final save

.gitattributes

@@ -34,3 +34,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
 tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-3200/tokenizer.json filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,59 @@
+---
+library_name: transformers
+base_model: Qwen/Qwen2.5-Coder-1.5B-Instruct
+tags:
+- llama-factory
+- generated_from_trainer
+model-index:
+- name: train-armv8-O2_epoch3_AMD
+  results: []
+---
+
+<!-- This model card has been generated automatically according to the information the Trainer had access to. You
+should probably proofread and complete it, then remove this comment. -->
+
+# train-armv8-O2_epoch3_AMD
+
+This model is a fine-tuned version of [saves/train-armv8-O2_epoch3/checkpoint-2400](https://huggingface.co/saves/train-armv8-O2_epoch3/checkpoint-2400) on the None dataset.
+
+## Model description
+
+More information needed
+
+## Intended uses & limitations
+
+More information needed
+
+## Training and evaluation data
+
+More information needed
+
+## Training procedure
+
+### Training hyperparameters
+
+The following hyperparameters were used during training:
+- learning_rate: 2e-05
+- train_batch_size: 8
+- eval_batch_size: 8
+- seed: 42
+- distributed_type: multi-GPU
+- num_devices: 8
+- gradient_accumulation_steps: 8
+- total_train_batch_size: 512
+- total_eval_batch_size: 64
+- optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
+- lr_scheduler_type: cosine
+- lr_scheduler_warmup_ratio: 0.1
+- num_epochs: 2.0
+
+### Training results
+
+
+
+### Framework versions
+
+- Transformers 4.55.0
+- Pytorch 2.8.0+rocm6.3
+- Datasets 3.6.0
+- Tokenizers 0.21.1

checkpoint-3200/added_tokens.json

@@ -0,0 +1,24 @@
+{
+  "</tool_call>": 151658,
+  "<tool_call>": 151657,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

checkpoint-3200/chat_template.jinja

@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}

checkpoint-3200/config.json

@@ -0,0 +1,58 @@
+{
+  "architectures": [
+    "Qwen2ForCausalLM"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151645,
+  "hidden_act": "silu",
+  "hidden_size": 1536,
+  "initializer_range": 0.02,
+  "intermediate_size": 8960,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen2",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 2,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000.0,
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.55.0",
+  "use_cache": false,
+  "use_sliding_window": false,
+  "vocab_size": 151936
+}

checkpoint-3200/generation_config.json

@@ -0,0 +1,14 @@
+{
+  "bos_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "repetition_penalty": 1.1,
+  "temperature": 0.7,
+  "top_k": 20,
+  "top_p": 0.8,
+  "transformers_version": "4.55.0"
+}

checkpoint-3200/global_step3199/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:635b447fd89d8445f6f51c5ec83b2468c4c9c34fc3f204c39255f9c74a4970e4
+size 2315578373

checkpoint-3200/global_step3199/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e7fdad219aa8be8a3f6485c8f3d923a49b61aea5330f61ebebea70080c36d97
+size 2315580229

checkpoint-3200/global_step3199/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:b1625a075fc78ca94a2ec3305d6e19eb02368fea27dee991a4845a5eb208cf6f
+size 2315580485

checkpoint-3200/global_step3199/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:82d2190996e4af13b71b4d00f2222aa0664fc8c0494089a9dd863f22b08704aa
+size 2315580485

checkpoint-3200/global_step3199/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:39c2ec34b7349dbf7b15256c33b1a252e61e3d196b469eec468484444c1fc974
+size 2315580485

checkpoint-3200/global_step3199/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:cb623eec884b135f5f9609f6bfd5a8725b143fca9cab7f4d23aadf18f4ba8203
+size 2315580485

checkpoint-3200/global_step3199/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:c200ed485efe31e148107b4429bac949f57960008d43e50896f99bd1b14b4c63
+size 2315580613

checkpoint-3200/global_step3199/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:ccc1cb052fc0ed34694c2401266a57470b27b6a08087aad407d5f7dc7bad8f43
+size 2315580357

checkpoint-3200/global_step3199/mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c07ba1138b35b820ec8dead83d6f2c774263bac566dcdc6aea3f3af55ca2794e
+size 3087519461

checkpoint-3200/latest

@@ -0,0 +1 @@
+global_step3199

checkpoint-3200/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:68568245a0b436d8cfe998cfc30ef096be55953626ad688db4acd2c1b6604aab
+size 3554214752

checkpoint-3200/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b0418786c564bb001fc5f4215375b1a99ed3ea3f30771057e1d8682214a0108f
+size 16389

checkpoint-3200/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:35deaabe10396919e53b2274736981b26bc690165c8839845f178b355a6a7588
+size 16389

checkpoint-3200/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1c5deee061a76a9e05b4634f5abad786b4042e744cc4c156a2ab76dc3c94fcac
+size 16389

checkpoint-3200/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:13579f7ca365ea6f9cef83929c34833cd97f9b5764a434989b4e3e2db87dce56
+size 16389

checkpoint-3200/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:073d0b1d7e2e3930e5c29fc418e6246462ab2a00396c3a86e1448d373503f1b2
+size 16389

checkpoint-3200/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7c60731015889939e665a308a0933ac6c327734a4699cf5615217db0cfaf03ca
+size 16389

checkpoint-3200/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:416648e7a35618f28d73625751a8fc2ad066533414d19f6fc0c39932429023fc
+size 16389

checkpoint-3200/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:619b5c49446c94dca2b233ad17098e42f7d58a945c2eced1aa23e50469a75431
+size 16389

checkpoint-3200/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e1c553521506bff85c2c06b3ec44627f7dd741cb0c5568884f7e3492dafb7de
+size 1465

checkpoint-3200/special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

checkpoint-3200/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9c5ae00e602b8860cbd784ba82a8aa14e8feecec692e7076590d014d7b7fdafa
+size 11421896

checkpoint-3200/tokenizer_config.json

@@ -0,0 +1,208 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 32768,
+  "pad_token": "<|endoftext|>",
+  "padding_side": "right",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

