grad_norm": 0.39250187112342494, + "learning_rate": 6.733461650126733e-05, + "loss": 1.0933, + "step": 548 + }, + { + "epoch": 1.0, + "grad_norm": 0.473522452217324, + "learning_rate": 6.727231286201752e-05, + "loss": 1.1124, + "step": 549 + }, + { + "epoch": 1.01, + "grad_norm": 0.4809062179622052, + "learning_rate": 6.720988533552582e-05, + "loss": 1.1585, + "step": 550 + }, + { + "epoch": 1.01, + "grad_norm": 0.3529662801059162, + "learning_rate": 6.714733420537559e-05, + "loss": 1.0501, + "step": 551 + }, + { + "epoch": 1.01, + "grad_norm": 0.5958247214391118, + "learning_rate": 6.708465975571168e-05, + "loss": 1.1086, + "step": 552 + }, + { + "epoch": 1.01, + "grad_norm": 0.5341364205022454, + "learning_rate": 6.70218622712391e-05, + "loss": 1.0518, + "step": 553 + }, + { + "epoch": 1.01, + "grad_norm": 0.3601805724462006, + "learning_rate": 6.695894203722181e-05, + "loss": 1.1779, + "step": 554 + }, + { + "epoch": 1.02, + "grad_norm": 0.43410190338280613, + "learning_rate": 6.68958993394813e-05, + "loss": 1.093, + "step": 555 + }, + { + "epoch": 1.02, + "grad_norm": 0.46217742572873594, + "learning_rate": 6.683273446439546e-05, + "loss": 1.0117, + "step": 556 + }, + { + "epoch": 1.02, + "grad_norm": 0.8591682373623357, + "learning_rate": 6.676944769889708e-05, + "loss": 1.1002, + "step": 557 + }, + { + "epoch": 1.02, + "grad_norm": 0.7383229487622726, + "learning_rate": 6.670603933047272e-05, + "loss": 1.0779, + "step": 558 + }, + { + "epoch": 1.02, + "grad_norm": 0.5965305891207813, + "learning_rate": 6.664250964716131e-05, + "loss": 1.0889, + "step": 559 + }, + { + "epoch": 1.02, + "grad_norm": 0.6030858606684543, + "learning_rate": 6.657885893755288e-05, + "loss": 1.0982, + "step": 560 + }, + { + "epoch": 1.03, + "grad_norm": 0.4644510682398409, + "learning_rate": 6.65150874907872e-05, + "loss": 1.1004, + "step": 561 + }, + { + "epoch": 1.03, + "grad_norm": 0.43943285132452564, + "learning_rate": 6.645119559655254e-05, + "loss": 1.0536, + "step": 562 + }, + { + "epoch": 1.03, + "grad_norm": 0.4456395978600012, + "learning_rate": 6.638718354508427e-05, + "loss": 1.0733, + "step": 563 + }, + { + "epoch": 1.03, + "grad_norm": 0.3303824433217466, + "learning_rate": 6.632305162716365e-05, + "loss": 1.0552, + "step": 564 + }, + { + "epoch": 1.03, + "grad_norm": 0.3617704823170143, + "learning_rate": 6.62588001341164e-05, + "loss": 1.1092, + "step": 565 + }, + { + "epoch": 1.04, + "grad_norm": 0.4465013349903427, + "learning_rate": 6.619442935781141e-05, + "loss": 1.0781, + "step": 566 + }, + { + "epoch": 1.04, + "grad_norm": 0.48516780613791277, + "learning_rate": 6.612993959065947e-05, + "loss": 1.0686, + "step": 567 + }, + { + "epoch": 1.04, + "grad_norm": 0.38867820318536633, + "learning_rate": 6.606533112561186e-05, + "loss": 1.1215, + "step": 568 + }, + { + "epoch": 1.04, + "grad_norm": 0.38566119820378336, + "learning_rate": 6.600060425615907e-05, + "loss": 1.1213, + "step": 569 + }, + { + "epoch": 1.04, + "grad_norm": 0.35534855445058544, + "learning_rate": 6.593575927632947e-05, + "loss": 1.0955, + "step": 570 + }, + { + "epoch": 1.04, + "grad_norm": 0.38124406233349717, + "learning_rate": 6.587079648068795e-05, + "loss": 1.0659, + "step": 571 + }, + { + "epoch": 1.05, + "grad_norm": 0.454750160923548, + "learning_rate": 6.580571616433457e-05, + "loss": 1.1149, + "step": 572 + }, + { + "epoch": 1.05, + "grad_norm": 0.35353190088025255, + "learning_rate": 6.574051862290325e-05, + "loss": 1.0388, + "step": 573 + }, + { + "epoch": 1.05, + "grad_norm": 0.3249395594793626, + "learning_rate": 6.567520415256045e-05, + "loss": 1.0784, + "step": 574 + }, + { + "epoch": 1.05, + "grad_norm": 0.40078898818247227, + "learning_rate": 6.560977305000375e-05, + "loss": 1.0859, + "step": 575 + }, + { + "epoch": 1.05, + "grad_norm": 0.4115264795060035, + "learning_rate": 6.554422561246054e-05, + "loss": 1.1828, + "step": 576 + }, + { + "epoch": 1.06, + "grad_norm": 0.30090229228069215, + "learning_rate": 6.54785621376867e-05, + "loss": 1.0901, + "step": 577 + }, + { + "epoch": 1.06, + "grad_norm": 0.28827860350299206, + "learning_rate": 6.541278292396523e-05, + "loss": 1.0277, + "step": 578 + }, + { + "epoch": 1.06, + "grad_norm": 0.34690404488996757, + "learning_rate": 6.534688827010484e-05, + "loss": 1.048, + "step": 579 + }, + { + "epoch": 1.06, + "grad_norm": 0.29943113556644785, + "learning_rate": 6.528087847543867e-05, + "loss": 1.0646, + "step": 580 + }, + { + "epoch": 1.06, + "grad_norm": 0.37318202575874415, + "learning_rate": 6.521475383982291e-05, + "loss": 1.1091, + "step": 581 + }, + { + "epoch": 1.06, + "grad_norm": 0.3049663659203959, + "learning_rate": 6.51485146636354e-05, + "loss": 1.0552, + "step": 582 + }, + { + "epoch": 1.07, + "grad_norm": 0.3342407867509692, + "learning_rate": 6.508216124777431e-05, + "loss": 1.2227, + "step": 583 + }, + { + "epoch": 1.07, + "grad_norm": 0.3348396047855952, + "learning_rate": 6.501569389365674e-05, + "loss": 1.0861, + "step": 584 + }, + { + "epoch": 1.07, + "grad_norm": 0.30951429367513383, + "learning_rate": 6.494911290321737e-05, + "loss": 1.0461, + "step": 585 + }, + { + "epoch": 1.07, + "grad_norm": 0.33898401361064606, + "learning_rate": 6.488241857890711e-05, + "loss": 1.0854, + "step": 586 + }, + { + "epoch": 1.07, + "grad_norm": 0.4901462068263497, + "learning_rate": 6.481561122369164e-05, + "loss": 1.1012, + "step": 587 + }, + { + "epoch": 1.08, + "grad_norm": 0.3179574879809652, + "learning_rate": 6.474869114105018e-05, + "loss": 1.0451, + "step": 588 + }, + { + "epoch": 1.08, + "grad_norm": 0.32159328915060714, + "learning_rate": 6.468165863497395e-05, + "loss": 1.0458, + "step": 589 + }, + { + "epoch": 1.08, + "grad_norm": 0.36462235008537297, + "learning_rate": 6.461451400996491e-05, + "loss": 1.1247, + "step": 590 + }, + { + "epoch": 1.08, + "grad_norm": 0.5373862753611778, + "learning_rate": 6.454725757103432e-05, + "loss": 1.0542, + "step": 591 + }, + { + "epoch": 1.08, + "grad_norm": 0.3160409270291303, + "learning_rate": 6.447988962370133e-05, + "loss": 1.0829, + "step": 592 + }, + { + "epoch": 1.08, + "grad_norm": 0.390452102978435, + "learning_rate": 6.441241047399169e-05, + "loss": 1.192, + "step": 593 + }, + { + "epoch": 1.09, + "grad_norm": 0.3802122712014928, + "learning_rate": 6.434482042843627e-05, + "loss": 1.1153, + "step": 594 + }, + { + "epoch": 1.09, + "grad_norm": 0.4081584328242501, + "learning_rate": 6.427711979406966e-05, + "loss": 1.1635, + "step": 595 + }, + { + "epoch": 1.09, + "grad_norm": 0.3791962989638633, + "learning_rate": 6.420930887842889e-05, + "loss": 1.1581, + "step": 596 + }, + { + "epoch": 1.09, + "grad_norm": 0.33239440056484193, + "learning_rate": 6.414138798955189e-05, + "loss": 1.0926, + "step": 597 + }, + { + "epoch": 1.09, + "grad_norm": 0.3279881540815014, + "learning_rate": 6.407335743597616e-05, + "loss": 1.1386, + "step": 598 + }, + { + "epoch": 1.1, + "grad_norm": 0.30309644763750837, + "learning_rate": 6.40052175267374e-05, + "loss": 1.0523, + "step": 599 + }, + { + "epoch": 1.1, + "grad_norm": 0.3349097308403333, + "learning_rate": 6.393696857136801e-05, + "loss": 1.0815, + "step": 600 + }, + { + "epoch": 1.1, + "grad_norm": 0.3288227593556618, + "learning_rate": 6.386861087989581e-05, + "loss": 1.015, + "step": 601 + }, + { + "epoch": 1.1, + "grad_norm": 0.36685586740843157, + "learning_rate": 6.380014476284255e-05, + "loss": 1.1232, + "step": 