upload

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip 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

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
+}

config.json

@@ -0,0 +1,30 @@
+{
+  "_name_or_path": "/lustre/fsn1/projects/rech/gkb/uua32zb/grand_challenge/checkpoints/Arab/Qwen__Qwen2.5-1.5B-annealing_only-Arab-fineweb2-0.0001LR-8192CL-2GAS-2BS-3EPOCHS-0.9BETA1-0.95BETA2-cosineLRSCHEDULER/",
+  "architectures": [
+    "Qwen2ForCausalLM"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151643,
+  "hidden_act": "silu",
+  "hidden_size": 1536,
+  "initializer_range": 0.02,
+  "intermediate_size": 8960,
+  "max_position_embeddings": 131072,
+  "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": "4.46.1",
+  "use_cache": false,
+  "use_mrope": false,
+  "use_sliding_window": false,
+  "vocab_size": 151936
+}

generation_config.json

@@ -0,0 +1,6 @@
+{
+  "bos_token_id": 151643,
+  "eos_token_id": 151643,
+  "max_new_tokens": 2048,
+  "transformers_version": "4.46.1"
+}

latest

@@ -0,0 +1 @@
+global_step8500

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f9666a226a65a6bad25dbcbe0920b45f26415bb57fcedcab371612b2766c433e
+size 3554214752

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
+  }
+}

tokenizer.json

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

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,
+  "chat_template": "{%- if tools %}\n    {{- '<|im_start|>system\\n' }}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- messages[0]['content'] }}\n    {%- else %}\n        {{- 'You are a helpful assistant.' }}\n    {%- endif %}\n    {{- \"\\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>\" }}\n    {%- for tool in tools %}\n        {{- \"\\n\" }}\n        {{- tool | tojson }}\n    {%- endfor %}\n    {{- \"\\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\" }}\n{%- else %}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- '<|im_start|>system\\n' + messages[0]['content'] + '<|im_end|>\\n' }}\n    {%- else %}\n        {{- '<|im_start|>system\\nYou are a helpful assistant.<|im_end|>\\n' }}\n    {%- endif %}\n{%- endif %}\n{%- for message in messages %}\n    {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) or (message.role == \"assistant\" and not message.tool_calls) %}\n        {{- '<|im_start|>' + message.role + '\\n' + message.content + '<|im_end|>' + '\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {{- '<|im_start|>' + message.role }}\n        {%- if message.content %}\n            {{- '\\n' + message.content }}\n        {%- endif %}\n        {%- for tool_call in message.tool_calls %}\n            {%- if tool_call.function is defined %}\n                {%- set tool_call = tool_call.function %}\n            {%- endif %}\n            {{- '\\n<tool_call>\\n{\"name\": \"' }}\n            {{- tool_call.name }}\n            {{- '\", \"arguments\": ' }}\n            {{- tool_call.arguments | tojson }}\n            {{- '}\\n</tool_call>' }}\n        {%- endfor %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"tool\" %}\n        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != \"tool\") %}\n            {{- '<|im_start|>user' }}\n        {%- endif %}\n        {{- '\\n<tool_response>\\n' }}\n        {{- message.content }}\n        {{- '\\n</tool_response>' }}\n        {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n            {{- '<|im_end|>\\n' }}\n        {%- endif %}\n    {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n    {{- '<|im_start|>assistant\\n' }}\n{%- endif %}\n",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "padding_side": "right",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