checkpoint-3200/trainer_state.json

@@ -0,0 +1,2274 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9782037409182254,
+  "eval_steps": 500,
+  "global_step": 3200,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.006183335909723296,
+      "grad_norm": 0.04143473133444786,
+      "learning_rate": 5.555555555555555e-07,
+      "loss": 0.0078,
+      "step": 10
+    },
+    {
+      "epoch": 0.012366671819446591,
+      "grad_norm": 0.039171501994132996,
+      "learning_rate": 1.1728395061728396e-06,
+      "loss": 0.0073,
+      "step": 20
+    },
+    {
+      "epoch": 0.018550007729169888,
+      "grad_norm": 0.045620836317539215,
+      "learning_rate": 1.7901234567901235e-06,
+      "loss": 0.0077,
+      "step": 30
+    },
+    {
+      "epoch": 0.024733343638893183,
+      "grad_norm": 0.042504891753196716,
+      "learning_rate": 2.4074074074074075e-06,
+      "loss": 0.0074,
+      "step": 40
+    },
+    {
+      "epoch": 0.030916679548616478,
+      "grad_norm": 0.04532310739159584,
+      "learning_rate": 3.0246913580246917e-06,
+      "loss": 0.0078,
+      "step": 50
+    },
+    {
+      "epoch": 0.037100015458339776,
+      "grad_norm": 0.046271760016679764,
+      "learning_rate": 3.641975308641976e-06,
+      "loss": 0.008,
+      "step": 60
+    },
+    {
+      "epoch": 0.04328335136806307,
+      "grad_norm": 0.046998776495456696,
+      "learning_rate": 4.2592592592592596e-06,
+      "loss": 0.0082,
+      "step": 70
+    },
+    {
+      "epoch": 0.049466687277786366,
+      "grad_norm": 0.06645342707633972,
+      "learning_rate": 4.876543209876544e-06,
+      "loss": 0.008,
+      "step": 80
+    },
+    {
+      "epoch": 0.05565002318750966,
+      "grad_norm": 0.05185055360198021,
+      "learning_rate": 5.493827160493828e-06,
+      "loss": 0.0077,
+      "step": 90
+    },
+    {
+      "epoch": 0.061833359097232955,
+      "grad_norm": 0.06297408044338226,
+      "learning_rate": 6.111111111111112e-06,
+      "loss": 0.0084,
+      "step": 100
+    },
+    {
+      "epoch": 0.06801669500695626,
+      "grad_norm": 0.08234790712594986,
+      "learning_rate": 6.728395061728395e-06,
+      "loss": 0.008,
+      "step": 110
+    },
+    {
+      "epoch": 0.07420003091667955,
+      "grad_norm": 0.05353984236717224,
+      "learning_rate": 7.34567901234568e-06,
+      "loss": 0.0083,
+      "step": 120
+    },
+    {
+      "epoch": 0.08038336682640285,
+      "grad_norm": 0.060335684567689896,
+      "learning_rate": 7.962962962962963e-06,
+      "loss": 0.0081,
+      "step": 130
+    },
+    {
+      "epoch": 0.08656670273612614,
+      "grad_norm": 0.06284262984991074,
+      "learning_rate": 8.580246913580249e-06,
+      "loss": 0.0084,
+      "step": 140
+    },
+    {
+      "epoch": 0.09275003864584944,
+      "grad_norm": 0.08134513348340988,
+      "learning_rate": 9.197530864197531e-06,
+      "loss": 0.0087,
+      "step": 150
+    },
+    {
+      "epoch": 0.09893337455557273,
+      "grad_norm": 0.06377002596855164,
+      "learning_rate": 9.814814814814815e-06,
+      "loss": 0.0087,
+      "step": 160
+    },
+    {
+      "epoch": 0.10511671046529603,
+      "grad_norm": 0.07254844158887863,
+      "learning_rate": 1.04320987654321e-05,
+      "loss": 0.0089,
+      "step": 170
+    },
+    {
+      "epoch": 0.11130004637501932,
+      "grad_norm": 0.07614797353744507,
+      "learning_rate": 1.1049382716049384e-05,
+      "loss": 0.0091,
+      "step": 180
+    },
+    {
+      "epoch": 0.11748338228474262,
+      "grad_norm": 0.09067586809396744,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 0.0084,
+      "step": 190
+    },
+    {
+      "epoch": 0.12366671819446591,
+      "grad_norm": 0.06307026743888855,
+      "learning_rate": 1.228395061728395e-05,
+      "loss": 0.0095,
+      "step": 200
+    },
+    {
+      "epoch": 0.1298500541041892,
+      "grad_norm": 0.05498047545552254,
+      "learning_rate": 1.2901234567901235e-05,
+      "loss": 0.0087,
+      "step": 210
+    },
+    {
+      "epoch": 0.13603339001391251,
+      "grad_norm": 0.08366186171770096,
+      "learning_rate": 1.351851851851852e-05,
+      "loss": 0.009,
+      "step": 220
+    },
+    {
+      "epoch": 0.1422167259236358,
+      "grad_norm": 0.08589412271976471,
+      "learning_rate": 1.4135802469135805e-05,
+      "loss": 0.009,
+      "step": 230
+    },
+    {
+      "epoch": 0.1484000618333591,
+      "grad_norm": 0.05334419757127762,
+      "learning_rate": 1.4753086419753087e-05,
+      "loss": 0.0093,
+      "step": 240
+    },
+    {
+      "epoch": 0.15458339774308238,
+      "grad_norm": 0.06228677183389664,
+      "learning_rate": 1.537037037037037e-05,
+      "loss": 0.0091,
+      "step": 250
+    },
+    {
+      "epoch": 0.1607667336528057,
+      "grad_norm": 0.059001028537750244,
+      "learning_rate": 1.5987654320987655e-05,
+      "loss": 0.0088,
+      "step": 260
+    },
+    {
+      "epoch": 0.16695006956252897,
+      "grad_norm": 0.056203559041023254,
+      "learning_rate": 1.660493827160494e-05,
+      "loss": 0.0092,
+      "step": 270
+    },
+    {
+      "epoch": 0.17313340547225228,
+      "grad_norm": 0.08194007724523544,
+      "learning_rate": 1.7222222222222224e-05,
+      "loss": 0.0095,
+      "step": 280
+    },
+    {
+      "epoch": 0.17931674138197556,
+      "grad_norm": 1.0690827369689941,
+      "learning_rate": 1.7839506172839506e-05,
+      "loss": 0.0282,
+      "step": 290
+    },
+    {
+      "epoch": 0.18550007729169887,
+      "grad_norm": 0.09711379557847977,
+      "learning_rate": 1.8456790123456792e-05,
+      "loss": 0.016,
+      "step": 300
+    },
+    {
+      "epoch": 0.19168341320142215,
+      "grad_norm": 0.05902743339538574,
+      "learning_rate": 1.9074074074074075e-05,
+      "loss": 0.0118,
+      "step": 310
+    },
+    {
+      "epoch": 0.19786674911114546,
+      "grad_norm": 0.058570075780153275,
+      "learning_rate": 1.969135802469136e-05,
+      "loss": 0.0105,
+      "step": 320
+    },
+    {
+      "epoch": 0.20405008502086877,
+      "grad_norm": 0.05820102244615555,
+      "learning_rate": 1.9999854512405252e-05,
+      "loss": 0.0098,
+      "step": 330
+    },
+    {
+      "epoch": 0.21023342093059205,
+      "grad_norm": 0.05199410021305084,
+      "learning_rate": 1.9998690637047088e-05,
+      "loss": 0.0092,
+      "step": 340
+    },
+    {
+      "epoch": 0.21641675684031536,
+      "grad_norm": 0.0505528561770916,
+      "learning_rate": 1.999636302179332e-05,
+      "loss": 0.0093,
+      "step": 350
+    },
+    {
+      "epoch": 0.22260009275003864,
+      "grad_norm": 0.052951931953430176,
+      "learning_rate": 1.9992871937553292e-05,
+      "loss": 0.0089,
+      "step": 360
+    },
+    {
+      "epoch": 0.22878342865976195,
+      "grad_norm": 0.053663648664951324,
+      "learning_rate": 1.9988217790651605e-05,
+      "loss": 0.0086,
+      "step": 370
+    },
+    {
+      "epoch": 0.23496676456948523,
+      "grad_norm": 0.052866872400045395,
+      "learning_rate": 1.9982401122780834e-05,
+      "loss": 0.0093,
+      "step": 380
+    },
+    {
+      "epoch": 0.24115010047920854,
+      "grad_norm": 0.044779375195503235,
+      "learning_rate": 1.9975422610938463e-05,
+      "loss": 0.0087,
+      "step": 390
+    },
+    {
+      "epoch": 0.24733343638893182,
+      "grad_norm": 0.05273430049419403,
+      "learning_rate": 1.9967283067348106e-05,
+      "loss": 0.0087,
+      "step": 400
+    },
+    {
+      "epoch": 0.2535167722986551,
+      "grad_norm": 0.05107806622982025,
+      "learning_rate": 1.9957983439364973e-05,
+      "loss": 0.0084,
+      "step": 410
+    },
+    {
+      "epoch": 0.2597001082083784,
+      "grad_norm": 0.05193234607577324,
+      "learning_rate": 1.994752480936559e-05,
+      "loss": 0.0085,
+      "step": 420
+    },
+    {
+      "epoch": 0.2658834441181017,
+      "grad_norm": 0.04062167927622795,
+      "learning_rate": 1.9935908394621844e-05,
+      "loss": 0.0085,
+      "step": 430
+    },
+    {
+      "epoch": 0.27206678002782503,
+      "grad_norm": 0.04734641686081886,
+      "learning_rate": 1.992313554715929e-05,
+      "loss": 0.0087,
+      "step": 440
+    },
+    {
+      "epoch": 0.2782501159375483,
+      "grad_norm": 0.0560256652534008,
+      "learning_rate": 1.99092077535998e-05,
+      "loss": 0.0086,
+      "step": 450
+    },
+    {
+      "epoch": 0.2844334518472716,
+      "grad_norm": 0.05063237249851227,
+      "learning_rate": 1.989412663498853e-05,
+      "loss": 0.0084,
+      "step": 460
+    },
+    {
+      "epoch": 0.2906167877569949,
+      "grad_norm": 0.05622229725122452,
+      "learning_rate": 1.9877893946605247e-05,
+      "loss": 0.0085,
+      "step": 470
+    },
+    {
+      "epoch": 0.2968001236667182,
+      "grad_norm": 0.06787122786045074,
+      "learning_rate": 1.9860511577760042e-05,
+      "loss": 0.0084,
+      "step": 480
+    },
+    {
+      "epoch": 0.3029834595764415,
+      "grad_norm": 0.05340688303112984,
+      "learning_rate": 1.9841981551573424e-05,
+      "loss": 0.0084,
+      "step": 490
+    },
+    {
+      "epoch": 0.30916679548616477,
+      "grad_norm": 0.05143291503190994,
+      "learning_rate": 1.9822306024740855e-05,
+      "loss": 0.0085,
+      "step": 500
+    },
+    {
+      "epoch": 0.3153501313958881,
+      "grad_norm": 0.04688792675733566,
+      "learning_rate": 1.980148728728174e-05,
+      "loss": 0.0085,
+      "step": 510
+    },
+    {
+      "epoch": 0.3215334673056114,
+      "grad_norm": 0.04486978054046631,
+      "learning_rate": 1.9779527762272877e-05,
+      "loss": 0.0082,
+      "step": 520
+    },
+    {
+      "epoch": 0.3277168032153347,
+      "grad_norm": 0.04129200056195259,
+      "learning_rate": 1.975643000556646e-05,
+      "loss": 0.0083,