602 + }, + { + "epoch": 1.1, + "grad_norm": 0.3620977714204643, + "learning_rate": 6.373157053122243e-05, + "loss": 1.1138, + "step": 603 + }, + { + "epoch": 1.1, + "grad_norm": 0.3130587018197183, + "learning_rate": 6.366288849654091e-05, + "loss": 1.1255, + "step": 604 + }, + { + "epoch": 1.11, + "grad_norm": 0.3602737087072766, + "learning_rate": 6.359409897079303e-05, + "loss": 1.0282, + "step": 605 + }, + { + "epoch": 1.11, + "grad_norm": 0.31168852571991945, + "learning_rate": 6.352520226646222e-05, + "loss": 1.0779, + "step": 606 + }, + { + "epoch": 1.11, + "grad_norm": 0.3516045580189353, + "learning_rate": 6.345619869651871e-05, + "loss": 1.1028, + "step": 607 + }, + { + "epoch": 1.11, + "grad_norm": 0.3231857927563657, + "learning_rate": 6.33870885744182e-05, + "loss": 1.1202, + "step": 608 + }, + { + "epoch": 1.11, + "grad_norm": 0.30205205129701157, + "learning_rate": 6.331787221410041e-05, + "loss": 1.1369, + "step": 609 + }, + { + "epoch": 1.12, + "grad_norm": 0.3198359813888166, + "learning_rate": 6.32485499299877e-05, + "loss": 1.1763, + "step": 610 + }, + { + "epoch": 1.12, + "grad_norm": 0.3128641370321787, + "learning_rate": 6.31791220369835e-05, + "loss": 1.0223, + "step": 611 + }, + { + "epoch": 1.12, + "grad_norm": 0.2989105616213649, + "learning_rate": 6.31095888504711e-05, + "loss": 1.0358, + "step": 612 + }, + { + "epoch": 1.12, + "grad_norm": 0.3103537906853337, + "learning_rate": 6.303995068631203e-05, + "loss": 1.1261, + "step": 613 + }, + { + "epoch": 1.12, + "grad_norm": 0.28598715532508207, + "learning_rate": 6.297020786084467e-05, + "loss": 1.0629, + "step": 614 + }, + { + "epoch": 1.12, + "grad_norm": 0.29809789918093255, + "learning_rate": 6.290036069088288e-05, + "loss": 1.035, + "step": 615 + }, + { + "epoch": 1.13, + "grad_norm": 0.33765270252261453, + "learning_rate": 6.283040949371451e-05, + "loss": 1.1221, + "step": 616 + }, + { + "epoch": 1.13, + "grad_norm": 0.3424617501293415, + "learning_rate": 6.276035458709993e-05, + "loss": 1.155, + "step": 617 + }, + { + "epoch": 1.13, + "grad_norm": 0.3799189737987811, + "learning_rate": 6.269019628927067e-05, + "loss": 1.0701, + "step": 618 + }, + { + "epoch": 1.13, + "grad_norm": 0.3358898935253196, + "learning_rate": 6.261993491892791e-05, + "loss": 1.1649, + "step": 619 + }, + { + "epoch": 1.13, + "grad_norm": 0.31569979424117356, + "learning_rate": 6.254957079524099e-05, + "loss": 1.0633, + "step": 620 + }, + { + "epoch": 1.14, + "grad_norm": 0.3002168156888237, + "learning_rate": 6.247910423784609e-05, + "loss": 1.0846, + "step": 621 + }, + { + "epoch": 1.14, + "grad_norm": 0.3097238823450595, + "learning_rate": 6.24085355668447e-05, + "loss": 1.0808, + "step": 622 + }, + { + "epoch": 1.14, + "grad_norm": 0.3120312761417578, + "learning_rate": 6.233786510280212e-05, + "loss": 1.0142, + "step": 623 + }, + { + "epoch": 1.14, + "grad_norm": 0.3335343015064923, + "learning_rate": 6.22670931667461e-05, + "loss": 1.0674, + "step": 624 + }, + { + "epoch": 1.14, + "grad_norm": 0.3234062304634526, + "learning_rate": 6.219622008016533e-05, + "loss": 1.0981, + "step": 625 + }, + { + "epoch": 1.14, + "grad_norm": 0.32152678786547273, + "learning_rate": 6.212524616500798e-05, + "loss": 1.0244, + "step": 626 + }, + { + "epoch": 1.15, + "grad_norm": 0.39031977608147594, + "learning_rate": 6.205417174368023e-05, + "loss": 1.1205, + "step": 627 + }, + { + "epoch": 1.15, + "grad_norm": 0.3806189090017157, + "learning_rate": 6.198299713904485e-05, + "loss": 1.1134, + "step": 628 + }, + { + "epoch": 1.15, + "grad_norm": 0.2978349276971668, + "learning_rate": 6.191172267441967e-05, + "loss": 1.0088, + "step": 629 + }, + { + "epoch": 1.15, + "grad_norm": 0.3190354077382501, + "learning_rate": 6.184034867357617e-05, + "loss": 1.108, + "step": 630 + }, + { + "epoch": 1.15, + "grad_norm": 0.32633048665038994, + "learning_rate": 6.176887546073797e-05, + "loss": 1.0825, + "step": 631 + }, + { + "epoch": 1.16, + "grad_norm": 0.3428026413020903, + "learning_rate": 6.169730336057939e-05, + "loss": 1.0765, + "step": 632 + }, + { + "epoch": 1.16, + "grad_norm": 0.3475737151929015, + "learning_rate": 6.162563269822391e-05, + "loss": 1.0693, + "step": 633 + }, + { + "epoch": 1.16, + "grad_norm": 0.3870252154591392, + "learning_rate": 6.15538637992428e-05, + "loss": 1.1081, + "step": 634 + }, + { + "epoch": 1.16, + "grad_norm": 0.33597355193652834, + "learning_rate": 6.148199698965352e-05, + "loss": 1.0893, + "step": 635 + }, + { + "epoch": 1.16, + "grad_norm": 0.30805894179787247, + "learning_rate": 6.141003259591834e-05, + "loss": 1.0995, + "step": 636 + }, + { + "epoch": 1.17, + "grad_norm": 0.3025073882734066, + "learning_rate": 6.133797094494281e-05, + "loss": 1.0388, + "step": 637 + }, + { + "epoch": 1.17, + "grad_norm": 0.3524395196391662, + "learning_rate": 6.126581236407429e-05, + "loss": 1.1196, + "step": 638 + }, + { + "epoch": 1.17, + "grad_norm": 0.3377646188130345, + "learning_rate": 6.119355718110039e-05, + "loss": 1.0382, + "step": 639 + }, + { + "epoch": 1.17, + "grad_norm": 0.35508400659785483, + "learning_rate": 6.112120572424763e-05, + "loss": 1.1402, + "step": 640 + }, + { + "epoch": 1.17, + "grad_norm": 0.3454418793700457, + "learning_rate": 6.104875832217982e-05, + "loss": 1.1032, + "step": 641 + }, + { + "epoch": 1.17, + "grad_norm": 0.32629806837059866, + "learning_rate": 6.097621530399661e-05, + "loss": 1.0959, + "step": 642 + }, + { + "epoch": 1.18, + "grad_norm": 0.3329536837751315, + "learning_rate": 6.090357699923202e-05, + "loss": 1.0467, + "step": 643 + }, + { + "epoch": 1.18, + "grad_norm": 0.32302233828349475, + "learning_rate": 6.083084373785287e-05, + "loss": 1.0858, + "step": 644 + }, + { + "epoch": 1.18, + "grad_norm": 0.3310358826507611, + "learning_rate": 6.075801585025739e-05, + "loss": 1.0715, + "step": 645 + }, + { + "epoch": 1.18, + "grad_norm": 0.319322035854079, + "learning_rate": 6.068509366727362e-05, + "loss": 1.177, + "step": 646 + }, + { + "epoch": 1.18, + "grad_norm": 0.3065230667302707, + "learning_rate": 6.061207752015797e-05, + "loss": 1.0649, + "step": 647 + }, + { + "epoch": 1.19, + "grad_norm": 0.29926795565748227, + "learning_rate": 6.053896774059368e-05, + "loss": 1.1325, + "step": 648 + }, + { + "epoch": 1.19, + "grad_norm": 0.3556069634279046, + "learning_rate": 6.046576466068931e-05, + "loss": 1.1366, + "step": 649 + }, + { + "epoch": 1.19, + "grad_norm": 0.3189191131461966, + "learning_rate": 6.039246861297727e-05, + "loss": 1.0693, + "step": 650 + }, + { + "epoch": 1.19, + "grad_norm": 0.3347197156648834, + "learning_rate": 6.031907993041227e-05, + "loss": 1.1009, + "step": 651 + }, + { + "epoch": 1.19, + "grad_norm": 0.32274156348185445, + "learning_rate": 6.0245598946369826e-05, + "loss": 1.1675, + "step": 652 + }, + { + "epoch": 1.19, + "grad_norm": 0.35534089035455224, + "learning_rate": 6.017202599464476e-05, + "loss": 1.1723, + "step": 653 + }, + { + "epoch": 1.2, + "grad_norm": 0.3106026578570133, + "learning_rate": 6.009836140944965e-05, + "loss": 1.0954, + "step": 654 + }, + { + "epoch": 1.2, + "grad_norm": 0.3309144454564729, + "learning_rate": 6.002460552541331e-05, + "loss": 1.0209, + "step": 655 + }, + { + "epoch": 1.2, + "grad_norm": 0.3023619281400003, + "learning_rate": 5.9950758677579345e-05, + "loss": 1.0363, + "step": 656 + }, + { + "epoch": 1.2, + "grad_norm": 0.3311182880219704, + "learning_rate": 