trainer_state.json

@@ -0,0 +1,1223 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.941124366817062,
+  "eval_steps": 500.0,
+  "global_step": 8500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.005536025687159188,
+      "grad_norm": 0.3439372479915619,
+      "learning_rate": 9.999243694725429e-05,
+      "loss": 0.5743,
+      "step": 50
+    },
+    {
+      "epoch": 0.011072051374318376,
+      "grad_norm": 0.3828931152820587,
+      "learning_rate": 9.99697500770078e-05,
+      "loss": 0.5609,
+      "step": 100
+    },
+    {
+      "epoch": 0.016608077061477565,
+      "grad_norm": 0.3481367528438568,
+      "learning_rate": 9.993194625254041e-05,
+      "loss": 0.5661,
+      "step": 150
+    },
+    {
+      "epoch": 0.022144102748636752,
+      "grad_norm": 0.3216044306755066,
+      "learning_rate": 9.987903691034485e-05,
+      "loss": 0.5729,
+      "step": 200
+    },
+    {
+      "epoch": 0.027680128435795943,
+      "grad_norm": 0.3202624022960663,
+      "learning_rate": 9.981103805666695e-05,
+      "loss": 0.5682,
+      "step": 250
+    },
+    {
+      "epoch": 0.03321615412295513,
+      "grad_norm": 0.32036539912223816,
+      "learning_rate": 9.972797026266339e-05,
+      "loss": 0.5744,
+      "step": 300
+    },
+    {
+      "epoch": 0.03875217981011432,
+      "grad_norm": 0.29840943217277527,
+      "learning_rate": 9.962985865817845e-05,
+      "loss": 0.5737,
+      "step": 350
+    },
+    {
+      "epoch": 0.044288205497273504,
+      "grad_norm": 0.3060721457004547,
+      "learning_rate": 9.951673292414175e-05,
+      "loss": 0.5748,
+      "step": 400
+    },
+    {
+      "epoch": 0.049824231184432695,
+      "grad_norm": 0.29271894693374634,
+      "learning_rate": 9.9388627283589e-05,
+      "loss": 0.5717,
+      "step": 450
+    },
+    {
+      "epoch": 0.055360256871591886,
+      "grad_norm": 0.30329838395118713,
+      "learning_rate": 9.92455804913089e-05,
+      "loss": 0.5751,
+      "step": 500
+    },
+    {
+      "epoch": 0.06089628255875107,
+      "grad_norm": 0.2716081142425537,
+      "learning_rate": 9.90876358221188e-05,
+      "loss": 0.573,
+      "step": 550
+    },
+    {
+      "epoch": 0.06643230824591026,
+      "grad_norm": 0.30109643936157227,
+      "learning_rate": 9.891484105777331e-05,
+      "loss": 0.5753,
+      "step": 600
+    },
+    {
+      "epoch": 0.07196833393306945,
+      "grad_norm": 0.2664714455604553,
+      "learning_rate": 9.872724847250907e-05,
+      "loss": 0.5747,
+      "step": 650
+    },
+    {
+      "epoch": 0.07750435962022864,
+      "grad_norm": 0.29528167843818665,
+      "learning_rate": 9.852491481723078e-05,
+      "loss": 0.5707,
+      "step": 700
+    },
+    {
+      "epoch": 0.08304038530738783,
+      "grad_norm": 0.2820889353752136,
+      "learning_rate": 9.830790130234274e-05,
+      "loss": 0.5738,
+      "step": 750
+    },
+    {
+      "epoch": 0.08857641099454701,
+      "grad_norm": 0.28580889105796814,
+      "learning_rate": 9.80762735792313e-05,
+      "loss": 0.5736,
+      "step": 800
+    },
+    {
+      "epoch": 0.0941124366817062,
+      "grad_norm": 0.2842045724391937,
+      "learning_rate": 9.7830101720404e-05,
+      "loss": 0.5712,
+      "step": 850
+    },
+    {
+      "epoch": 0.09964846236886539,
+      "grad_norm": 0.29261285066604614,
+      "learning_rate": 9.756946019829088e-05,
+      "loss": 0.5699,
+      "step": 900
+    },
+    {
+      "epoch": 0.10518448805602458,