+      "step": 530
+    },
+    {
+      "epoch": 0.33390013912505795,
+      "grad_norm": 0.04800448194146156,
+      "learning_rate": 1.973219670549259e-05,
+      "loss": 0.0083,
+      "step": 540
+    },
+    {
+      "epoch": 0.34008347503478126,
+      "grad_norm": 0.05428948998451233,
+      "learning_rate": 1.970683068254637e-05,
+      "loss": 0.0081,
+      "step": 550
+    },
+    {
+      "epoch": 0.34626681094450457,
+      "grad_norm": 0.062362976372241974,
+      "learning_rate": 1.9680334889059668e-05,
+      "loss": 0.0081,
+      "step": 560
+    },
+    {
+      "epoch": 0.3524501468542279,
+      "grad_norm": 0.04794848710298538,
+      "learning_rate": 1.9652712408857452e-05,
+      "loss": 0.0081,
+      "step": 570
+    },
+    {
+      "epoch": 0.35863348276395113,
+      "grad_norm": 0.04279520735144615,
+      "learning_rate": 1.9623966456898914e-05,
+      "loss": 0.0078,
+      "step": 580
+    },
+    {
+      "epoch": 0.36481681867367444,
+      "grad_norm": 0.04028687626123428,
+      "learning_rate": 1.959410037890323e-05,
+      "loss": 0.0083,
+      "step": 590
+    },
+    {
+      "epoch": 0.37100015458339775,
+      "grad_norm": 0.04868888854980469,
+      "learning_rate": 1.9563117650960208e-05,
+      "loss": 0.0085,
+      "step": 600
+    },
+    {
+      "epoch": 0.37718349049312105,
+      "grad_norm": 0.048066020011901855,
+      "learning_rate": 1.9531021879125654e-05,
+      "loss": 0.0082,
+      "step": 610
+    },
+    {
+      "epoch": 0.3833668264028443,
+      "grad_norm": 0.051193494349718094,
+      "learning_rate": 1.9497816799001726e-05,
+      "loss": 0.0081,
+      "step": 620
+    },
+    {
+      "epoch": 0.3895501623125676,
+      "grad_norm": 0.046053316444158554,
+      "learning_rate": 1.9463506275302094e-05,
+      "loss": 0.0078,
+      "step": 630
+    },
+    {
+      "epoch": 0.3957334982222909,
+      "grad_norm": 0.04008670151233673,
+      "learning_rate": 1.9428094301402164e-05,
+      "loss": 0.008,
+      "step": 640
+    },
+    {
+      "epoch": 0.40191683413201423,
+      "grad_norm": 0.048819512128829956,
+      "learning_rate": 1.939158499887428e-05,
+      "loss": 0.0084,
+      "step": 650
+    },
+    {
+      "epoch": 0.40810017004173754,
+      "grad_norm": 0.04148195683956146,
+      "learning_rate": 1.9353982617008024e-05,
+      "loss": 0.0077,
+      "step": 660
+    },
+    {
+      "epoch": 0.4142835059514608,
+      "grad_norm": 0.04567483067512512,
+      "learning_rate": 1.9315291532315626e-05,
+      "loss": 0.0082,
+      "step": 670
+    },
+    {
+      "epoch": 0.4204668418611841,
+      "grad_norm": 0.0510326586663723,
+      "learning_rate": 1.9275516248022613e-05,
+      "loss": 0.0082,
+      "step": 680
+    },
+    {
+      "epoch": 0.4266501777709074,
+      "grad_norm": 0.053588561713695526,
+      "learning_rate": 1.9234661393543668e-05,
+      "loss": 0.0078,
+      "step": 690
+    },
+    {
+      "epoch": 0.4328335136806307,
+      "grad_norm": 0.0437391996383667,
+      "learning_rate": 1.91927317239438e-05,
+      "loss": 0.0082,
+      "step": 700
+    },
+    {
+      "epoch": 0.439016849590354,
+      "grad_norm": 0.05658455193042755,
+      "learning_rate": 1.9149732119384942e-05,
+      "loss": 0.0083,
+      "step": 710
+    },
+    {
+      "epoch": 0.4452001855000773,
+      "grad_norm": 0.039588961750268936,
+      "learning_rate": 1.9105667584557914e-05,
+      "loss": 0.0077,
+      "step": 720
+    },
+    {
+      "epoch": 0.4513835214098006,
+      "grad_norm": 0.05383581295609474,
+      "learning_rate": 1.9060543248099958e-05,
+      "loss": 0.0081,
+      "step": 730
+    },
+    {
+      "epoch": 0.4575668573195239,
+      "grad_norm": 0.047476038336753845,
+      "learning_rate": 1.9014364361997813e-05,
+      "loss": 0.0082,
+      "step": 740
+    },
+    {
+      "epoch": 0.46375019322924715,
+      "grad_norm": 0.057258907705545425,
+      "learning_rate": 1.896713630097643e-05,
+      "loss": 0.0081,
+      "step": 750
+    },
+    {
+      "epoch": 0.46993352913897046,
+      "grad_norm": 0.046649232506752014,
+      "learning_rate": 1.8918864561873428e-05,
+      "loss": 0.0078,
+      "step": 760
+    },
+    {
+      "epoch": 0.47611686504869377,
+      "grad_norm": 0.04877932369709015,
+      "learning_rate": 1.8869554762999307e-05,
+      "loss": 0.0079,
+      "step": 770
+    },
+    {
+      "epoch": 0.4823002009584171,
+      "grad_norm": 0.048988137394189835,
+      "learning_rate": 1.881921264348355e-05,
+      "loss": 0.008,
+      "step": 780
+    },
+    {
+      "epoch": 0.4884835368681404,
+      "grad_norm": 0.04841630905866623,
+      "learning_rate": 1.876784406260664e-05,
+      "loss": 0.0078,
+      "step": 790
+    },
+    {
+      "epoch": 0.49466687277786364,
+      "grad_norm": 0.05180073156952858,
+      "learning_rate": 1.871545499911811e-05,
+      "loss": 0.0079,
+      "step": 800
+    },
+    {
+      "epoch": 0.500850208687587,
+      "grad_norm": 0.04108748957514763,
+      "learning_rate": 1.866205155054067e-05,
+      "loss": 0.0079,
+      "step": 810
+    },
+    {
+      "epoch": 0.5070335445973102,
+      "grad_norm": 0.03657471388578415,
+      "learning_rate": 1.860763993246053e-05,
+      "loss": 0.0077,
+      "step": 820
+    },
+    {
+      "epoch": 0.5132168805070335,
+      "grad_norm": 0.04646613076329231,
+      "learning_rate": 1.8552226477803984e-05,
+      "loss": 0.0076,
+      "step": 830
+    },
+    {
+      "epoch": 0.5194002164167568,
+      "grad_norm": 0.050987377762794495,
+      "learning_rate": 1.8495817636100293e-05,
+      "loss": 0.0079,
+      "step": 840
+    },
+    {
+      "epoch": 0.5255835523264801,
+      "grad_norm": 0.0483928844332695,
+      "learning_rate": 1.8438419972731066e-05,
+      "loss": 0.0076,
+      "step": 850
+    },
+    {
+      "epoch": 0.5317668882362034,
+      "grad_norm": 0.05614335834980011,
+      "learning_rate": 1.8380040168166096e-05,
+      "loss": 0.0079,
+      "step": 860
+    },
+    {
+      "epoch": 0.5379502241459267,
+      "grad_norm": 0.045122936367988586,
+      "learning_rate": 1.8320685017185832e-05,
+      "loss": 0.0077,
+      "step": 870
+    },
+    {
+      "epoch": 0.5441335600556501,
+      "grad_norm": 0.03678731620311737,
+      "learning_rate": 1.8260361428090542e-05,
+      "loss": 0.0077,
+      "step": 880
+    },
+    {
+      "epoch": 0.5503168959653734,
+      "grad_norm": 0.04048831760883331,
+      "learning_rate": 1.8199076421896256e-05,
+      "loss": 0.0075,
+      "step": 890
+    },
+    {
+      "epoch": 0.5565002318750966,
+      "grad_norm": 0.04655023291707039,
+      "learning_rate": 1.813683713151761e-05,
+      "loss": 0.0076,
+      "step": 900
+    },
+    {
+      "epoch": 0.5626835677848199,
+      "grad_norm": 0.04041978716850281,
+      "learning_rate": 1.8073650800937627e-05,
+      "loss": 0.0076,
+      "step": 910
+    },
+    {
+      "epoch": 0.5688669036945432,
+      "grad_norm": 0.051544759422540665,
+      "learning_rate": 1.8009524784364615e-05,
+      "loss": 0.0076,
+      "step": 920
+    },
+    {
+      "epoch": 0.5750502396042665,
+      "grad_norm": 0.04109196737408638,
+      "learning_rate": 1.7944466545376215e-05,
+      "loss": 0.0079,
+      "step": 930
+    },
+    {
+      "epoch": 0.5812335755139898,
+      "grad_norm": 0.04459492862224579,
+      "learning_rate": 1.7878483656050713e-05,
+      "loss": 0.0075,
+      "step": 940
+    },
+    {
+      "epoch": 0.5874169114237131,
+      "grad_norm": 0.051292601972818375,
+      "learning_rate": 1.7811583796085722e-05,
+      "loss": 0.0077,
+      "step": 950
+    },
+    {
+      "epoch": 0.5936002473334364,
+      "grad_norm": 0.04666479676961899,
+      "learning_rate": 1.7743774751904385e-05,
+      "loss": 0.0077,
+      "step": 960
+    },
+    {
+      "epoch": 0.5997835832431597,
+      "grad_norm": 0.043166983872652054,
+      "learning_rate": 1.7675064415749078e-05,
+      "loss": 0.0074,
+      "step": 970
+    },
+    {
+      "epoch": 0.605966919152883,
+      "grad_norm": 0.0443737618625164,
+      "learning_rate": 1.760546078476286e-05,
+      "loss": 0.0077,
+      "step": 980
+    },
+    {
+      "epoch": 0.6121502550626062,
+      "grad_norm": 0.04099385812878609,
+      "learning_rate": 1.7534971960058684e-05,
+      "loss": 0.0076,
+      "step": 990
+    },
+    {
+      "epoch": 0.6183335909723295,
+      "grad_norm": 0.04541454464197159,
+      "learning_rate": 1.7463606145776525e-05,
+      "loss": 0.0076,
+      "step": 1000
+    },
+    {
+      "epoch": 0.6245169268820528,
+      "grad_norm": 0.0451682023704052,
+      "learning_rate": 1.739137164812849e-05,
+      "loss": 0.0075,
+      "step": 1010
+    },
+    {
+      "epoch": 0.6307002627917762,
+      "grad_norm": 0.04592340439558029,
+      "learning_rate": 1.731827687443208e-05,
+      "loss": 0.0077,
+      "step": 1020
+    },
+    {
+      "epoch": 0.6368835987014995,
+      "grad_norm": 0.04665357992053032,
+      "learning_rate": 1.7244330332131657e-05,
+      "loss": 0.0072,
+      "step": 1030
+    },
+    {
+      "epoch": 0.6430669346112228,
+      "grad_norm": 0.046974148601293564,
+      "learning_rate": 1.7169540627808276e-05,
+      "loss": 0.0074,
+      "step": 1040
+    },
+    {
+      "epoch": 0.6492502705209461,
+      "grad_norm": 0.04424068331718445,
+      "learning_rate": 1.7093916466177972e-05,
+      "loss": 0.0073,
+      "step": 1050
+    },
+    {
+      "epoch": 0.6554336064306694,
+      "grad_norm": 0.04412393271923065,
+      "learning_rate": 1.701746664907862e-05,
+      "loss": 0.0072,
+      "step": 1060
+    },
+    {
+      "epoch": 0.6616169423403926,
+      "grad_norm": 0.046212710440158844,
+      "learning_rate": 1.6940200074445503e-05,