5.987682120140451e-05, + "loss": 1.0515, + "step": 657 + }, + { + "epoch": 1.2, + "grad_norm": 0.33396486010030413, + "learning_rate": 5.980279343275729e-05, + "loss": 1.1251, + "step": 658 + }, + { + "epoch": 1.21, + "grad_norm": 0.3465764556678002, + "learning_rate": 5.97286757079163e-05, + "loss": 1.165, + "step": 659 + }, + { + "epoch": 1.21, + "grad_norm": 0.304193441363374, + "learning_rate": 5.965446836356882e-05, + "loss": 1.0228, + "step": 660 + }, + { + "epoch": 1.21, + "grad_norm": 0.3415149030413082, + "learning_rate": 5.9580171736809224e-05, + "loss": 1.0742, + "step": 661 + }, + { + "epoch": 1.21, + "grad_norm": 0.33138658321132064, + "learning_rate": 5.950578616513746e-05, + "loss": 1.0843, + "step": 662 + }, + { + "epoch": 1.21, + "grad_norm": 0.30774403421162994, + "learning_rate": 5.943131198645752e-05, + "loss": 1.065, + "step": 663 + }, + { + "epoch": 1.21, + "grad_norm": 0.3428877492183819, + "learning_rate": 5.9356749539075885e-05, + "loss": 1.1101, + "step": 664 + }, + { + "epoch": 1.22, + "grad_norm": 0.3621290546130101, + "learning_rate": 5.928209916170003e-05, + "loss": 1.1372, + "step": 665 + }, + { + "epoch": 1.22, + "grad_norm": 0.3482375945469884, + "learning_rate": 5.9207361193436865e-05, + "loss": 1.132, + "step": 666 + }, + { + "epoch": 1.22, + "grad_norm": 0.31754384974068384, + "learning_rate": 5.9132535973791156e-05, + "loss": 1.148, + "step": 667 + }, + { + "epoch": 1.22, + "grad_norm": 0.36003834782050365, + "learning_rate": 5.9057623842664044e-05, + "loss": 1.1099, + "step": 668 + }, + { + "epoch": 1.22, + "grad_norm": 0.2963701622969662, + "learning_rate": 5.8982625140351464e-05, + "loss": 1.0755, + "step": 669 + }, + { + "epoch": 1.23, + "grad_norm": 0.32579569606066516, + "learning_rate": 5.8907540207542616e-05, + "loss": 1.0809, + "step": 670 + }, + { + "epoch": 1.23, + "grad_norm": 0.4247563451753457, + "learning_rate": 5.8832369385318416e-05, + "loss": 1.097, + "step": 671 + }, + { + "epoch": 1.23, + "grad_norm": 0.33076932102169776, + "learning_rate": 5.875711301514992e-05, + "loss": 1.1078, + "step": 672 + }, + { + "epoch": 1.23, + "grad_norm": 0.3609238032332309, + "learning_rate": 5.8681771438896815e-05, + "loss": 1.1031, + "step": 673 + }, + { + "epoch": 1.23, + "grad_norm": 0.325159585649425, + "learning_rate": 5.860634499880583e-05, + "loss": 1.0707, + "step": 674 + }, + { + "epoch": 1.23, + "grad_norm": 0.4620687271068983, + "learning_rate": 5.853083403750922e-05, + "loss": 1.1017, + "step": 675 + }, + { + "epoch": 1.24, + "grad_norm": 0.33485279064365936, + "learning_rate": 5.845523889802316e-05, + "loss": 1.0989, + "step": 676 + }, + { + "epoch": 1.24, + "grad_norm": 0.30952573170841513, + "learning_rate": 5.8379559923746214e-05, + "loss": 1.0393, + "step": 677 + }, + { + "epoch": 1.24, + "grad_norm": 0.33498605810588283, + "learning_rate": 5.830379745845781e-05, + "loss": 1.1259, + "step": 678 + }, + { + "epoch": 1.24, + "grad_norm": 0.35771921163037307, + "learning_rate": 5.822795184631659e-05, + "loss": 1.0815, + "step": 679 + }, + { + "epoch": 1.24, + "grad_norm": 0.3329650192347647, + "learning_rate": 5.815202343185894e-05, + "loss": 1.1344, + "step": 680 + }, + { + "epoch": 1.25, + "grad_norm": 0.3356634465845771, + "learning_rate": 5.807601255999736e-05, + "loss": 1.1297, + "step": 681 + }, + { + "epoch": 1.25, + "grad_norm": 0.3289442034151235, + "learning_rate": 5.7999919576018934e-05, + "loss": 1.022, + "step": 682 + }, + { + "epoch": 1.25, + "grad_norm": 0.3207007334784113, + "learning_rate": 5.7923744825583745e-05, + "loss": 1.0571, + "step": 683 + }, + { + "epoch": 1.25, + "grad_norm": 0.3582460325329284, + "learning_rate": 5.7847488654723304e-05, + "loss": 1.0778, + "step": 684 + }, + { + "epoch": 1.25, + "grad_norm": 0.3563317666176927, + "learning_rate": 5.777115140983899e-05, + "loss": 1.1003, + "step": 685 + }, + { + "epoch": 1.25, + "grad_norm": 3.4694912945702105, + "learning_rate": 5.769473343770047e-05, + "loss": 1.121, + "step": 686 + }, + { + "epoch": 1.26, + "grad_norm": 0.43002349520483113, + "learning_rate": 5.761823508544411e-05, + "loss": 1.0765, + "step": 687 + }, + { + "epoch": 1.26, + "grad_norm": 0.39467783104839754, + "learning_rate": 5.754165670057142e-05, + "loss": 1.0788, + "step": 688 + }, + { + "epoch": 1.26, + "grad_norm": 0.39629029674867916, + "learning_rate": 5.7464998630947464e-05, + "loss": 1.0812, + "step": 689 + }, + { + "epoch": 1.26, + "grad_norm": 0.3880152093965208, + "learning_rate": 5.738826122479929e-05, + "loss": 1.1228, + "step": 690 + }, + { + "epoch": 1.26, + "grad_norm": 0.3777874121959188, + "learning_rate": 5.7311444830714324e-05, + "loss": 1.0907, + "step": 691 + }, + { + "epoch": 1.27, + "grad_norm": 0.38004041653523696, + "learning_rate": 5.723454979763882e-05, + "loss": 1.1263, + "step": 692 + }, + { + "epoch": 1.27, + "grad_norm": 0.37049672627797636, + "learning_rate": 5.7157576474876246e-05, + "loss": 1.1438, + "step": 693 + }, + { + "epoch": 1.27, + "grad_norm": 0.32973606103437614, + "learning_rate": 5.7080525212085725e-05, + "loss": 1.0553, + "step": 694 + }, + { + "epoch": 1.27, + "grad_norm": 0.31674639252070325, + "learning_rate": 5.700339635928038e-05, + "loss": 1.06, + "step": 695 + }, + { + "epoch": 1.27, + "grad_norm": 0.32282199426553837, + "learning_rate": 5.692619026682588e-05, + "loss": 1.0841, + "step": 696 + }, + { + "epoch": 1.27, + "grad_norm": 0.4810882958061859, + "learning_rate": 5.684890728543869e-05, + "loss": 1.0803, + "step": 697 + }, + { + "epoch": 1.28, + "grad_norm": 0.3995638550178378, + "learning_rate": 5.6771547766184566e-05, + "loss": 1.1187, + "step": 698 + }, + { + "epoch": 1.28, + "grad_norm": 0.35264932960583484, + "learning_rate": 5.669411206047699e-05, + "loss": 1.0641, + "step": 699 + }, + { + "epoch": 1.28, + "grad_norm": 0.35240640524733, + "learning_rate": 5.661660052007547e-05, + "loss": 1.076, + "step": 700 + }, + { + "epoch": 1.28, + "grad_norm": 0.3540694609860389, + "learning_rate": 5.653901349708401e-05, + "loss": 1.1369, + "step": 701 + }, + { + "epoch": 1.28, + "grad_norm": 0.3196055112925304, + "learning_rate": 5.646135134394955e-05, + "loss": 1.0677, + "step": 702 + }, + { + "epoch": 1.29, + "grad_norm": 0.4214141007955914, + "learning_rate": 5.6383614413460266e-05, + "loss": 1.1139, + "step": 703 + }, + { + "epoch": 1.29, + "grad_norm": 0.3625611311798579, + "learning_rate": 5.630580305874402e-05, + "loss": 1.1845, + "step": 704 + }, + { + "epoch": 1.29, + "grad_norm": 0.3425208672181188, + "learning_rate": 5.62279176332668e-05, + "loss": 1.174, + "step": 705 + }, + { + "epoch": 1.29, + "grad_norm": 0.3108419862818321, + "learning_rate": 5.6149958490830996e-05, + "loss": 1.0331, + "step": 706 + }, + { + "epoch": 1.29, + "grad_norm": 0.3274644181571904, + "learning_rate": 5.607192598557394e-05, + "loss": 1.0664, + "step": 707 + }, + { + "epoch": 1.29, + "grad_norm": 0.346218197215145, + "learning_rate": 5.599382047196617e-05, + "loss": 1.2088, + "step": 708 + }, + { + "epoch": 1.3, + "grad_norm": 0.328497632267458, + "learning_rate": 5.591564230480989e-05, + "loss": 1.0287, + "step": 709 + }, + { + "epoch": 1.3, + "grad_norm": 0.3708173720611468, + "learning_rate": 5.583739183923732e-05, + "loss": 1.0883, + "step": 710 + }, + { + "epoch": 1.3, + "grad_norm": 0.3631427403535479, + "learning_rate": 5.575906943070915e-05, + "loss": 1.1155, + "step": 711 + }, + { + "epoch": 1.3, + "grad_norm": 