+      "grad_norm": 0.2930528223514557,
+      "learning_rate": 9.729442786271517e-05,
+      "loss": 0.5741,
+      "step": 950
+    },
+    {
+      "epoch": 0.11072051374318377,
+      "grad_norm": 0.2633638083934784,
+      "learning_rate": 9.70050879170393e-05,
+      "loss": 0.5699,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11625653943034296,
+      "grad_norm": 0.262563556432724,
+      "learning_rate": 9.670152789299407e-05,
+      "loss": 0.5727,
+      "step": 1050
+    },
+    {
+      "epoch": 0.12179256511750214,
+      "grad_norm": 0.261879026889801,
+      "learning_rate": 9.638383962419843e-05,
+      "loss": 0.572,
+      "step": 1100
+    },
+    {
+      "epoch": 0.12732859080466133,
+      "grad_norm": 0.41501718759536743,
+      "learning_rate": 9.605211921837773e-05,
+      "loss": 0.5699,
+      "step": 1150
+    },
+    {
+      "epoch": 0.13286461649182052,
+      "grad_norm": 0.2684990465641022,
+      "learning_rate": 9.5706467028289e-05,
+      "loss": 0.5705,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1384006421789797,
+      "grad_norm": 0.2522321045398712,
+      "learning_rate": 9.534698762136204e-05,
+      "loss": 0.5696,
+      "step": 1250
+    },
+    {
+      "epoch": 0.1439366678661389,
+      "grad_norm": 0.3583656847476959,
+      "learning_rate": 9.497378974806551e-05,
+      "loss": 0.5677,
+      "step": 1300
+    },
+    {
+      "epoch": 0.1494726935532981,
+      "grad_norm": 0.24989576637744904,
+      "learning_rate": 9.45869863090074e-05,
+      "loss": 0.5672,
+      "step": 1350
+    },
+    {
+      "epoch": 0.15500871924045728,
+      "grad_norm": 0.25384247303009033,
+      "learning_rate": 9.418669432078016e-05,
+      "loss": 0.5669,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16054474492761647,
+      "grad_norm": 0.2404128462076187,
+      "learning_rate": 9.377303488056066e-05,
+      "loss": 0.5657,
+      "step": 1450
+    },
+    {
+      "epoch": 0.16608077061477566,
+      "grad_norm": 0.2735093832015991,
+      "learning_rate": 9.334613312947549e-05,
+      "loss": 0.5662,
+      "step": 1500
+    },
+    {
+      "epoch": 0.17161679630193485,
+      "grad_norm": 0.25135621428489685,
+      "learning_rate": 9.290611821474307e-05,
+      "loss": 0.5661,
+      "step": 1550
+    },
+    {
+      "epoch": 0.17715282198909402,
+      "grad_norm": 0.2490808069705963,
+      "learning_rate": 9.245312325060375e-05,
+      "loss": 0.5655,
+      "step": 1600
+    },
+    {
+      "epoch": 0.1826888476762532,
+      "grad_norm": 0.254464715719223,
+      "learning_rate": 9.198728527804987e-05,
+      "loss": 0.5652,
+      "step": 1650
+    },
+    {
+      "epoch": 0.1882248733634124,
+      "grad_norm": 0.2614707052707672,
+      "learning_rate": 9.150874522336767e-05,
+      "loss": 0.5647,
+      "step": 1700
+    },
+    {
+      "epoch": 0.1937608990505716,
+      "grad_norm": 0.265312135219574,
+      "learning_rate": 9.101764785550415e-05,
+      "loss": 0.5637,
+      "step": 1750
+    },
+    {
+      "epoch": 0.19929692473773078,
+      "grad_norm": 0.23875299096107483,
+      "learning_rate": 9.051414174227118e-05,
+      "loss": 0.5614,
+      "step": 1800
+    },
+    {
+      "epoch": 0.20483295042488997,
+      "grad_norm": 0.25199180841445923,
+      "learning_rate": 8.999837920540044e-05,
+      "loss": 0.5618,
+      "step": 1850
+    },
+    {
+      "epoch": 0.21036897611204916,
+      "grad_norm": 0.24338677525520325,