+      "loss": 0.0073,
+      "step": 1070
+    },
+    {
+      "epoch": 0.6678002782501159,
+      "grad_norm": 0.05339352414011955,
+      "learning_rate": 1.6862125735275675e-05,
+      "loss": 0.0077,
+      "step": 1080
+    },
+    {
+      "epoch": 0.6739836141598392,
+      "grad_norm": 0.053103432059288025,
+      "learning_rate": 1.6783252718581296e-05,
+      "loss": 0.0078,
+      "step": 1090
+    },
+    {
+      "epoch": 0.6801669500695625,
+      "grad_norm": 0.04800717905163765,
+      "learning_rate": 1.6703590204331967e-05,
+      "loss": 0.0073,
+      "step": 1100
+    },
+    {
+      "epoch": 0.6863502859792858,
+      "grad_norm": 0.04264698177576065,
+      "learning_rate": 1.6623147464386324e-05,
+      "loss": 0.0075,
+      "step": 1110
+    },
+    {
+      "epoch": 0.6925336218890091,
+      "grad_norm": 0.040609877556562424,
+      "learning_rate": 1.6541933861412848e-05,
+      "loss": 0.0073,
+      "step": 1120
+    },
+    {
+      "epoch": 0.6987169577987324,
+      "grad_norm": 0.03998512774705887,
+      "learning_rate": 1.645995884780019e-05,
+      "loss": 0.0076,
+      "step": 1130
+    },
+    {
+      "epoch": 0.7049002937084557,
+      "grad_norm": 0.03671044111251831,
+      "learning_rate": 1.637723196455699e-05,
+      "loss": 0.0073,
+      "step": 1140
+    },
+    {
+      "epoch": 0.7110836296181791,
+      "grad_norm": 0.041929442435503006,
+      "learning_rate": 1.629376284020143e-05,
+      "loss": 0.0075,
+      "step": 1150
+    },
+    {
+      "epoch": 0.7172669655279023,
+      "grad_norm": 0.042013250291347504,
+      "learning_rate": 1.620956118964053e-05,
+      "loss": 0.0074,
+      "step": 1160
+    },
+    {
+      "epoch": 0.7234503014376256,
+      "grad_norm": 0.04549422115087509,
+      "learning_rate": 1.6124636813039502e-05,
+      "loss": 0.0073,
+      "step": 1170
+    },
+    {
+      "epoch": 0.7296336373473489,
+      "grad_norm": 0.04200827330350876,
+      "learning_rate": 1.6038999594681074e-05,
+      "loss": 0.0072,
+      "step": 1180
+    },
+    {
+      "epoch": 0.7358169732570722,
+      "grad_norm": 0.03912513703107834,
+      "learning_rate": 1.5952659501815073e-05,
+      "loss": 0.0073,
+      "step": 1190
+    },
+    {
+      "epoch": 0.7420003091667955,
+      "grad_norm": 0.03626884147524834,
+      "learning_rate": 1.5865626583498355e-05,
+      "loss": 0.0073,
+      "step": 1200
+    },
+    {
+      "epoch": 0.7481836450765188,
+      "grad_norm": 0.04432886838912964,
+      "learning_rate": 1.5777910969425184e-05,
+      "loss": 0.0071,
+      "step": 1210
+    },
+    {
+      "epoch": 0.7543669809862421,
+      "grad_norm": 0.03677462041378021,
+      "learning_rate": 1.5689522868748276e-05,
+      "loss": 0.0074,
+      "step": 1220
+    },
+    {
+      "epoch": 0.7605503168959654,
+      "grad_norm": 0.03566281870007515,
+      "learning_rate": 1.5600472568890523e-05,
+      "loss": 0.0071,
+      "step": 1230
+    },
+    {
+      "epoch": 0.7667336528056886,
+      "grad_norm": 0.037937846034765244,
+      "learning_rate": 1.551077043434769e-05,
+      "loss": 0.0072,
+      "step": 1240
+    },
+    {
+      "epoch": 0.7729169887154119,
+      "grad_norm": 0.034936949610710144,
+      "learning_rate": 1.5420426905482056e-05,
+      "loss": 0.0071,
+      "step": 1250
+    },
+    {
+      "epoch": 0.7791003246251352,
+      "grad_norm": 0.0420398972928524,
+      "learning_rate": 1.5329452497307316e-05,
+      "loss": 0.0072,
+      "step": 1260
+    },
+    {
+      "epoch": 0.7852836605348585,
+      "grad_norm": 0.04124725982546806,
+      "learning_rate": 1.5237857798264701e-05,
+      "loss": 0.0072,
+      "step": 1270
+    },
+    {
+      "epoch": 0.7914669964445818,
+      "grad_norm": 0.042076483368873596,
+      "learning_rate": 1.5145653468990626e-05,
+      "loss": 0.0071,
+      "step": 1280
+    },
+    {
+      "epoch": 0.7976503323543052,
+      "grad_norm": 0.03725574165582657,
+      "learning_rate": 1.50528502410759e-05,
+      "loss": 0.0069,
+      "step": 1290
+    },
+    {
+      "epoch": 0.8038336682640285,
+      "grad_norm": 0.04050353169441223,
+      "learning_rate": 1.4959458915816681e-05,
+      "loss": 0.0072,
+      "step": 1300
+    },
+    {
+      "epoch": 0.8100170041737518,
+      "grad_norm": 0.03882165625691414,
+      "learning_rate": 1.4865490362957317e-05,
+      "loss": 0.0073,
+      "step": 1310
+    },
+    {
+      "epoch": 0.8162003400834751,
+      "grad_norm": 0.03756791725754738,
+      "learning_rate": 1.477095551942525e-05,
+      "loss": 0.0074,
+      "step": 1320
+    },
+    {
+      "epoch": 0.8223836759931983,
+      "grad_norm": 0.05022745579481125,
+      "learning_rate": 1.4675865388058037e-05,
+      "loss": 0.007,
+      "step": 1330
+    },
+    {
+      "epoch": 0.8285670119029216,
+      "grad_norm": 0.04937414452433586,
+      "learning_rate": 1.458023103632277e-05,
+      "loss": 0.0071,
+      "step": 1340
+    },
+    {
+      "epoch": 0.8347503478126449,
+      "grad_norm": 0.051321037113666534,
+      "learning_rate": 1.4484063595027924e-05,
+      "loss": 0.0071,
+      "step": 1350
+    },
+    {
+      "epoch": 0.8409336837223682,
+      "grad_norm": 0.035162657499313354,
+      "learning_rate": 1.438737425702786e-05,
+      "loss": 0.0071,
+      "step": 1360
+    },
+    {
+      "epoch": 0.8471170196320915,
+      "grad_norm": 0.03311579301953316,
+      "learning_rate": 1.4290174275920095e-05,
+      "loss": 0.0072,
+      "step": 1370
+    },
+    {
+      "epoch": 0.8533003555418148,
+      "grad_norm": 0.04103536903858185,
+      "learning_rate": 1.41924749647355e-05,
+      "loss": 0.0071,
+      "step": 1380
+    },
+    {
+      "epoch": 0.8594836914515381,
+      "grad_norm": 0.051637377589941025,
+      "learning_rate": 1.4094287694621589e-05,
+      "loss": 0.0069,
+      "step": 1390
+    },
+    {
+      "epoch": 0.8656670273612614,
+      "grad_norm": 0.033984847366809845,
+      "learning_rate": 1.3995623893519044e-05,
+      "loss": 0.007,
+      "step": 1400
+    },
+    {
+      "epoch": 0.8718503632709848,
+      "grad_norm": 0.03714025020599365,
+      "learning_rate": 1.3896495044831622e-05,
+      "loss": 0.0066,
+      "step": 1410
+    },
+    {
+      "epoch": 0.878033699180708,
+      "grad_norm": 0.041570328176021576,
+      "learning_rate": 1.37969126860896e-05,
+      "loss": 0.0067,
+      "step": 1420
+    },
+    {
+      "epoch": 0.8842170350904313,
+      "grad_norm": 0.04341095685958862,
+      "learning_rate": 1.3696888407606952e-05,
+      "loss": 0.0069,
+      "step": 1430
+    },
+    {
+      "epoch": 0.8904003710001546,
+      "grad_norm": 0.039629433304071426,
+      "learning_rate": 1.3596433851132342e-05,
+      "loss": 0.0069,
+      "step": 1440
+    },
+    {
+      "epoch": 0.8965837069098779,
+      "grad_norm": 0.04006121680140495,
+      "learning_rate": 1.3495560708494167e-05,
+      "loss": 0.007,
+      "step": 1450
+    },
+    {
+      "epoch": 0.9027670428196012,
+      "grad_norm": 0.04321839660406113,
+      "learning_rate": 1.3394280720239733e-05,
+      "loss": 0.007,
+      "step": 1460
+    },
+    {
+      "epoch": 0.9089503787293245,
+      "grad_norm": 0.038396500051021576,
+      "learning_rate": 1.3292605674268803e-05,
+      "loss": 0.0069,
+      "step": 1470
+    },
+    {
+      "epoch": 0.9151337146390478,
+      "grad_norm": 0.040967196226119995,
+      "learning_rate": 1.31905474044616e-05,
+      "loss": 0.007,
+      "step": 1480
+    },
+    {
+      "epoch": 0.9213170505487711,
+      "grad_norm": 0.048842981457710266,
+      "learning_rate": 1.3088117789301473e-05,
+      "loss": 0.0067,
+      "step": 1490
+    },
+    {
+      "epoch": 0.9275003864584943,
+      "grad_norm": 0.03724457696080208,
+      "learning_rate": 1.298532875049237e-05,
+      "loss": 0.0069,
+      "step": 1500
+    },
+    {
+      "epoch": 0.9336837223682176,
+      "grad_norm": 0.039507653564214706,
+      "learning_rate": 1.2882192251571282e-05,
+      "loss": 0.0068,
+      "step": 1510
+    },
+    {
+      "epoch": 0.9398670582779409,
+      "grad_norm": 0.035300228744745255,
+      "learning_rate": 1.2778720296515816e-05,
+      "loss": 0.0068,
+      "step": 1520
+    },
+    {
+      "epoch": 0.9460503941876642,
+      "grad_norm": 0.05400003492832184,
+      "learning_rate": 1.2674924928347055e-05,
+      "loss": 0.0067,
+      "step": 1530
+    },
+    {
+      "epoch": 0.9522337300973875,
+      "grad_norm": 0.04334673658013344,
+      "learning_rate": 1.2570818227727889e-05,
+      "loss": 0.0068,
+      "step": 1540
+    },
+    {
+      "epoch": 0.9584170660071109,
+      "grad_norm": 0.03568002209067345,
+      "learning_rate": 1.2466412311556952e-05,
+      "loss": 0.0068,
+      "step": 1550
+    },
+    {
+      "epoch": 0.9646004019168342,
+      "grad_norm": 0.047103166580200195,
+      "learning_rate": 1.2361719331558346e-05,
+      "loss": 0.0068,
+      "step": 1560
+    },
+    {
+      "epoch": 0.9707837378265575,
+      "grad_norm": 0.036910735070705414,
+      "learning_rate": 1.2256751472867304e-05,
+      "loss": 0.0067,
+      "step": 1570
+    },
+    {
+      "epoch": 0.9769670737362808,
+      "grad_norm": 0.04214445874094963,
+      "learning_rate": 1.2151520952611984e-05,
+      "loss": 0.0067,
+      "step": 1580
+    },
+    {
+      "epoch": 0.983150409646004,
+      "grad_norm": 0.05339273810386658,
+      "learning_rate": 1.2046040018491523e-05,
+      "loss": 0.0066,
+      "step": 1590
+    },
+    {
+      "epoch": 0.9893337455557273,
+      "grad_norm": 0.04332943633198738,
+      "learning_rate": 1.1940320947350532e-05,
+      "loss": 0.007,
+      "step": 1600
+    },
+    {
+      "epoch": 0.9955170814654506,
+      "grad_norm": 0.0429212786257267,