0.3305201458598695, + "learning_rate": 5.5680675435012834e-05, + "loss": 1.0958, + "step": 712 + }, + { + "epoch": 1.3, + "grad_norm": 0.34978833532083714, + "learning_rate": 5.5602210208261036e-05, + "loss": 1.1437, + "step": 713 + }, + { + "epoch": 1.31, + "grad_norm": 0.3510553882510229, + "learning_rate": 5.552367410688999e-05, + "loss": 1.0941, + "step": 714 + }, + { + "epoch": 1.31, + "grad_norm": 0.3523747462465078, + "learning_rate": 5.544506748765789e-05, + "loss": 1.1289, + "step": 715 + }, + { + "epoch": 1.31, + "grad_norm": 0.38262637783927445, + "learning_rate": 5.5366390707643266e-05, + "loss": 1.099, + "step": 716 + }, + { + "epoch": 1.31, + "grad_norm": 0.38620065989073454, + "learning_rate": 5.528764412424334e-05, + "loss": 1.083, + "step": 717 + }, + { + "epoch": 1.31, + "grad_norm": 0.3401355276121096, + "learning_rate": 5.520882809517245e-05, + "loss": 1.028, + "step": 718 + }, + { + "epoch": 1.32, + "grad_norm": 0.3392061008943934, + "learning_rate": 5.512994297846039e-05, + "loss": 1.1083, + "step": 719 + }, + { + "epoch": 1.32, + "grad_norm": 0.34219480421015414, + "learning_rate": 5.505098913245077e-05, + "loss": 1.1108, + "step": 720 + }, + { + "epoch": 1.32, + "grad_norm": 0.3275058061553761, + "learning_rate": 5.497196691579945e-05, + "loss": 1.111, + "step": 721 + }, + { + "epoch": 1.32, + "grad_norm": 0.36800249746509384, + "learning_rate": 5.489287668747283e-05, + "loss": 1.1221, + "step": 722 + }, + { + "epoch": 1.32, + "grad_norm": 0.4129005533101575, + "learning_rate": 5.481371880674628e-05, + "loss": 1.0966, + "step": 723 + }, + { + "epoch": 1.32, + "grad_norm": 0.36563906596251655, + "learning_rate": 5.4734493633202505e-05, + "loss": 1.0927, + "step": 724 + }, + { + "epoch": 1.33, + "grad_norm": 0.3614650536839971, + "learning_rate": 5.465520152672986e-05, + "loss": 1.13, + "step": 725 + }, + { + "epoch": 1.33, + "grad_norm": 0.36419665098633497, + "learning_rate": 5.4575842847520765e-05, + "loss": 1.1183, + "step": 726 + }, + { + "epoch": 1.33, + "grad_norm": 0.34490689807258995, + "learning_rate": 5.449641795607005e-05, + "loss": 1.0919, + "step": 727 + }, + { + "epoch": 1.33, + "grad_norm": 0.3627643746876298, + "learning_rate": 5.441692721317334e-05, + "loss": 1.0411, + "step": 728 + }, + { + "epoch": 1.33, + "grad_norm": 0.323620411949565, + "learning_rate": 5.433737097992537e-05, + "loss": 1.0725, + "step": 729 + }, + { + "epoch": 1.34, + "grad_norm": 0.3521599501824965, + "learning_rate": 5.425774961771838e-05, + "loss": 1.0926, + "step": 730 + }, + { + "epoch": 1.34, + "grad_norm": 0.3302390546764222, + "learning_rate": 5.417806348824047e-05, + "loss": 1.0468, + "step": 731 + }, + { + "epoch": 1.34, + "grad_norm": 0.3833325802616019, + "learning_rate": 5.4098312953473956e-05, + "loss": 1.1291, + "step": 732 + }, + { + "epoch": 1.34, + "grad_norm": 0.3708621126835512, + "learning_rate": 5.401849837569372e-05, + "loss": 1.0887, + "step": 733 + }, + { + "epoch": 1.34, + "grad_norm": 0.3625834373416278, + "learning_rate": 5.393862011746555e-05, + "loss": 1.0981, + "step": 734 + }, + { + "epoch": 1.34, + "grad_norm": 0.3583343965080617, + "learning_rate": 5.385867854164451e-05, + "loss": 1.1021, + "step": 735 + }, + { + "epoch": 1.35, + "grad_norm": 0.34598320594096066, + "learning_rate": 5.377867401137332e-05, + "loss": 1.1376, + "step": 736 + }, + { + "epoch": 1.35, + "grad_norm": 0.3046382791315433, + "learning_rate": 5.369860689008066e-05, + "loss": 1.0206, + "step": 737 + }, + { + "epoch": 1.35, + "grad_norm": 0.34464948380043725, + "learning_rate": 5.3618477541479505e-05, + "loss": 1.1084, + "step": 738 + }, + { + "epoch": 1.35, + "grad_norm": 0.3203242519627101, + "learning_rate": 5.353828632956557e-05, + "loss": 1.0731, + "step": 739 + }, + { + "epoch": 1.35, + "grad_norm": 0.3431169960355163, + "learning_rate": 5.3458033618615516e-05, + "loss": 1.091, + "step": 740 + }, + { + "epoch": 1.36, + "grad_norm": 0.33492074521678705, + "learning_rate": 5.337771977318543e-05, + "loss": 1.1112, + "step": 741 + }, + { + "epoch": 1.36, + "grad_norm": 0.32576546585541344, + "learning_rate": 5.3297345158109086e-05, + "loss": 1.0993, + "step": 742 + }, + { + "epoch": 1.36, + "grad_norm": 0.3410007245037574, + "learning_rate": 5.3216910138496286e-05, + "loss": 1.094, + "step": 743 + }, + { + "epoch": 1.36, + "grad_norm": 0.34891180680896833, + "learning_rate": 5.313641507973128e-05, + "loss": 1.1331, + "step": 744 + }, + { + "epoch": 1.36, + "grad_norm": 0.37135766946717214, + "learning_rate": 5.3055860347471006e-05, + "loss": 1.1, + "step": 745 + }, + { + "epoch": 1.36, + "grad_norm": 0.3465019415478411, + "learning_rate": 5.297524630764349e-05, + "loss": 1.1256, + "step": 746 + }, + { + "epoch": 1.37, + "grad_norm": 0.37035388481626563, + "learning_rate": 5.289457332644615e-05, + "loss": 1.0366, + "step": 747 + }, + { + "epoch": 1.37, + "grad_norm": 0.33853883270759155, + "learning_rate": 5.281384177034421e-05, + "loss": 1.0547, + "step": 748 + }, + { + "epoch": 1.37, + "grad_norm": 0.364306618627317, + "learning_rate": 5.2733052006068897e-05, + "loss": 1.0768, + "step": 749 + }, + { + "epoch": 1.37, + "grad_norm": 0.4021754315731627, + "learning_rate": 5.2652204400615916e-05, + "loss": 1.1382, + "step": 750 + }, + { + "epoch": 1.37, + "grad_norm": 0.3332185389039008, + "learning_rate": 5.257129932124368e-05, + "loss": 1.0815, + "step": 751 + }, + { + "epoch": 1.38, + "grad_norm": 0.3453105709879854, + "learning_rate": 5.249033713547173e-05, + "loss": 1.1109, + "step": 752 + }, + { + "epoch": 1.38, + "grad_norm": 0.3385397539717797, + "learning_rate": 5.2409318211078966e-05, + "loss": 1.0529, + "step": 753 + }, + { + "epoch": 1.38, + "grad_norm": 0.33197994450130447, + "learning_rate": 5.232824291610206e-05, + "loss": 1.0721, + "step": 754 + }, + { + "epoch": 1.38, + "grad_norm": 0.32836289576124167, + "learning_rate": 5.224711161883375e-05, + "loss": 1.0459, + "step": 755 + }, + { + "epoch": 1.38, + "grad_norm": 0.32491620058831744, + "learning_rate": 5.216592468782117e-05, + "loss": 1.0897, + "step": 756 + }, + { + "epoch": 1.38, + "grad_norm": 0.3137879047811153, + "learning_rate": 5.2084682491864155e-05, + "loss": 1.096, + "step": 757 + }, + { + "epoch": 1.39, + "grad_norm": 0.3356938043023012, + "learning_rate": 5.200338540001364e-05, + "loss": 1.0827, + "step": 758 + }, + { + "epoch": 1.39, + "grad_norm": 0.36044340490819055, + "learning_rate": 5.192203378156984e-05, + "loss": 1.0617, + "step": 759 + }, + { + "epoch": 1.39, + "grad_norm": 0.34674262047888293, + "learning_rate": 5.184062800608077e-05, + "loss": 1.1267, + "step": 760 + }, + { + "epoch": 1.39, + "grad_norm": 0.32469442322149333, + "learning_rate": 5.1759168443340375e-05, + "loss": 1.1483, + "step": 761 + }, + { + "epoch": 1.39, + "grad_norm": 0.3290384307774216, + "learning_rate": 5.167765546338698e-05, + "loss": 1.047, + "step": 762 + }, + { + "epoch": 1.4, + "grad_norm": 0.31637612188770403, + "learning_rate": 5.1596089436501525e-05, + "loss": 1.0311, + "step": 763 + }, + { + "epoch": 1.4, + "grad_norm": 0.3168693829641207, + "learning_rate": 5.151447073320597e-05, + "loss": 1.1405, + "step": 764 + }, + { + "epoch": 1.4, + "grad_norm": 0.34322421571238926, + "learning_rate": 5.143279972426153e-05, + "loss": 1.1428, + "step": 765 + }, + { + "epoch": 1.4, + "grad_norm": 0.3291030435830325, + "learning_rate": 5.1351076780667026e-05, + "loss": 