+      "learning_rate": 8.947051627446275e-05,
+      "loss": 0.5649,
+      "step": 1900
+    },
+    {
+      "epoch": 0.21590500179920835,
+      "grad_norm": 0.24273960292339325,
+      "learning_rate": 8.89307126396657e-05,
+      "loss": 0.5581,
+      "step": 1950
+    },
+    {
+      "epoch": 0.22144102748636754,
+      "grad_norm": 0.40545448660850525,
+      "learning_rate": 8.837913160354374e-05,
+      "loss": 0.5618,
+      "step": 2000
+    },
+    {
+      "epoch": 0.22697705317352673,
+      "grad_norm": 0.25062185525894165,
+      "learning_rate": 8.781594003155569e-05,
+      "loss": 0.5603,
+      "step": 2050
+    },
+    {
+      "epoch": 0.23251307886068592,
+      "grad_norm": 0.23254728317260742,
+      "learning_rate": 8.724130830160414e-05,
+      "loss": 0.5607,
+      "step": 2100
+    },
+    {
+      "epoch": 0.23804910454784511,
+      "grad_norm": 0.25338929891586304,
+      "learning_rate": 8.66554102524924e-05,
+      "loss": 0.5579,
+      "step": 2150
+    },
+    {
+      "epoch": 0.24358513023500428,
+      "grad_norm": 0.31788700819015503,
+      "learning_rate": 8.605842313133444e-05,
+      "loss": 0.5591,
+      "step": 2200
+    },
+    {
+      "epoch": 0.24912115592216347,
+      "grad_norm": 0.24001486599445343,
+      "learning_rate": 8.545052753993368e-05,
+      "loss": 0.5598,
+      "step": 2250
+    },
+    {
+      "epoch": 0.25465718160932266,
+      "grad_norm": 1.3120613098144531,
+      "learning_rate": 8.483190738014701e-05,
+      "loss": 0.5539,
+      "step": 2300
+    },
+    {
+      "epoch": 0.2601932072964819,
+      "grad_norm": 0.23131923377513885,
+      "learning_rate": 8.420274979825033e-05,
+      "loss": 0.5548,
+      "step": 2350
+    },
+    {
+      "epoch": 0.26572923298364104,
+      "grad_norm": 0.23763211071491241,
+      "learning_rate": 8.356324512832276e-05,
+      "loss": 0.5548,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27126525867080026,
+      "grad_norm": 0.2329683154821396,
+      "learning_rate": 8.291358683466625e-05,
+      "loss": 0.5537,
+      "step": 2450
+    },
+    {
+      "epoch": 0.2768012843579594,
+      "grad_norm": 0.23742693662643433,
+      "learning_rate": 8.225397145327851e-05,
+      "loss": 0.5534,
+      "step": 2500
+    },
+    {
+      "epoch": 0.2823373100451186,
+      "grad_norm": 0.244510680437088,
+      "learning_rate": 8.158459853239634e-05,
+      "loss": 0.5552,
+      "step": 2550
+    },
+    {
+      "epoch": 0.2878733357322778,
+      "grad_norm": 0.23719392716884613,
+      "learning_rate": 8.090567057212808e-05,
+      "loss": 0.5533,
+      "step": 2600
+    },
+    {
+      "epoch": 0.29340936141943696,
+      "grad_norm": 0.22732912003993988,
+      "learning_rate": 8.021739296319266e-05,
+      "loss": 0.551,
+      "step": 2650
+    },
+    {
+      "epoch": 0.2989453871065962,
+      "grad_norm": 0.23341135680675507,
+      "learning_rate": 7.95199739247845e-05,
+      "loss": 0.5498,
+      "step": 2700
+    },
+    {
+      "epoch": 0.30448141279375535,
+      "grad_norm": 0.26524463295936584,
+      "learning_rate": 7.881362444158249e-05,
+      "loss": 0.5481,
+      "step": 2750
+    },
+    {
+      "epoch": 0.31001743848091456,
+      "grad_norm": 1.088100552558899,
+      "learning_rate": 7.809855819992263e-05,
+      "loss": 0.5506,
+      "step": 2800
+    },
+    {
+      "epoch": 0.3155534641680737,
+      "grad_norm": 0.22746309638023376,
+      "learning_rate": 7.737499152315298e-05,