+      "learning_rate": 1.1834376043750201e-05,
+      "loss": 0.0063,
+      "step": 1610
+    },
+    {
+      "epoch": 1.0012366671819446,
+      "grad_norm": 0.04421928524971008,
+      "learning_rate": 1.1728217638536196e-05,
+      "loss": 0.0071,
+      "step": 1620
+    },
+    {
+      "epoch": 1.007420003091668,
+      "grad_norm": 0.04365362226963043,
+      "learning_rate": 1.1621858087403466e-05,
+      "loss": 0.0062,
+      "step": 1630
+    },
+    {
+      "epoch": 1.0136033390013912,
+      "grad_norm": 0.03700665757060051,
+      "learning_rate": 1.1515309769458174e-05,
+      "loss": 0.0059,
+      "step": 1640
+    },
+    {
+      "epoch": 1.0197866749111146,
+      "grad_norm": 0.0455857552587986,
+      "learning_rate": 1.1408585085776916e-05,
+      "loss": 0.0059,
+      "step": 1650
+    },
+    {
+      "epoch": 1.0259700108208378,
+      "grad_norm": 0.03983158990740776,
+      "learning_rate": 1.1301696457963352e-05,
+      "loss": 0.0062,
+      "step": 1660
+    },
+    {
+      "epoch": 1.0321533467305612,
+      "grad_norm": 0.03655843436717987,
+      "learning_rate": 1.1194656326702474e-05,
+      "loss": 0.0062,
+      "step": 1670
+    },
+    {
+      "epoch": 1.0383366826402844,
+      "grad_norm": 0.048476699739694595,
+      "learning_rate": 1.1087477150312653e-05,
+      "loss": 0.006,
+      "step": 1680
+    },
+    {
+      "epoch": 1.0445200185500076,
+      "grad_norm": 0.04713127389550209,
+      "learning_rate": 1.098017140329561e-05,
+      "loss": 0.0059,
+      "step": 1690
+    },
+    {
+      "epoch": 1.050703354459731,
+      "grad_norm": 0.03435790538787842,
+      "learning_rate": 1.0872751574884512e-05,
+      "loss": 0.006,
+      "step": 1700
+    },
+    {
+      "epoch": 1.0568866903694543,
+      "grad_norm": 0.04033854603767395,
+      "learning_rate": 1.0765230167590385e-05,
+      "loss": 0.0056,
+      "step": 1710
+    },
+    {
+      "epoch": 1.0630700262791777,
+      "grad_norm": 0.03484448045492172,
+      "learning_rate": 1.0657619695746944e-05,
+      "loss": 0.0061,
+      "step": 1720
+    },
+    {
+      "epoch": 1.0692533621889009,
+      "grad_norm": 0.041448500007390976,
+      "learning_rate": 1.054993268405405e-05,
+      "loss": 0.006,
+      "step": 1730
+    },
+    {
+      "epoch": 1.0754366980986243,
+      "grad_norm": 0.045591264963150024,
+      "learning_rate": 1.0442181666119984e-05,
+      "loss": 0.006,
+      "step": 1740
+    },
+    {
+      "epoch": 1.0816200340083475,
+      "grad_norm": 0.036942582577466965,
+      "learning_rate": 1.033437918300266e-05,
+      "loss": 0.006,
+      "step": 1750
+    },
+    {
+      "epoch": 1.087803369918071,
+      "grad_norm": 0.03570098802447319,
+      "learning_rate": 1.0226537781749988e-05,
+      "loss": 0.0058,
+      "step": 1760
+    },
+    {
+      "epoch": 1.0939867058277941,
+      "grad_norm": 0.0382462739944458,
+      "learning_rate": 1.0118670013939526e-05,
+      "loss": 0.0058,
+      "step": 1770
+    },
+    {
+      "epoch": 1.1001700417375173,
+      "grad_norm": 0.03386927768588066,
+      "learning_rate": 1.0010788434217625e-05,
+      "loss": 0.0059,
+      "step": 1780
+    },
+    {
+      "epoch": 1.1063533776472407,
+      "grad_norm": 0.04100915789604187,
+      "learning_rate": 9.902905598838177e-06,
+      "loss": 0.0059,
+      "step": 1790
+    },
+    {
+      "epoch": 1.112536713556964,
+      "grad_norm": 0.03817896544933319,
+      "learning_rate": 9.795034064201235e-06,
+      "loss": 0.0058,
+      "step": 1800
+    },
+    {
+      "epoch": 1.1187200494666873,
+      "grad_norm": 0.04328783601522446,
+      "learning_rate": 9.68718638539156e-06,
+      "loss": 0.0059,
+      "step": 1810
+    },
+    {
+      "epoch": 1.1249033853764105,
+      "grad_norm": 0.038214631378650665,
+      "learning_rate": 9.579375114717352e-06,
+      "loss": 0.0058,
+      "step": 1820
+    },
+    {
+      "epoch": 1.1310867212861337,
+      "grad_norm": 0.03657764196395874,
+      "learning_rate": 9.471612800249295e-06,
+      "loss": 0.0059,
+      "step": 1830
+    },
+    {
+      "epoch": 1.1372700571958572,
+      "grad_norm": 0.045281436294317245,
+      "learning_rate": 9.363911984360104e-06,
+      "loss": 0.0059,
+      "step": 1840
+    },
+    {
+      "epoch": 1.1434533931055806,
+      "grad_norm": 0.03649020567536354,
+      "learning_rate": 9.256285202264716e-06,
+      "loss": 0.0061,
+      "step": 1850
+    },
+    {
+      "epoch": 1.1496367290153038,
+      "grad_norm": 0.03974698483943939,
+      "learning_rate": 9.148744980561332e-06,
+      "loss": 0.0056,
+      "step": 1860
+    },
+    {
+      "epoch": 1.155820064925027,
+      "grad_norm": 0.041133228689432144,
+      "learning_rate": 9.041303835773458e-06,
+      "loss": 0.006,
+      "step": 1870
+    },
+    {
+      "epoch": 1.1620034008347504,
+      "grad_norm": 0.039168015122413635,
+      "learning_rate": 8.933974272893113e-06,
+      "loss": 0.0058,
+      "step": 1880
+    },
+    {
+      "epoch": 1.1681867367444736,
+      "grad_norm": 0.06035231426358223,
+      "learning_rate": 8.826768783925385e-06,
+      "loss": 0.0058,
+      "step": 1890
+    },
+    {
+      "epoch": 1.174370072654197,
+      "grad_norm": 0.04310515522956848,
+      "learning_rate": 8.719699846434493e-06,
+      "loss": 0.0057,
+      "step": 1900
+    },
+    {
+      "epoch": 1.1805534085639202,
+      "grad_norm": 0.03753894194960594,
+      "learning_rate": 8.612779922091534e-06,
+      "loss": 0.0055,
+      "step": 1910
+    },
+    {
+      "epoch": 1.1867367444736434,
+      "grad_norm": 0.0496654249727726,
+      "learning_rate": 8.506021455224083e-06,
+      "loss": 0.0056,
+      "step": 1920
+    },
+    {
+      "epoch": 1.1929200803833668,
+      "grad_norm": 0.03786466270685196,
+      "learning_rate": 8.399436871367804e-06,
+      "loss": 0.0059,
+      "step": 1930
+    },
+    {
+      "epoch": 1.19910341629309,
+      "grad_norm": 0.034229736775159836,
+      "learning_rate": 8.293038575820249e-06,
+      "loss": 0.0058,
+      "step": 1940
+    },
+    {
+      "epoch": 1.2052867522028134,
+      "grad_norm": 0.03900885581970215,
+      "learning_rate": 8.186838952197019e-06,
+      "loss": 0.0057,
+      "step": 1950
+    },
+    {
+      "epoch": 1.2114700881125366,
+      "grad_norm": 0.036625415086746216,
+      "learning_rate": 8.080850360990439e-06,
+      "loss": 0.0056,
+      "step": 1960
+    },
+    {
+      "epoch": 1.21765342402226,
+      "grad_norm": 0.032340049743652344,
+      "learning_rate": 7.975085138130938e-06,
+      "loss": 0.0056,
+      "step": 1970
+    },
+    {
+      "epoch": 1.2238367599319833,
+      "grad_norm": 0.03607684373855591,
+      "learning_rate": 7.869555593551277e-06,
+      "loss": 0.0059,
+      "step": 1980
+    },
+    {
+      "epoch": 1.2300200958417067,
+      "grad_norm": 0.032486408948898315,
+      "learning_rate": 7.764274009753796e-06,
+      "loss": 0.0059,
+      "step": 1990
+    },
+    {
+      "epoch": 1.2362034317514299,
+      "grad_norm": 0.03523161634802818,
+      "learning_rate": 7.659252640380876e-06,
+      "loss": 0.0056,
+      "step": 2000
+    },
+    {
+      "epoch": 1.242386767661153,
+      "grad_norm": 0.036453727632761,
+      "learning_rate": 7.554503708788738e-06,
+      "loss": 0.0055,
+      "step": 2010
+    },
+    {
+      "epoch": 1.2485701035708765,
+      "grad_norm": 0.040288910269737244,
+      "learning_rate": 7.450039406624792e-06,
+      "loss": 0.0057,
+      "step": 2020
+    },
+    {
+      "epoch": 1.2547534394806,
+      "grad_norm": 0.0430009663105011,
+      "learning_rate": 7.345871892408631e-06,
+      "loss": 0.0056,
+      "step": 2030
+    },
+    {
+      "epoch": 1.2609367753903231,
+      "grad_norm": 0.03574775531888008,
+      "learning_rate": 7.242013290116944e-06,
+      "loss": 0.0057,
+      "step": 2040
+    },
+    {
+      "epoch": 1.2671201113000463,
+      "grad_norm": 0.03812326118350029,
+      "learning_rate": 7.138475687772381e-06,
+      "loss": 0.0056,
+      "step": 2050
+    },
+    {
+      "epoch": 1.2733034472097697,
+      "grad_norm": 0.03299403935670853,
+      "learning_rate": 7.03527113603667e-06,
+      "loss": 0.0058,
+      "step": 2060
+    },
+    {
+      "epoch": 1.279486783119493,
+      "grad_norm": 0.03786725923418999,
+      "learning_rate": 6.9324116468080305e-06,
+      "loss": 0.0056,
+      "step": 2070
+    },
+    {
+      "epoch": 1.2856701190292164,
+      "grad_norm": 0.03449017554521561,
+      "learning_rate": 6.829909191823121e-06,
+      "loss": 0.0054,
+      "step": 2080
+    },
+    {
+      "epoch": 1.2918534549389395,
+      "grad_norm": 0.03949199244379997,
+      "learning_rate": 6.727775701263666e-06,
+      "loss": 0.0058,
+      "step": 2090
+    },
+    {
+      "epoch": 1.2980367908486627,
+      "grad_norm": 0.03537849709391594,
+      "learning_rate": 6.626023062367894e-06,
+      "loss": 0.006,
+      "step": 2100
+    },
+    {
+      "epoch": 1.3042201267583862,
+      "grad_norm": 0.03561354801058769,
+      "learning_rate": 6.524663118047015e-06,
+      "loss": 0.0058,
+      "step": 2110
+    },
+    {
+      "epoch": 1.3104034626681094,
+      "grad_norm": 0.037109751254320145,
+      "learning_rate": 6.423707665506812e-06,
+      "loss": 0.0056,
+      "step": 2120
+    },
+    {
+      "epoch": 1.3165867985778328,
+      "grad_norm": 0.03227913752198219,
+      "learning_rate": 6.323168454874582e-06,
+      "loss": 0.0054,
+      "step": 2130
+    },
+    {
+      "epoch": 1.322770134487556,
+      "grad_norm": 0.034376900643110275,
+      "learning_rate": 6.223057187831548e-06,
+      "loss": 0.0057,
+      "step": 2140
+    },
+    {
+      "epoch": 1.3289534703972794,