1.0473, + "step": 766 + }, + { + "epoch": 1.4, + "grad_norm": 0.33772039158758044, + "learning_rate": 5.1269302273657195e-05, + "loss": 1.0909, + "step": 767 + }, + { + "epoch": 1.4, + "grad_norm": 0.3802031736890876, + "learning_rate": 5.118747657470102e-05, + "loss": 1.1482, + "step": 768 + }, + { + "epoch": 1.41, + "grad_norm": 0.3296067628997962, + "learning_rate": 5.1105600055500025e-05, + "loss": 1.0085, + "step": 769 + }, + { + "epoch": 1.41, + "grad_norm": 0.3707139982828035, + "learning_rate": 5.102367308798658e-05, + "loss": 1.0746, + "step": 770 + }, + { + "epoch": 1.41, + "grad_norm": 0.3378537316757011, + "learning_rate": 5.094169604432225e-05, + "loss": 1.0482, + "step": 771 + }, + { + "epoch": 1.41, + "grad_norm": 0.4008417246255145, + "learning_rate": 5.085966929689601e-05, + "loss": 1.1065, + "step": 772 + }, + { + "epoch": 1.41, + "grad_norm": 0.3244385106988064, + "learning_rate": 5.077759321832271e-05, + "loss": 1.0827, + "step": 773 + }, + { + "epoch": 1.42, + "grad_norm": 0.37228575732812336, + "learning_rate": 5.0695468181441215e-05, + "loss": 1.1146, + "step": 774 + }, + { + "epoch": 1.42, + "grad_norm": 0.33761714797540276, + "learning_rate": 5.061329455931283e-05, + "loss": 1.092, + "step": 775 + }, + { + "epoch": 1.42, + "grad_norm": 0.3158158390913494, + "learning_rate": 5.053107272521955e-05, + "loss": 1.1058, + "step": 776 + }, + { + "epoch": 1.42, + "grad_norm": 0.3691501929738938, + "learning_rate": 5.044880305266239e-05, + "loss": 1.1599, + "step": 777 + }, + { + "epoch": 1.42, + "grad_norm": 0.33730914019805525, + "learning_rate": 5.0366485915359645e-05, + "loss": 1.0615, + "step": 778 + }, + { + "epoch": 1.42, + "grad_norm": 0.34970059240017, + "learning_rate": 5.0284121687245257e-05, + "loss": 1.1475, + "step": 779 + }, + { + "epoch": 1.43, + "grad_norm": 0.3374028029407197, + "learning_rate": 5.020171074246707e-05, + "loss": 1.0926, + "step": 780 + }, + { + "epoch": 1.43, + "grad_norm": 0.3350020681123992, + "learning_rate": 5.011925345538514e-05, + "loss": 1.1276, + "step": 781 + }, + { + "epoch": 1.43, + "grad_norm": 0.3224228965786606, + "learning_rate": 5.003675020057003e-05, + "loss": 1.0183, + "step": 782 + }, + { + "epoch": 1.43, + "grad_norm": 0.3357310714740298, + "learning_rate": 4.995420135280114e-05, + "loss": 1.1114, + "step": 783 + }, + { + "epoch": 1.43, + "grad_norm": 0.3590203255363759, + "learning_rate": 4.9871607287064966e-05, + "loss": 1.1504, + "step": 784 + }, + { + "epoch": 1.44, + "grad_norm": 0.33011195419611655, + "learning_rate": 4.9788968378553396e-05, + "loss": 1.0826, + "step": 785 + }, + { + "epoch": 1.44, + "grad_norm": 0.31088868195439445, + "learning_rate": 4.970628500266207e-05, + "loss": 1.0704, + "step": 786 + }, + { + "epoch": 1.44, + "grad_norm": 0.3144996103179409, + "learning_rate": 4.962355753498858e-05, + "loss": 1.1403, + "step": 787 + }, + { + "epoch": 1.44, + "grad_norm": 0.3147269555419068, + "learning_rate": 4.954078635133081e-05, + "loss": 1.0898, + "step": 788 + }, + { + "epoch": 1.44, + "grad_norm": 0.3280151747783868, + "learning_rate": 4.945797182768524e-05, + "loss": 1.1115, + "step": 789 + }, + { + "epoch": 1.44, + "grad_norm": 0.3551996569232493, + "learning_rate": 4.937511434024524e-05, + "loss": 1.1731, + "step": 790 + }, + { + "epoch": 1.45, + "grad_norm": 0.343863208057807, + "learning_rate": 4.9292214265399336e-05, + "loss": 1.0866, + "step": 791 + }, + { + "epoch": 1.45, + "grad_norm": 0.37316699385322466, + "learning_rate": 4.920927197972949e-05, + "loss": 1.1083, + "step": 792 + }, + { + "epoch": 1.45, + "grad_norm": 0.3635739774067832, + "learning_rate": 4.9126287860009453e-05, + "loss": 1.1393, + "step": 793 + }, + { + "epoch": 1.45, + "grad_norm": 0.3755910554972886, + "learning_rate": 4.9043262283202974e-05, + "loss": 1.1624, + "step": 794 + }, + { + "epoch": 1.45, + "grad_norm": 0.3635899120146823, + "learning_rate": 4.8960195626462145e-05, + "loss": 1.2095, + "step": 795 + }, + { + "epoch": 1.46, + "grad_norm": 0.3642202684342816, + "learning_rate": 4.8877088267125664e-05, + "loss": 1.1099, + "step": 796 + }, + { + "epoch": 1.46, + "grad_norm": 0.3339946548799316, + "learning_rate": 4.879394058271712e-05, + "loss": 1.1157, + "step": 797 + }, + { + "epoch": 1.46, + "grad_norm": 0.3457189703100475, + "learning_rate": 4.871075295094329e-05, + "loss": 1.129, + "step": 798 + }, + { + "epoch": 1.46, + "grad_norm": 0.3550931839691424, + "learning_rate": 4.862752574969241e-05, + "loss": 1.076, + "step": 799 + }, + { + "epoch": 1.46, + "grad_norm": 0.36139108917966734, + "learning_rate": 4.8544259357032475e-05, + "loss": 1.1577, + "step": 800 + }, + { + "epoch": 1.0, + "grad_norm": 0.39569703665247874, + "learning_rate": 4.8460954151209486e-05, + "loss": 1.0543, + "step": 801 + }, + { + "epoch": 1.0, + "grad_norm": 0.3879033670170866, + "learning_rate": 4.837761051064579e-05, + "loss": 1.0688, + "step": 802 + }, + { + "epoch": 1.01, + "grad_norm": 0.3796846713967255, + "learning_rate": 4.8294228813938285e-05, + "loss": 0.9911, + "step": 803 + }, + { + "epoch": 1.01, + "grad_norm": 0.4007831430409375, + "learning_rate": 4.8210809439856804e-05, + "loss": 1.0126, + "step": 804 + }, + { + "epoch": 1.01, + "grad_norm": 0.37588078665500885, + "learning_rate": 4.8127352767342276e-05, + "loss": 0.9302, + "step": 805 + }, + { + "epoch": 1.01, + "grad_norm": 0.4078509175013281, + "learning_rate": 4.8043859175505095e-05, + "loss": 0.9982, + "step": 806 + }, + { + "epoch": 1.01, + "grad_norm": 0.379096046185539, + "learning_rate": 4.7960329043623344e-05, + "loss": 1.0035, + "step": 807 + }, + { + "epoch": 1.01, + "grad_norm": 0.3813938568133554, + "learning_rate": 4.787676275114111e-05, + "loss": 0.9579, + "step": 808 + }, + { + "epoch": 1.02, + "grad_norm": 0.3686863564511168, + "learning_rate": 4.779316067766673e-05, + "loss": 1.0105, + "step": 809 + }, + { + "epoch": 1.02, + "grad_norm": 0.4263940878847523, + "learning_rate": 4.770952320297109e-05, + "loss": 1.0677, + "step": 810 + }, + { + "epoch": 1.02, + "grad_norm": 0.37178778374665006, + "learning_rate": 4.7625850706985886e-05, + "loss": 1.0019, + "step": 811 + }, + { + "epoch": 1.02, + "grad_norm": 0.36803355429187945, + "learning_rate": 4.7542143569801894e-05, + "loss": 0.9937, + "step": 812 + }, + { + "epoch": 1.02, + "grad_norm": 0.3897072472941179, + "learning_rate": 4.745840217166725e-05, + "loss": 1.0877, + "step": 813 + }, + { + "epoch": 1.03, + "grad_norm": 0.35571833841716255, + "learning_rate": 4.737462689298577e-05, + "loss": 1.0015, + "step": 814 + }, + { + "epoch": 1.03, + "grad_norm": 0.38930229991094323, + "learning_rate": 4.7290818114315086e-05, + "loss": 1.028, + "step": 815 + }, + { + "epoch": 1.03, + "grad_norm": 0.411005007105147, + "learning_rate": 4.72069762163651e-05, + "loss": 1.0068, + "step": 816 + }, + { + "epoch": 1.03, + "grad_norm": 0.3980240190337736, + "learning_rate": 4.7123101579996106e-05, + "loss": 0.9919, + "step": 817 + }, + { + "epoch": 1.03, + "grad_norm": 0.36369517703115467, + "learning_rate": 4.7039194586217136e-05, + "loss": 0.967, + "step": 818 + }, + { + "epoch": 1.03, + "grad_norm": 0.38591148840458894, + "learning_rate": 4.695525561618418e-05, + "loss": 0.9743, + "step": 819 + }, + { + "epoch": 1.04, + "grad_norm": 0.45873135108949337, + "learning_rate": 4.687128505119853e-05, + "loss": 1.0516, + "step": 820 + }, + { + "epoch": 1.04, + "grad_norm": 