+      "loss": 0.5469,
+      "step": 2850
+    },
+    {
+      "epoch": 0.32108948985523295,
+      "grad_norm": 0.23455417156219482,
+      "learning_rate": 7.664314330619122e-05,
+      "loss": 0.5442,
+      "step": 2900
+    },
+    {
+      "epoch": 0.3266255155423921,
+      "grad_norm": 0.2275807410478592,
+      "learning_rate": 7.590323494930396e-05,
+      "loss": 0.5488,
+      "step": 2950
+    },
+    {
+      "epoch": 0.3321615412295513,
+      "grad_norm": 0.23548050224781036,
+      "learning_rate": 7.515549029112848e-05,
+      "loss": 0.5457,
+      "step": 3000
+    },
+    {
+      "epoch": 0.3376975669167105,
+      "grad_norm": 0.27617892622947693,
+      "learning_rate": 7.440013554095635e-05,
+      "loss": 0.5477,
+      "step": 3050
+    },
+    {
+      "epoch": 0.3432335926038697,
+      "grad_norm": 0.22806765139102936,
+      "learning_rate": 7.363739921030026e-05,
+      "loss": 0.5438,
+      "step": 3100
+    },
+    {
+      "epoch": 0.34876961829102887,
+      "grad_norm": 0.22560247778892517,
+      "learning_rate": 7.286751204376419e-05,
+      "loss": 0.5434,
+      "step": 3150
+    },
+    {
+      "epoch": 0.35430564397818803,
+      "grad_norm": 0.24336422979831696,
+      "learning_rate": 7.209070694923812e-05,
+      "loss": 0.5441,
+      "step": 3200
+    },
+    {
+      "epoch": 0.35984166966534725,
+      "grad_norm": 0.2279517948627472,
+      "learning_rate": 7.130721892743816e-05,
+      "loss": 0.5392,
+      "step": 3250
+    },
+    {
+      "epoch": 0.3653776953525064,
+      "grad_norm": 0.23315465450286865,
+      "learning_rate": 7.05172850008137e-05,
+      "loss": 0.5421,
+      "step": 3300
+    },
+    {
+      "epoch": 0.37091372103966563,
+      "grad_norm": 0.21792641282081604,
+      "learning_rate": 6.972114414184281e-05,
+      "loss": 0.5385,
+      "step": 3350
+    },
+    {
+      "epoch": 0.3764497467268248,
+      "grad_norm": 0.22232070565223694,
+      "learning_rate": 6.891903720073791e-05,
+      "loss": 0.5393,
+      "step": 3400
+    },
+    {
+      "epoch": 0.381985772413984,
+      "grad_norm": 0.22461949288845062,
+      "learning_rate": 6.811120683258312e-05,
+      "loss": 0.539,
+      "step": 3450
+    },
+    {
+      "epoch": 0.3875217981011432,
+      "grad_norm": 0.2534426748752594,
+      "learning_rate": 6.729789742392578e-05,
+      "loss": 0.5388,
+      "step": 3500
+    },
+    {
+      "epoch": 0.3930578237883024,
+      "grad_norm": 0.23156966269016266,
+      "learning_rate": 6.647935501884414e-05,
+      "loss": 0.5374,
+      "step": 3550
+    },
+    {
+      "epoch": 0.39859384947546156,
+      "grad_norm": 0.22473926842212677,
+      "learning_rate": 6.565582724451361e-05,
+      "loss": 0.5352,
+      "step": 3600
+    },
+    {
+      "epoch": 0.4041298751626208,
+      "grad_norm": 0.21904204785823822,
+      "learning_rate": 6.482756323629396e-05,
+      "loss": 0.5323,
+      "step": 3650
+    },
+    {
+      "epoch": 0.40966590084977994,
+      "grad_norm": 0.2203177511692047,
+      "learning_rate": 6.399481356236042e-05,
+      "loss": 0.5328,
+      "step": 3700
+    },
+    {
+      "epoch": 0.41520192653693916,
+      "grad_norm": 0.22808462381362915,
+      "learning_rate": 6.315783014790135e-05,
+      "loss": 0.5369,
+      "step": 3750
+    },
+    {
+      "epoch": 0.4207379522240983,
+      "grad_norm": 0.22881828248500824,
+      "learning_rate": 6.231686619890517e-05,
+      "loss": 0.5331,