+      "grad_norm": 0.042623914778232574,
+      "learning_rate": 6.123385516250911e-06,
+      "loss": 0.0054,
+      "step": 2150
+    },
+    {
+      "epoch": 1.3351368063070026,
+      "grad_norm": 0.03284977748990059,
+      "learning_rate": 6.02416504084169e-06,
+      "loss": 0.0056,
+      "step": 2160
+    },
+    {
+      "epoch": 1.341320142216726,
+      "grad_norm": 0.03773827850818634,
+      "learning_rate": 5.925407309798535e-06,
+      "loss": 0.0061,
+      "step": 2170
+    },
+    {
+      "epoch": 1.3475034781264492,
+      "grad_norm": 0.03684335947036743,
+      "learning_rate": 5.827123817457631e-06,
+      "loss": 0.0053,
+      "step": 2180
+    },
+    {
+      "epoch": 1.3536868140361724,
+      "grad_norm": 0.03831595927476883,
+      "learning_rate": 5.729326002958889e-06,
+      "loss": 0.0057,
+      "step": 2190
+    },
+    {
+      "epoch": 1.3598701499458958,
+      "grad_norm": 0.03576178103685379,
+      "learning_rate": 5.632025248914546e-06,
+      "loss": 0.0054,
+      "step": 2200
+    },
+    {
+      "epoch": 1.366053485855619,
+      "grad_norm": 0.04004263877868652,
+      "learning_rate": 5.535232880084373e-06,
+      "loss": 0.0057,
+      "step": 2210
+    },
+    {
+      "epoch": 1.3722368217653425,
+      "grad_norm": 0.03518558666110039,
+      "learning_rate": 5.438960162057548e-06,
+      "loss": 0.0054,
+      "step": 2220
+    },
+    {
+      "epoch": 1.3784201576750656,
+      "grad_norm": 0.03575367107987404,
+      "learning_rate": 5.343218299941517e-06,
+      "loss": 0.0055,
+      "step": 2230
+    },
+    {
+      "epoch": 1.384603493584789,
+      "grad_norm": 0.03193347156047821,
+      "learning_rate": 5.248018437057807e-06,
+      "loss": 0.0055,
+      "step": 2240
+    },
+    {
+      "epoch": 1.3907868294945123,
+      "grad_norm": 0.03565206378698349,
+      "learning_rate": 5.15337165364507e-06,
+      "loss": 0.0054,
+      "step": 2250
+    },
+    {
+      "epoch": 1.3969701654042357,
+      "grad_norm": 0.0318601168692112,
+      "learning_rate": 5.059288965569471e-06,
+      "loss": 0.0054,
+      "step": 2260
+    },
+    {
+      "epoch": 1.4031535013139589,
+      "grad_norm": 0.03948667645454407,
+      "learning_rate": 4.965781323042547e-06,
+      "loss": 0.0054,
+      "step": 2270
+    },
+    {
+      "epoch": 1.409336837223682,
+      "grad_norm": 0.03156143054366112,
+      "learning_rate": 4.872859609346733e-06,
+      "loss": 0.0053,
+      "step": 2280
+    },
+    {
+      "epoch": 1.4155201731334055,
+      "grad_norm": 0.03467686474323273,
+      "learning_rate": 4.780534639568651e-06,
+      "loss": 0.0053,
+      "step": 2290
+    },
+    {
+      "epoch": 1.4217035090431287,
+      "grad_norm": 0.03605823591351509,
+      "learning_rate": 4.68881715934037e-06,
+      "loss": 0.0055,
+      "step": 2300
+    },
+    {
+      "epoch": 1.4278868449528521,
+      "grad_norm": 0.03542206436395645,
+      "learning_rate": 4.597717843588708e-06,
+      "loss": 0.0054,
+      "step": 2310
+    },
+    {
+      "epoch": 1.4340701808625753,
+      "grad_norm": 0.03765241429209709,
+      "learning_rate": 4.5072472952928015e-06,
+      "loss": 0.0055,
+      "step": 2320
+    },
+    {
+      "epoch": 1.4402535167722987,
+      "grad_norm": 0.03652080148458481,
+      "learning_rate": 4.417416044250026e-06,
+      "loss": 0.0057,
+      "step": 2330
+    },
+    {
+      "epoch": 1.446436852682022,
+      "grad_norm": 0.03262539580464363,
+      "learning_rate": 4.328234545850441e-06,
+      "loss": 0.0052,
+      "step": 2340
+    },
+    {
+      "epoch": 1.4526201885917454,
+      "grad_norm": 0.03427337110042572,
+      "learning_rate": 4.239713179859901e-06,
+      "loss": 0.0053,
+      "step": 2350
+    },
+    {
+      "epoch": 1.4588035245014686,
+      "grad_norm": 0.03885317221283913,
+      "learning_rate": 4.151862249211953e-06,
+      "loss": 0.0054,
+      "step": 2360
+    },
+    {
+      "epoch": 1.4649868604111917,
+      "grad_norm": 0.03729204460978508,
+      "learning_rate": 4.064691978808699e-06,
+      "loss": 0.0053,
+      "step": 2370
+    },
+    {
+      "epoch": 1.4711701963209152,
+      "grad_norm": 0.032668232917785645,
+      "learning_rate": 3.978212514330703e-06,
+      "loss": 0.0053,
+      "step": 2380
+    },
+    {
+      "epoch": 1.4773535322306384,
+      "grad_norm": 0.03400380164384842,
+      "learning_rate": 3.892433921056184e-06,
+      "loss": 0.0053,
+      "step": 2390
+    },
+    {
+      "epoch": 1.4835368681403618,
+      "grad_norm": 0.03154066205024719,
+      "learning_rate": 3.8073661826894837e-06,
+      "loss": 0.0052,
+      "step": 2400
+    },
+    {
+      "epoch": 1.489720204050085,
+      "grad_norm": 0.0407693088054657,
+      "learning_rate": 3.7230192001990963e-06,
+      "loss": 0.0047,
+      "step": 2410
+    },
+    {
+      "epoch": 1.4959035399598082,
+      "grad_norm": 0.03409425541758537,
+      "learning_rate": 3.6394027906652927e-06,
+      "loss": 0.0048,
+      "step": 2420
+    },
+    {
+      "epoch": 1.5020868758695316,
+      "grad_norm": 0.03952354937791824,
+      "learning_rate": 3.5565266861375236e-06,
+      "loss": 0.0052,
+      "step": 2430
+    },
+    {
+      "epoch": 1.508270211779255,
+      "grad_norm": 0.035029612481594086,
+      "learning_rate": 3.4744005325017073e-06,
+      "loss": 0.0046,
+      "step": 2440
+    },
+    {
+      "epoch": 1.5144535476889782,
+      "grad_norm": 0.04004015028476715,
+      "learning_rate": 3.3930338883575564e-06,
+      "loss": 0.0048,
+      "step": 2450
+    },
+    {
+      "epoch": 1.5206368835987014,
+      "grad_norm": 0.04242485761642456,
+      "learning_rate": 3.3124362239060627e-06,
+      "loss": 0.0051,
+      "step": 2460
+    },
+    {
+      "epoch": 1.5268202195084248,
+      "grad_norm": 0.035334572196006775,
+      "learning_rate": 3.2326169198472555e-06,
+      "loss": 0.005,
+      "step": 2470
+    },
+    {
+      "epoch": 1.5330035554181483,
+      "grad_norm": 0.03957268223166466,
+      "learning_rate": 3.1535852662884138e-06,
+      "loss": 0.0048,
+      "step": 2480
+    },
+    {
+      "epoch": 1.5391868913278715,
+      "grad_norm": 0.03457095846533775,
+      "learning_rate": 3.0753504616627694e-06,
+      "loss": 0.0048,
+      "step": 2490
+    },
+    {
+      "epoch": 1.5453702272375947,
+      "grad_norm": 0.03342267498373985,
+      "learning_rate": 2.9979216116589203e-06,
+      "loss": 0.0048,
+      "step": 2500
+    },
+    {
+      "epoch": 1.5515535631473178,
+      "grad_norm": 0.0400426909327507,
+      "learning_rate": 2.9213077281610336e-06,
+      "loss": 0.0046,
+      "step": 2510
+    },
+    {
+      "epoch": 1.5577368990570413,
+      "grad_norm": 0.034529346972703934,
+      "learning_rate": 2.8455177281999515e-06,
+      "loss": 0.0048,
+      "step": 2520
+    },
+    {
+      "epoch": 1.5639202349667647,
+      "grad_norm": 0.042129479348659515,
+      "learning_rate": 2.7705604329153434e-06,
+      "loss": 0.0048,
+      "step": 2530
+    },
+    {
+      "epoch": 1.5701035708764879,
+      "grad_norm": 0.03391699492931366,
+      "learning_rate": 2.6964445665290274e-06,
+      "loss": 0.0048,
+      "step": 2540
+    },
+    {
+      "epoch": 1.576286906786211,
+      "grad_norm": 0.03239929676055908,
+      "learning_rate": 2.6231787553295583e-06,
+      "loss": 0.005,
+      "step": 2550
+    },
+    {
+      "epoch": 1.5824702426959345,
+      "grad_norm": 0.03297235071659088,
+      "learning_rate": 2.550771526668221e-06,
+      "loss": 0.0047,
+      "step": 2560
+    },
+    {
+      "epoch": 1.5886535786056577,
+      "grad_norm": 0.03652636706829071,
+      "learning_rate": 2.47923130796655e-06,
+      "loss": 0.0048,
+      "step": 2570
+    },
+    {
+      "epoch": 1.5948369145153811,
+      "grad_norm": 0.03381171077489853,
+      "learning_rate": 2.408566425735446e-06,
+      "loss": 0.0047,
+      "step": 2580
+    },
+    {
+      "epoch": 1.6010202504251043,
+      "grad_norm": 0.03702075779438019,
+      "learning_rate": 2.338785104606082e-06,
+      "loss": 0.0046,
+      "step": 2590
+    },
+    {
+      "epoch": 1.6072035863348275,
+      "grad_norm": 0.03340402618050575,
+      "learning_rate": 2.26989546637263e-06,
+      "loss": 0.0047,
+      "step": 2600
+    },
+    {
+      "epoch": 1.613386922244551,
+      "grad_norm": 0.030743056908249855,
+      "learning_rate": 2.2019055290469816e-06,
+      "loss": 0.0048,
+      "step": 2610
+    },
+    {
+      "epoch": 1.6195702581542744,
+      "grad_norm": 0.03447384014725685,
+      "learning_rate": 2.1348232059255293e-06,
+      "loss": 0.0049,
+      "step": 2620
+    },
+    {
+      "epoch": 1.6257535940639976,
+      "grad_norm": 0.03731811046600342,
+      "learning_rate": 2.0686563046681507e-06,
+      "loss": 0.0047,
+      "step": 2630
+    },
+    {
+      "epoch": 1.6319369299737208,
+      "grad_norm": 0.03979882597923279,
+      "learning_rate": 2.0034125263894777e-06,
+      "loss": 0.0047,
+      "step": 2640
+    },
+    {
+      "epoch": 1.6381202658834442,
+      "grad_norm": 0.030915752053260803,
+      "learning_rate": 1.9390994647625693e-06,
+      "loss": 0.0043,
+      "step": 2650
+    },
+    {
+      "epoch": 1.6443036017931674,
+      "grad_norm": 0.03616530820727348,
+      "learning_rate": 1.8757246051350908e-06,
+      "loss": 0.0047,
+      "step": 2660
+    },
+    {
+      "epoch": 1.6504869377028908,
+      "grad_norm": 0.03190992400050163,
+      "learning_rate": 1.813295323658103e-06,
+      "loss": 0.0045,
+      "step": 2670
+    },
+    {
+      "epoch": 1.656670273612614,
+      "grad_norm": 0.032364994287490845,
+      "learning_rate": 1.7518188864275498e-06,
+      "loss": 0.0045,
+      "step": 2680
+    },
+    {
+      "epoch": 1.6628536095223372,