0.3866330351411308, + "learning_rate": 4.6787283272704966e-05, + "loss": 0.9939, + "step": 821 + }, + { + "epoch": 1.04, + "grad_norm": 0.4620340173291326, + "learning_rate": 4.670325066229009e-05, + "loss": 1.0526, + "step": 822 + }, + { + "epoch": 1.04, + "grad_norm": 0.38877454299870284, + "learning_rate": 4.661918760168052e-05, + "loss": 0.9904, + "step": 823 + }, + { + "epoch": 1.04, + "grad_norm": 0.3880489386116793, + "learning_rate": 4.653509447274121e-05, + "loss": 0.9623, + "step": 824 + }, + { + "epoch": 1.05, + "grad_norm": 0.3827392356186151, + "learning_rate": 4.6450971657473743e-05, + "loss": 1.0772, + "step": 825 + }, + { + "epoch": 1.05, + "grad_norm": 0.4132814641854327, + "learning_rate": 4.63668195380145e-05, + "loss": 1.0533, + "step": 826 + }, + { + "epoch": 1.05, + "grad_norm": 0.3703610182402835, + "learning_rate": 4.628263849663301e-05, + "loss": 0.9336, + "step": 827 + }, + { + "epoch": 1.05, + "grad_norm": 0.4152053683299823, + "learning_rate": 4.619842891573016e-05, + "loss": 0.9801, + "step": 828 + }, + { + "epoch": 1.05, + "grad_norm": 0.41791059043554274, + "learning_rate": 4.6114191177836514e-05, + "loss": 1.0617, + "step": 829 + }, + { + "epoch": 1.05, + "grad_norm": 0.46363896517299136, + "learning_rate": 4.6029925665610524e-05, + "loss": 0.9687, + "step": 830 + }, + { + "epoch": 1.06, + "grad_norm": 0.41141959057512445, + "learning_rate": 4.59456327618368e-05, + "loss": 1.0965, + "step": 831 + }, + { + "epoch": 1.06, + "grad_norm": 0.3789192764519836, + "learning_rate": 4.5861312849424386e-05, + "loss": 0.9793, + "step": 832 + }, + { + "epoch": 1.06, + "grad_norm": 0.4047291581107866, + "learning_rate": 4.5776966311405035e-05, + "loss": 1.0342, + "step": 833 + }, + { + "epoch": 1.06, + "grad_norm": 0.4425157400959256, + "learning_rate": 4.5692593530931416e-05, + "loss": 1.0892, + "step": 834 + }, + { + "epoch": 1.06, + "grad_norm": 0.3707332144806616, + "learning_rate": 4.560819489127545e-05, + "loss": 0.9815, + "step": 835 + }, + { + "epoch": 1.07, + "grad_norm": 0.3897444102572823, + "learning_rate": 4.552377077582646e-05, + "loss": 0.9884, + "step": 836 + }, + { + "epoch": 1.07, + "grad_norm": 0.42725787957019346, + "learning_rate": 4.543932156808959e-05, + "loss": 0.9972, + "step": 837 + }, + { + "epoch": 1.07, + "grad_norm": 0.40615269781820007, + "learning_rate": 4.535484765168386e-05, + "loss": 0.9529, + "step": 838 + }, + { + "epoch": 1.07, + "grad_norm": 0.3505829736050887, + "learning_rate": 4.527034941034063e-05, + "loss": 0.9492, + "step": 839 + }, + { + "epoch": 1.07, + "grad_norm": 0.36688064686440497, + "learning_rate": 4.51858272279017e-05, + "loss": 0.9592, + "step": 840 + }, + { + "epoch": 1.07, + "grad_norm": 0.4043468777955929, + "learning_rate": 4.5101281488317634e-05, + "loss": 1.048, + "step": 841 + }, + { + "epoch": 1.08, + "grad_norm": 0.3811489793242706, + "learning_rate": 4.501671257564602e-05, + "loss": 1.0138, + "step": 842 + }, + { + "epoch": 1.08, + "grad_norm": 0.39813004142325986, + "learning_rate": 4.49321208740497e-05, + "loss": 1.071, + "step": 843 + }, + { + "epoch": 1.08, + "grad_norm": 0.3809751022095503, + "learning_rate": 4.484750676779504e-05, + "loss": 1.0351, + "step": 844 + }, + { + "epoch": 1.08, + "grad_norm": 0.384312178013823, + "learning_rate": 4.4762870641250185e-05, + "loss": 0.9737, + "step": 845 + }, + { + "epoch": 1.08, + "grad_norm": 0.40769404907923557, + "learning_rate": 4.467821287888331e-05, + "loss": 0.9659, + "step": 846 + }, + { + "epoch": 1.09, + "grad_norm": 0.39594136851937817, + "learning_rate": 4.459353386526086e-05, + "loss": 0.9405, + "step": 847 + }, + { + "epoch": 1.09, + "grad_norm": 0.37180161011562185, + "learning_rate": 4.450883398504584e-05, + "loss": 1.0732, + "step": 848 + }, + { + "epoch": 1.09, + "grad_norm": 0.3772603623154663, + "learning_rate": 4.442411362299602e-05, + "loss": 0.9646, + "step": 849 + }, + { + "epoch": 1.09, + "grad_norm": 0.4346142368506476, + "learning_rate": 4.433937316396224e-05, + "loss": 0.9572, + "step": 850 + } + ], + "logging_steps": 1.0, + "max_steps": 1638, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 50, + "total_flos": 881130794385408.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-850/training_args.bin b/checkpoint-850/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..8c2dfa20e1da5754719c3d7e300b9b86407f077f --- /dev/null +++ b/checkpoint-850/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3f2f7bd873b9dca108c5ca2e32ea140480fabeed2dec60f702daabd0a44d071e +size 6776 diff --git a/checkpoint-850/zero_to_fp32.py b/checkpoint-850/zero_to_fp32.py new file mode 100755 index 0000000000000000000000000000000000000000..24cc342e78d1a006c782b3a4cd68d9ce786d8fd8 --- /dev/null +++ b/checkpoint-850/zero_to_fp32.py @@ -0,0 +1,604 @@ +#!/usr/bin/env python + +# Copyright (c) Microsoft Corporation. +# SPDX-License-Identifier: Apache-2.0 + +# DeepSpeed Team + +# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets +# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in +# the future. Once extracted, the weights don't require DeepSpeed and can be used in any +# application. +# +# example: python zero_to_fp32.py . pytorch_model.bin + +import argparse +import torch +import glob +import math +import os +import re +from collections import OrderedDict +from dataclasses import dataclass + +# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with +# DeepSpeed data structures it has to be available in the current python environment. +from deepspeed.utils import logger +from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS, + FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES, + FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS) + + +@dataclass +class zero_model_state: + buffers: dict() + param_shapes: dict() + shared_params: list + ds_version: int + frozen_param_shapes: dict() + frozen_param_fragments: dict() + + +debug = 0 + +# load to cpu +device = torch.device('cpu') + + +def atoi(text): + return int(text) if text.isdigit() else text + + +def natural_keys(text): + ''' + alist.sort(key=natural_keys) sorts in human order + http://nedbatchelder.com/blog/200712/human_sorting.html + (See Toothy's implementation in the comments) + ''' + return [atoi(c) for c in re.split(r'(\d+)', text)] + + +def get_model_state_file(checkpoint_dir, zero_stage): + if not os.path.isdir(checkpoint_dir): + raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist") + + # there should be only one file + if zero_stage <= 2: + file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt") + elif zero_stage == 3: + file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt") + + if not os.path.exists(file): + raise FileNotFoundError(f"can't find model states file at '{file}'") + + return file + + +def get_checkpoint_files(checkpoint_dir, glob_pattern): + # XXX: need to test that this simple glob rule works for multi-node setup too + ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys) + + if len(ckpt_files) == 0: + raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'") + + return ckpt_files + + +def get_optim_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt") + + +def get_model_state_files(checkpoint_dir): + return get_checkpoint_files(checkpoint_dir, "*_model_states.pt") + + +def parse_model_states(files): + zero_model_states = [] + for file in files: + state_dict = torch.load(file, map_location=device) + + if