+      "step": 3800
+    },
+    {
+      "epoch": 0.4262739779112575,
+      "grad_norm": 0.2256467193365097,
+      "learning_rate": 6.147217612556002e-05,
+      "loss": 0.5289,
+      "step": 3850
+    },
+    {
+      "epoch": 0.4318100035984167,
+      "grad_norm": 0.2211819589138031,
+      "learning_rate": 6.062401546528903e-05,
+      "loss": 0.5312,
+      "step": 3900
+    },
+    {
+      "epoch": 0.43734602928557587,
+      "grad_norm": 0.22896280884742737,
+      "learning_rate": 5.9772640805444626e-05,
+      "loss": 0.5297,
+      "step": 3950
+    },
+    {
+      "epoch": 0.4428820549727351,
+      "grad_norm": 0.2265992909669876,
+      "learning_rate": 5.891830970568517e-05,
+      "loss": 0.5331,
+      "step": 4000
+    },
+    {
+      "epoch": 0.44841808065989425,
+      "grad_norm": 0.22474776208400726,
+      "learning_rate": 5.806128062005744e-05,
+      "loss": 0.528,
+      "step": 4050
+    },
+    {
+      "epoch": 0.45395410634705347,
+      "grad_norm": 0.21792100369930267,
+      "learning_rate": 5.720181281880862e-05,
+      "loss": 0.5329,
+      "step": 4100
+    },
+    {
+      "epoch": 0.45949013203421263,
+      "grad_norm": 0.224033921957016,
+      "learning_rate": 5.634016630995125e-05,
+      "loss": 0.5275,
+      "step": 4150
+    },
+    {
+      "epoch": 0.46502615772137185,
+      "grad_norm": 0.22141297161579132,
+      "learning_rate": 5.547660176060513e-05,
+      "loss": 0.5285,
+      "step": 4200
+    },
+    {
+      "epoch": 0.470562183408531,
+      "grad_norm": 0.22379058599472046,
+      "learning_rate": 5.4611380418139714e-05,
+      "loss": 0.5267,
+      "step": 4250
+    },
+    {
+      "epoch": 0.47609820909569023,
+      "grad_norm": 0.21981391310691833,
+      "learning_rate": 5.374476403114096e-05,
+      "loss": 0.5276,
+      "step": 4300
+    },
+    {
+      "epoch": 0.4816342347828494,
+      "grad_norm": 0.21584829688072205,
+      "learning_rate": 5.287701477022671e-05,
+      "loss": 0.5259,
+      "step": 4350
+    },
+    {
+      "epoch": 0.48717026047000855,
+      "grad_norm": 0.21814610064029694,
+      "learning_rate": 5.2008395148734135e-05,
+      "loss": 0.5229,
+      "step": 4400
+    },
+    {
+      "epoch": 0.49270628615716777,
+      "grad_norm": 0.21903090178966522,
+      "learning_rate": 5.113916794330382e-05,
+      "loss": 0.5225,
+      "step": 4450
+    },
+    {
+      "epoch": 0.49824231184432693,
+      "grad_norm": 0.21363414824008942,
+      "learning_rate": 5.0269596114383844e-05,
+      "loss": 0.5226,
+      "step": 4500
+    },
+    {
+      "epoch": 0.5037783375314862,
+      "grad_norm": 0.21449269354343414,
+      "learning_rate": 4.939994272667855e-05,
+      "loss": 0.5194,
+      "step": 4550
+    },
+    {
+      "epoch": 0.5093143632186453,
+      "grad_norm": 0.2134004533290863,
+      "learning_rate": 4.853047086956558e-05,
+      "loss": 0.5201,
+      "step": 4600
+    },
+    {
+      "epoch": 0.5148503889058045,
+      "grad_norm": 0.2227301150560379,
+      "learning_rate": 4.766144357750563e-05,
+      "loss": 0.5231,
+      "step": 4650
+    },
+    {
+      "epoch": 0.5203864145929638,
+      "grad_norm": 0.21659956872463226,
+      "learning_rate": 4.679312375046853e-05,
+      "loss": 0.5186,
+      "step": 4700
+    },
+    {
+      "epoch": 0.5259224402801229,
+      "grad_norm": 0.22292132675647736,
+      "learning_rate": 4.5925774074400415e-05,
+      "loss": 0.517,
+      "step": 4750