+      "grad_norm": 0.03670560568571091,
+      "learning_rate": 1.6913024486385733e-06,
+      "loss": 0.0047,
+      "step": 2690
+    },
+    {
+      "epoch": 1.6690369454320606,
+      "grad_norm": 0.0347522608935833,
+      "learning_rate": 1.6317530537527148e-06,
+      "loss": 0.0047,
+      "step": 2700
+    },
+    {
+      "epoch": 1.675220281341784,
+      "grad_norm": 0.03402375429868698,
+      "learning_rate": 1.5731776326781435e-06,
+      "loss": 0.0047,
+      "step": 2710
+    },
+    {
+      "epoch": 1.6814036172515072,
+      "grad_norm": 0.034099772572517395,
+      "learning_rate": 1.5155830029629614e-06,
+      "loss": 0.0045,
+      "step": 2720
+    },
+    {
+      "epoch": 1.6875869531612304,
+      "grad_norm": 0.035209812223911285,
+      "learning_rate": 1.4589758680017263e-06,
+      "loss": 0.0047,
+      "step": 2730
+    },
+    {
+      "epoch": 1.6937702890709536,
+      "grad_norm": 0.03288722783327103,
+      "learning_rate": 1.403362816255236e-06,
+      "loss": 0.0044,
+      "step": 2740
+    },
+    {
+      "epoch": 1.699953624980677,
+      "grad_norm": 0.031503114849328995,
+      "learning_rate": 1.3487503204837182e-06,
+      "loss": 0.005,
+      "step": 2750
+    },
+    {
+      "epoch": 1.7061369608904005,
+      "grad_norm": 0.03957793861627579,
+      "learning_rate": 1.2951447369934544e-06,
+      "loss": 0.0046,
+      "step": 2760
+    },
+    {
+      "epoch": 1.7123202968001237,
+      "grad_norm": 0.03468756750226021,
+      "learning_rate": 1.2425523048969835e-06,
+      "loss": 0.0048,
+      "step": 2770
+    },
+    {
+      "epoch": 1.7185036327098469,
+      "grad_norm": 0.03237258270382881,
+      "learning_rate": 1.190979145386939e-06,
+      "loss": 0.0046,
+      "step": 2780
+    },
+    {
+      "epoch": 1.7246869686195703,
+      "grad_norm": 0.032560866326093674,
+      "learning_rate": 1.1404312610235968e-06,
+      "loss": 0.0045,
+      "step": 2790
+    },
+    {
+      "epoch": 1.7308703045292937,
+      "grad_norm": 0.03548423945903778,
+      "learning_rate": 1.0909145350362516e-06,
+      "loss": 0.0046,
+      "step": 2800
+    },
+    {
+      "epoch": 1.7370536404390169,
+      "grad_norm": 0.04064382612705231,
+      "learning_rate": 1.042434730638473e-06,
+      "loss": 0.0046,
+      "step": 2810
+    },
+    {
+      "epoch": 1.74323697634874,
+      "grad_norm": 0.03297872841358185,
+      "learning_rate": 9.949974903573223e-07,
+      "loss": 0.0044,
+      "step": 2820
+    },
+    {
+      "epoch": 1.7494203122584633,
+      "grad_norm": 0.03142615780234337,
+      "learning_rate": 9.486083353766262e-07,
+      "loss": 0.0045,
+      "step": 2830
+    },
+    {
+      "epoch": 1.7556036481681867,
+      "grad_norm": 0.032335978001356125,
+      "learning_rate": 9.032726648943779e-07,
+      "loss": 0.0047,
+      "step": 2840
+    },
+    {
+      "epoch": 1.7617869840779101,
+      "grad_norm": 0.032847121357917786,
+      "learning_rate": 8.58995755494314e-07,
+      "loss": 0.0043,
+      "step": 2850
+    },
+    {
+      "epoch": 1.7679703199876333,
+      "grad_norm": 0.03723368048667908,
+      "learning_rate": 8.157827605317892e-07,
+      "loss": 0.0045,
+      "step": 2860
+    },
+    {
+      "epoch": 1.7741536558973565,
+      "grad_norm": 0.03200025483965874,
+      "learning_rate": 7.736387095339748e-07,
+      "loss": 0.0046,
+      "step": 2870
+    },
+    {
+      "epoch": 1.78033699180708,
+      "grad_norm": 0.03175206854939461,
+      "learning_rate": 7.325685076144795e-07,
+      "loss": 0.0044,
+      "step": 2880
+    },
+    {
+      "epoch": 1.7865203277168034,
+      "grad_norm": 0.03391426429152489,
+      "learning_rate": 6.925769349024459e-07,
+      "loss": 0.0046,
+      "step": 2890
+    },
+    {
+      "epoch": 1.7927036636265266,
+      "grad_norm": 0.033382490277290344,
+      "learning_rate": 6.536686459861918e-07,
+      "loss": 0.0044,
+      "step": 2900
+    },
+    {
+      "epoch": 1.7988869995362498,
+      "grad_norm": 0.03503987565636635,
+      "learning_rate": 6.158481693714735e-07,
+      "loss": 0.0045,
+      "step": 2910
+    },
+    {
+      "epoch": 1.805070335445973,
+      "grad_norm": 0.03369370475411415,
+      "learning_rate": 5.791199069544074e-07,
+      "loss": 0.0043,
+      "step": 2920
+    },
+    {
+      "epoch": 1.8112536713556964,
+      "grad_norm": 0.03411993384361267,
+      "learning_rate": 5.434881335091458e-07,
+      "loss": 0.0047,
+      "step": 2930
+    },
+    {
+      "epoch": 1.8174370072654198,
+      "grad_norm": 0.031462300568819046,
+      "learning_rate": 5.089569961903318e-07,
+      "loss": 0.0048,
+      "step": 2940
+    },
+    {
+      "epoch": 1.823620343175143,
+      "grad_norm": 0.03133223205804825,
+      "learning_rate": 4.755305140504185e-07,
+      "loss": 0.0044,
+      "step": 2950
+    },
+    {
+      "epoch": 1.8298036790848662,
+      "grad_norm": 0.03286576643586159,
+      "learning_rate": 4.432125775718965e-07,
+      "loss": 0.0044,
+      "step": 2960
+    },
+    {
+      "epoch": 1.8359870149945896,
+      "grad_norm": 0.0382905974984169,
+      "learning_rate": 4.120069482144773e-07,
+      "loss": 0.0045,
+      "step": 2970
+    },
+    {
+      "epoch": 1.842170350904313,
+      "grad_norm": 0.0318513922393322,
+      "learning_rate": 3.8191725797730897e-07,
+      "loss": 0.0045,
+      "step": 2980
+    },
+    {
+      "epoch": 1.8483536868140362,
+      "grad_norm": 0.03260161355137825,
+      "learning_rate": 3.529470089762421e-07,
+      "loss": 0.0045,
+      "step": 2990
+    },
+    {
+      "epoch": 1.8545370227237594,
+      "grad_norm": 0.03626921400427818,
+      "learning_rate": 3.2509957303622496e-07,
+      "loss": 0.0045,
+      "step": 3000
+    },
+    {
+      "epoch": 1.8607203586334826,
+      "grad_norm": 0.033349767327308655,
+      "learning_rate": 2.983781912988604e-07,
+      "loss": 0.0044,
+      "step": 3010
+    },
+    {
+      "epoch": 1.866903694543206,
+      "grad_norm": 0.03295992314815521,
+      "learning_rate": 2.727859738451721e-07,
+      "loss": 0.0044,
+      "step": 3020
+    },
+    {
+      "epoch": 1.8730870304529295,
+      "grad_norm": 0.03412463888525963,
+      "learning_rate": 2.4832589933361907e-07,
+      "loss": 0.0042,
+      "step": 3030
+    },
+    {
+      "epoch": 1.8792703663626527,
+      "grad_norm": 0.03095577284693718,
+      "learning_rate": 2.2500081465341993e-07,
+      "loss": 0.0043,
+      "step": 3040
+    },
+    {
+      "epoch": 1.8854537022723759,
+      "grad_norm": 0.03507233411073685,
+      "learning_rate": 2.0281343459319868e-07,
+      "loss": 0.0044,
+      "step": 3050
+    },
+    {
+      "epoch": 1.8916370381820993,
+      "grad_norm": 0.03278861194849014,
+      "learning_rate": 1.8176634152501816e-07,
+      "loss": 0.0043,
+      "step": 3060
+    },
+    {
+      "epoch": 1.8978203740918227,
+      "grad_norm": 0.03288375213742256,
+      "learning_rate": 1.618619851038161e-07,
+      "loss": 0.0045,
+      "step": 3070
+    },
+    {
+      "epoch": 1.904003710001546,
+      "grad_norm": 0.03299954906105995,
+      "learning_rate": 1.4310268198229115e-07,
+      "loss": 0.0044,
+      "step": 3080
+    },
+    {
+      "epoch": 1.910187045911269,
+      "grad_norm": 0.03464515134692192,
+      "learning_rate": 1.2549061554127494e-07,
+      "loss": 0.0043,
+      "step": 3090
+    },
+    {
+      "epoch": 1.9163703818209923,
+      "grad_norm": 0.03539302200078964,
+      "learning_rate": 1.0902783563560226e-07,
+      "loss": 0.0046,
+      "step": 3100
+    },
+    {
+      "epoch": 1.9225537177307157,
+      "grad_norm": 0.03203083202242851,
+      "learning_rate": 9.371625835553622e-08,
+      "loss": 0.0045,
+      "step": 3110
+    },
+    {
+      "epoch": 1.9287370536404391,
+      "grad_norm": 0.03126191347837448,
+      "learning_rate": 7.955766580375334e-08,
+      "loss": 0.0042,
+      "step": 3120
+    },
+    {
+      "epoch": 1.9349203895501623,
+      "grad_norm": 0.031476061791181564,
+      "learning_rate": 6.655370588792732e-08,
+      "loss": 0.0045,
+      "step": 3130
+    },
+    {
+      "epoch": 1.9411037254598855,
+      "grad_norm": 0.03566085919737816,
+      "learning_rate": 5.470589212892785e-08,
+      "loss": 0.0044,
+      "step": 3140
+    },
+    {
+      "epoch": 1.947287061369609,
+      "grad_norm": 0.030942443758249283,
+      "learning_rate": 4.401560348466505e-08,
+      "loss": 0.0043,
+      "step": 3150
+    },
+    {
+      "epoch": 1.9534703972793324,
+      "grad_norm": 0.03156135976314545,
+      "learning_rate": 3.448408418959326e-08,
+      "loss": 0.0043,
+      "step": 3160
+    },
+    {
+      "epoch": 1.9596537331890556,
+      "grad_norm": 0.03293183073401451,
+      "learning_rate": 2.6112443609895845e-08,
+      "loss": 0.0044,
+      "step": 3170
+    },
+    {
+      "epoch": 1.9658370690987788,
+      "grad_norm": 0.03247111290693283,
+      "learning_rate": 1.890165611436734e-08,
+      "loss": 0.0045,
+      "step": 3180
+    },
+    {
+      "epoch": 1.972020405008502,
+      "grad_norm": 0.03142822906374931,
+      "learning_rate": 1.285256096100529e-08,
+      "loss": 0.0044,
+      "step": 3190
+    },
+    {
+      "epoch": 1.9782037409182254,
+      "grad_norm": 0.034565262496471405,
+      "learning_rate": 7.965862199330598e-09,
+      "loss": 0.0045,
+      "step": 3200
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 3236,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 100,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 7.045427946497612e+19,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-3200/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cc8b833b867ea533744d2414dc3b6b6a0d0eef687754e0cce57e2883b75a64ca
+size 7953