BUFFER_NAMES not in state_dict: + raise ValueError(f"{file} is not a model state checkpoint") + buffer_names = state_dict[BUFFER_NAMES] + if debug: + print("Found buffers:", buffer_names) + + # recover just the buffers while restoring them to fp32 if they were saved in fp16 + buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names} + param_shapes = state_dict[PARAM_SHAPES] + + # collect parameters that are included in param_shapes + param_names = [] + for s in param_shapes: + for name in s.keys(): + param_names.append(name) + + # update with frozen parameters + frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None) + if frozen_param_shapes is not None: + if debug: + print(f"Found frozen_param_shapes: {frozen_param_shapes}") + param_names += list(frozen_param_shapes.keys()) + + # handle shared params + shared_params = [[k, v] for k, v in state_dict["shared_params"].items()] + + ds_version = state_dict.get(DS_VERSION, None) + + frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None) + + z_model_state = zero_model_state(buffers=buffers, + param_shapes=param_shapes, + shared_params=shared_params, + ds_version=ds_version, + frozen_param_shapes=frozen_param_shapes, + frozen_param_fragments=frozen_param_fragments) + zero_model_states.append(z_model_state) + + return zero_model_states + + +def parse_optim_states(files, ds_checkpoint_dir): + + total_files = len(files) + state_dicts = [] + for f in files: + state_dict = torch.load(f, map_location=device) + # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights + # and also handle the case where it was already removed by another helper script + state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None) + state_dicts.append(state_dict) + + if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]: + raise ValueError(f"{files[0]} is not a zero checkpoint") + zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE] + world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT] + + # For ZeRO-2 each param group can have different partition_count as data parallelism for expert + # parameters can be different from data parallelism for non-expert parameters. So we can just + # use the max of the partition_count to get the dp world_size. + + if type(world_size) is list: + world_size = max(world_size) + + if world_size != total_files: + raise ValueError( + f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. " + "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes." + ) + + # the groups are named differently in each stage + if zero_stage <= 2: + fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS + elif zero_stage == 3: + fp32_groups_key = FP32_FLAT_GROUPS + else: + raise ValueError(f"unknown zero stage {zero_stage}") + + if zero_stage <= 2: + fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))] + elif zero_stage == 3: + # if there is more than one param group, there will be multiple flattened tensors - one + # flattened tensor per group - for simplicity merge them into a single tensor + # + # XXX: could make the script more memory efficient for when there are multiple groups - it + # will require matching the sub-lists of param_shapes for each param group flattened tensor + + fp32_flat_groups = [ + torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts)) + ] + + return zero_stage, world_size, fp32_flat_groups + + +def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters): + """ + Returns fp32 state_dict reconstructed from ds checkpoint + + Args: + - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are) + + """ + print(f"Processing zero checkpoint '{ds_checkpoint_dir}'") + + optim_files = get_optim_files(ds_checkpoint_dir) + zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir) + print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}") + + model_files = get_model_state_files(ds_checkpoint_dir) + + zero_model_states = parse_model_states(model_files) + print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}') + + if zero_stage <= 2: + return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + elif zero_stage == 3: + return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters) + + +def _zero2_merge_frozen_params(state_dict, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + frozen_param_fragments = zero_model_states[0].frozen_param_fragments + + if debug: + num_elem = sum(s.numel() for s in frozen_param_shapes.values()) + print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in frozen_param_fragments.values()]) + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + state_dict[name] = frozen_param_fragments[name] + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _has_callable(obj, fn): + attr = getattr(obj, fn, None) + return callable(attr) + + +def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + + # Reconstruction protocol: + # + # XXX: document this + + if debug: + for i in range(world_size): + for j in range(len(fp32_flat_groups[0])): + print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}") + + # XXX: memory usage doubles here (zero2) + num_param_groups = len(fp32_flat_groups[0]) + merged_single_partition_of_fp32_groups = [] + for i in range(num_param_groups): + merged_partitions = [sd[i] for sd in fp32_flat_groups] + full_single_fp32_vector = torch.cat(merged_partitions, 0) + merged_single_partition_of_fp32_groups.append(full_single_fp32_vector) + avail_numel = sum( + [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups]) + + if debug: + wanted_params = sum([len(shapes) for shapes in param_shapes]) + wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes]) + # not asserting if there is a mismatch due to possible padding + print(f"Have {avail_numel} numels to process.") + print(f"Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + total_numel = 0 + total_params = 0 + for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups): + offset = 0 + avail_numel = full_single_fp32_vector.numel() + for name, shape in shapes.items(): + + unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape) + total_numel += unpartitioned_numel + total_params += 1 + + if debug: + print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ") + state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape) + offset += unpartitioned_numel + + # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and + # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex + # paddings performed in the code it's almost impossible to predict the exact numbers w/o the + # live optimizer object, so we are checking that the numbers are within the right range + align_to = 2 * world_size + + def zero2_align(x): + return align_to * math.ceil(x / align_to) + + if debug: + print(f"original offset={offset}, avail_numel={avail_numel}") + + offset = zero2_align(offset) + avail_numel = zero2_align(avail_numel) + + if debug: + print(f"aligned offset={offset}, avail_numel={avail_numel}") + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero2_merge_frozen_params(state_dict, zero_model_states) + + _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def zero3_partitioned_param_info(unpartitioned_numel, world_size): + remainder = unpartitioned_numel % world_size + padding_numel = (world_size - remainder) if remainder else 0 + partitioned_numel = math.ceil(unpartitioned_numel / world_size) + return partitioned_numel, padding_numel + + +def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states): + if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0: + return + + if debug: + for i in range(world_size): + num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values()) + print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}') + + frozen_param_shapes = zero_model_states[0].frozen_param_shapes + wanted_params = len(frozen_param_shapes) + wanted_numel = sum(s.numel() for s in frozen_param_shapes.values()) + avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size + print(f'Frozen params: Have {avail_numel} numels to process.') + print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params') + + total_params = 0 + total_numel = 0 + for name, shape in zero_model_states[0].frozen_param_shapes.items(): + total_params += 1 + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + + param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states) + state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape) + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements") + + +def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states): + param_shapes = zero_model_states[0].param_shapes + avail_numel = fp32_flat_groups[0].numel() * world_size + # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each + # param, re-consolidating each param, while dealing with padding if any + + # merge list of dicts, preserving order + param_shapes = {k: v for d in param_shapes for k, v in d.items()} + + if debug: + for i in range(world_size): + print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}") + + wanted_params = len(param_shapes) + wanted_numel = sum(shape.numel() for shape in param_shapes.values()) + # not asserting if there is a mismatch due to possible padding + avail_numel = fp32_flat_groups[0].numel() * world_size + print(f"Trainable params: Have {avail_numel} numels to process.") + print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.") + + # params + # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support + # out-of-core computing solution + offset = 0 + total_numel = 0 + total_params = 0 + for name, shape in param_shapes.items(): + + unpartitioned_numel = shape.numel() + total_numel += unpartitioned_numel + total_params += 1 + + partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size) + + if debug: + print( + f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}" + ) + + # XXX: memory usage doubles here + state_dict[name] = torch.cat( + tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)), + 0).narrow(0, 0, unpartitioned_numel).view(shape) + offset += partitioned_numel + + offset *= world_size + + # Sanity check + if offset != avail_numel: + raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong") + + print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements") + + +def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states, + exclude_frozen_parameters): + state_dict = OrderedDict() + + # buffers + buffers = zero_model_states[0].buffers + state_dict.update(buffers) + if debug: + print(f"added {len(buffers)} buffers") + + if not exclude_frozen_parameters: + _zero3_merge_frozen_params(state_dict, world_size, zero_model_states) + + _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states) + + # recover shared parameters + for pair in zero_model_states[0].shared_params: + if pair[1] in state_dict: + state_dict[pair[0]] = state_dict[pair[1]] + + return state_dict + + +def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with + ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example + via a model hub. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + + Returns: + - pytorch ``state_dict`` + + Note: this approach may not work if your application doesn't have sufficient free CPU memory and + you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with + the checkpoint. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint + # do the training and checkpoint saving + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu + model = model.cpu() # move to cpu + model.load_state_dict(state_dict) + # submit to model hub or save the model to share with others + + In this example the ``model`` will no longer be usable in the deepspeed context of the same + application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead. + + """ + if tag is None: + latest_path = os.path.join(checkpoint_dir, 'latest') + if os.path.isfile(latest_path): + with open(latest_path, 'r') as fd: + tag = fd.read().strip() + else: + raise ValueError(f"Unable to find 'latest' file at {latest_path}") + + ds_checkpoint_dir = os.path.join(checkpoint_dir, tag) + + if not os.path.isdir(ds_checkpoint_dir): + raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist") + + return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters) + + +def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False): + """ + Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be + loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed. + + Args: + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + - ``exclude_frozen_parameters``: exclude frozen parameters + """ + + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters) + print(f"Saving fp32 state dict to {output_file}") + torch.save(state_dict, output_file) + + +def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None): + """ + 1. Put the provided model to cpu + 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` + 3. Load it into the provided model + + Args: + - ``model``: the model object to update + - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``) + - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14`` + + Returns: + - ``model`: modified model + + Make sure you have plenty of CPU memory available before you call this function. If you don't + have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it + conveniently placed for you in the checkpoint folder. + + A typical usage might be :: + + from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint + model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir) + # submit to model hub or save the model to share with others + + Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context + of the same application. i.e. you will need to re-initialize the deepspeed engine, since + ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it. + + """ + logger.info(f"Extracting fp32 weights") + state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag) + + logger.info(f"Overwriting model with fp32 weights") + model = model.cpu() + model.load_state_dict(state_dict, strict=False) + + return model + + +if __name__ == "__main__": + + parser = argparse.ArgumentParser() + parser.add_argument("checkpoint_dir", + type=str, + help="path to the desired checkpoint folder, e.g., path/checkpoint-12") + parser.add_argument( + "output_file", + type=str, + help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)") + parser.add_argument("-t", + "--tag", + type=str, + default=None, + help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1") + parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters") + parser.add_argument("-d", "--debug", action='store_true', help="enable debug") + args = parser.parse_args() + + debug = args.debug + + convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, + args.output_file, + tag=args.tag, + exclude_frozen_parameters=args.exclude_frozen_parameters)