checkpoint-3200/zero_to_fp32.py

@@ -0,0 +1,760 @@
+#!/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 . output_dir/
+#   or
+#   python zero_to_fp32.py . output_dir/ --safe_serialization
+
+import argparse
+import torch
+import glob
+import math
+import os
+import re
+import gc
+import json
+import numpy as np
+from tqdm import tqdm
+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, weights_only=False)
+
+        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 tqdm(files, desc='Loading checkpoint shards'):
+        state_dict = torch.load(f, map_location=device, mmap=True, weights_only=False)
+        # 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}")
+
+    fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] 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")
+
+
+class GatheredTensor:
+    """
+    A pseudo tensor that collects partitioned weights.
+    It is more memory efficient when there are multiple groups.
+    """
+
+    def __init__(self, flat_groups, flat_groups_offset, offset, partitioned_numel, shape):
+        self.flat_groups = flat_groups
+        self.flat_groups_offset = flat_groups_offset
+        self.offset = offset
+        self.partitioned_numel = partitioned_numel
+        self.shape = shape
+        self.dtype = self.flat_groups[0][0].dtype
+
+    def contiguous(self):
+        """
+        Merge partitioned weights from flat_groups into a single tensor.
+        """
+        end_idx = self.offset + self.partitioned_numel
+        world_size = len(self.flat_groups)
+        pad_flat_param_chunks = []
+
+        for rank_i in range(world_size):
+            # for each rank, we need to collect weights from related group/groups
+            flat_groups_at_rank_i = self.flat_groups[rank_i]
+            start_group_id = None
+            end_group_id = None
+            for group_id in range(len(self.flat_groups_offset)):
+                if self.flat_groups_offset[group_id] <= self.offset < self.flat_groups_offset[group_id + 1]:
+                    start_group_id = group_id
+                if self.flat_groups_offset[group_id] < end_idx <= self.flat_groups_offset[group_id + 1]:
+                    end_group_id = group_id
+                    break
+            # collect weights from related group/groups
+            for group_id in range(start_group_id, end_group_id + 1):
+                flat_tensor = flat_groups_at_rank_i[group_id]
+                start_offset = self.offset - self.flat_groups_offset[group_id]
+                end_offset = min(end_idx, self.flat_groups_offset[group_id + 1]) - self.flat_groups_offset[group_id]
+                pad_flat_param_chunks.append(flat_tensor[start_offset:end_offset])
+
+        # collect weights from all ranks
+        pad_flat_param = torch.cat(pad_flat_param_chunks, dim=0)
+        param = pad_flat_param[:self.shape.numel()].view(self.shape).contiguous()
+        return param
+
+
+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 = sum([flat_group.numel() for flat_group in fp32_flat_groups[0]]) * 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
+    flat_groups_offset = [0] + list(np.cumsum([flat_tensor.numel() for flat_tensor in fp32_flat_groups[0]]))
+    for name, shape in tqdm(param_shapes.items(), desc='Gathering sharded weights'):
+        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}"
+            )
+
+        # memory efficient tensor
+        tensor = GatheredTensor(fp32_flat_groups, flat_groups_offset, offset, partitioned_numel, shape)
+        state_dict[name] = tensor
+        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 to_torch_tensor(state_dict, return_empty_tensor=False):
+    """
+    Convert state_dict of GatheredTensor to torch tensor
+    """
+    torch_state_dict = {}
+    converted_tensors = {}
+    for name, tensor in state_dict.items():
+        tensor_id = id(tensor)
+        if tensor_id in converted_tensors:  # shared tensors
+            shared_tensor = torch_state_dict[converted_tensors[tensor_id]]
+            torch_state_dict[name] = shared_tensor
+        else:
+            converted_tensors[tensor_id] = name
+            if return_empty_tensor:
+                torch_state_dict[name] = torch.empty(tensor.shape, dtype=tensor.dtype)
+            else:
+                torch_state_dict[name] = tensor.contiguous()
+    return torch_state_dict
+
+
+def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir,
+                                             tag=None,
+                                             exclude_frozen_parameters=False,
+                                             lazy_mode=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
+        - ``lazy_mode``: get state_dict in lazy mode. It returns a dict of pesduo tensor instead of torch tensor, which is more memory efficient.
+          Convert the pesduo tensor to torch tensor by ``.contiguous()``
+
+    Returns:
+        - pytorch ``state_dict``
+
+    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.
+
+    Note: the above usage may not work if your application doesn't have sufficient free CPU memory.
+    You may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
+    the checkpoint. Or you can load state_dict in lazy mode ::
+
+        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, lazy_mode=True) # not on cpu
+        for name, lazy_tensor in state_dict.item():
+            tensor = lazy_tensor.contiguous()  # to cpu
+            print(name, tensor)
+            # del tensor to release memory if it no longer in use
+    """
+    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")
+
+    state_dict = _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
+    if lazy_mode:
+        return state_dict
+    else:
+        return to_torch_tensor(state_dict)
+
+
+def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir,
+                                               output_dir,
+                                               max_shard_size="5GB",
+                                               safe_serialization=False,
+                                               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_dir``: directory to the pytorch fp32 state_dict output files
+        - ``max_shard_size``: the maximum size for a checkpoint before being sharded, default value is 5GB
+        - ``safe_serialization``:  whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).
+        - ``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
+    """
+
+    # Dependency pre-check
+    if safe_serialization:
+        try:
+            from safetensors.torch import save_file
+        except ImportError:
+            print('If you want to use `safe_serialization`, please `pip install safetensors`')
+            raise
+    if max_shard_size is not None:
+        try:
+            from huggingface_hub import split_torch_state_dict_into_shards
+        except ImportError:
+            print('If you want to use `max_shard_size`, please `pip install huggingface_hub`')
+            raise
+
+    # Convert zero checkpoint to state_dict
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir,
+                                                          tag,
+                                                          exclude_frozen_parameters,
+                                                          lazy_mode=True)
+
+    # Shard the model if it is too big.
+    weights_name = "model.safetensors" if safe_serialization else "pytorch_model.bin"
+    if max_shard_size is not None:
+        filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors")
+        # an memory-efficient approach for sharding
+        empty_state_dict = to_torch_tensor(state_dict, return_empty_tensor=True)
+        state_dict_split = split_torch_state_dict_into_shards(empty_state_dict,
+                                                              filename_pattern=filename_pattern,
+                                                              max_shard_size=max_shard_size)
+    else:
+        from collections import namedtuple
+        StateDictSplit = namedtuple("StateDictSplit", ["is_sharded", "filename_to_tensors"])
+        state_dict_split = StateDictSplit(is_sharded=False,
+                                          filename_to_tensors={weights_name: list(state_dict.keys())})
+
+    # Save the model by shard
+    os.makedirs(output_dir, exist_ok=True)
+    filename_to_tensors = state_dict_split.filename_to_tensors.items()
+    for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"):
+        shard_state_dict = {tensor_name: state_dict[tensor_name] for tensor_name in tensors}
+        shard_state_dict = to_torch_tensor(shard_state_dict)
+        output_path = os.path.join(output_dir, shard_file)
+        if safe_serialization:
+            save_file(shard_state_dict, output_path, metadata={"format": "pt"})
+        else:
+            torch.save(shard_state_dict, output_path)
+        # release the memory of current shard
+        for tensor_name in list(shard_state_dict.keys()):
+            del state_dict[tensor_name]
+            del shard_state_dict[tensor_name]
+        del shard_state_dict
+        gc.collect()
+
+    # Save index if sharded
+    if state_dict_split.is_sharded:
+        index = {
+            "metadata": state_dict_split.metadata,
+            "weight_map": state_dict_split.tensor_to_filename,
+        }
+        save_index_file = "model.safetensors.index.json" if safe_serialization else "pytorch_model.bin.index.json"
+        save_index_file = os.path.join(output_dir, save_index_file)
+        with open(save_index_file, "w", encoding="utf-8") as f:
+            content = json.dumps(index, indent=2, sort_keys=True) + "\n"
+            f.write(content)
+
+
+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_dir",
+                        type=str,
+                        help="directory to the pytorch fp32 state_dict output files"
+                        "(e.g. path/checkpoint-12-output/)")
+    parser.add_argument(
+        "--max_shard_size",
+        type=str,
+        default="5GB",
+        help="The maximum size for a checkpoint before being sharded. Checkpoints shard will then be each of size"
+        "lower than this size. If expressed as a string, needs to be digits followed by a unit (like `5MB`"
+        "We default it to 5GB in order for models to be able to run easily on free-tier google colab instances"
+        "without CPU OOM issues.")
+    parser.add_argument(
+        "--safe_serialization",
+        default=False,
+        action='store_true',
+        help="Whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).")
+    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_dir,
+                                               max_shard_size=args.max_shard_size,
+                                               safe_serialization=args.safe_serialization,
+                                               tag=args.tag,
+                                               exclude_frozen_parameters=args.exclude_frozen_parameters)

generation_config.json

@@ -0,0 +1,14 @@
+{
+  "bos_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "repetition_penalty": 1.1,
+  "temperature": 0.7,
+  "top_k": 20,
+  "top_p": 0.8,
+  "transformers_version": "4.55.0"
+}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:68568245a0b436d8cfe998cfc30ef096be55953626ad688db4acd2c1b6604aab
+oid sha256:89a6e59a26bc98ff9cd9e0c0c62ddf5a19ddb7ad689cafa1e7781185299c9272
 size 3554214752

trainer_log.jsonl

@@ -78,3 +78,7 @@
 {"current_steps": 3180, "total_steps": 3236, "loss": 0.0045, "lr": 1.890165611436734e-08, "epoch": 1.9658370690987788, "percentage": 98.27, "elapsed_time": "14:10:39", "remaining_time": "0:14:58"}
 {"current_steps": 3190, "total_steps": 3236, "loss": 0.0044, "lr": 1.285256096100529e-08, "epoch": 1.972020405008502, "percentage": 98.58, "elapsed_time": "14:21:25", "remaining_time": "0:12:25"}
 {"current_steps": 3200, "total_steps": 3236, "loss": 0.0045, "lr": 7.965862199330598e-09, "epoch": 1.9782037409182254, "percentage": 98.89, "elapsed_time": "14:32:10", "remaining_time": "0:09:48"}
+{"current_steps": 3210, "total_steps": 3236, "loss": 0.0043, "lr": 4.242128588446415e-09, "epoch": 1.9843870768279488, "percentage": 99.2, "elapsed_time": "14:44:32", "remaining_time": "0:07:09"}
+{"current_steps": 3220, "total_steps": 3236, "loss": 0.0047, "lr": 1.6817935308366395e-09, "epoch": 1.990570412737672, "percentage": 99.51, "elapsed_time": "14:55:19", "remaining_time": "0:04:26"}
+{"current_steps": 3230, "total_steps": 3236, "loss": 0.0042, "lr": 2.851550219240551e-10, "epoch": 1.9967537486473952, "percentage": 99.81, "elapsed_time": "15:06:05", "remaining_time": "0:01:40"}
+{"current_steps": 3235, "total_steps": 3236, "epoch": 1.9998454166022568, "percentage": 99.97, "elapsed_time": "15:11:45", "remaining_time": "0:00:16"}