+    },
+    {
+      "epoch": 0.5314584659672821,
+      "grad_norm": 0.21309538185596466,
+      "learning_rate": 4.5059656941755244e-05,
+      "loss": 0.5207,
+      "step": 4800
+    },
+    {
+      "epoch": 0.5369944916544412,
+      "grad_norm": 0.222327321767807,
+      "learning_rate": 4.419503437211533e-05,
+      "loss": 0.5184,
+      "step": 4850
+    },
+    {
+      "epoch": 0.5425305173416005,
+      "grad_norm": 0.21690316498279572,
+      "learning_rate": 4.333216793292463e-05,
+      "loss": 0.5176,
+      "step": 4900
+    },
+    {
+      "epoch": 0.5480665430287597,
+      "grad_norm": 0.21055321395397186,
+      "learning_rate": 4.247131866035886e-05,
+      "loss": 0.515,
+      "step": 4950
+    },
+    {
+      "epoch": 0.5536025687159188,
+      "grad_norm": 0.2189202904701233,
+      "learning_rate": 4.161274698035618e-05,
+      "loss": 0.5128,
+      "step": 5000
+    },
+    {
+      "epoch": 0.559138594403078,
+      "grad_norm": 0.2133471965789795,
+      "learning_rate": 4.0756712629832675e-05,
+      "loss": 0.5141,
+      "step": 5050
+    },
+    {
+      "epoch": 0.5646746200902372,
+      "grad_norm": 0.2180759310722351,
+      "learning_rate": 3.990347457810612e-05,
+      "loss": 0.5151,
+      "step": 5100
+    },
+    {
+      "epoch": 0.5702106457773964,
+      "grad_norm": 0.22086185216903687,
+      "learning_rate": 3.9053290948552164e-05,
+      "loss": 0.5151,
+      "step": 5150
+    },
+    {
+      "epoch": 0.5757466714645556,
+      "grad_norm": 0.2056632786989212,
+      "learning_rate": 3.820641894051613e-05,
+      "loss": 0.5131,
+      "step": 5200
+    },
+    {
+      "epoch": 0.5812826971517148,
+      "grad_norm": 0.21264417469501495,
+      "learning_rate": 3.736311475150464e-05,
+      "loss": 0.5116,
+      "step": 5250
+    },
+    {
+      "epoch": 0.5868187228388739,
+      "grad_norm": 0.2104145884513855,
+      "learning_rate": 3.65236334996802e-05,
+      "loss": 0.5107,
+      "step": 5300
+    },
+    {
+      "epoch": 0.5923547485260332,
+      "grad_norm": 0.21011343598365784,
+      "learning_rate": 3.568822914668225e-05,
+      "loss": 0.5109,
+      "step": 5350
+    },
+    {
+      "epoch": 0.5978907742131924,
+      "grad_norm": 0.20390474796295166,
+      "learning_rate": 3.4857154420798246e-05,
+      "loss": 0.5103,
+      "step": 5400
+    },
+    {
+      "epoch": 0.6034267999003515,
+      "grad_norm": 0.211042582988739,
+      "learning_rate": 3.4030660740507654e-05,
+      "loss": 0.5125,
+      "step": 5450
+    },
+    {
+      "epoch": 0.6089628255875107,
+      "grad_norm": 0.20937225222587585,
+      "learning_rate": 3.32089981384224e-05,
+      "loss": 0.5083,
+      "step": 5500
+    },
+    {
+      "epoch": 0.61449885127467,
+      "grad_norm": 0.21997790038585663,
+      "learning_rate": 3.239241518564649e-05,
+      "loss": 0.5075,
+      "step": 5550
+    },
+    {
+      "epoch": 0.6200348769618291,
+      "grad_norm": 0.21026423573493958,
+      "learning_rate": 3.158115891657757e-05,
+      "loss": 0.5059,
+      "step": 5600
+    },
+    {
+      "epoch": 0.6255709026489883,
+      "grad_norm": 0.2174540013074875,
+      "learning_rate": 3.0775474754173836e-05,
+      "loss": 0.5068,
+      "step": 5650
+    },
+    {
+      "epoch": 0.6311069283361475,
+      "grad_norm": 0.34200409054756165,
+      "learning_rate": 2.9975606435707905e-05,
+      "loss": 0.5057,
+      "step": 5700
+    },
+    {
+      "epoch": 0.6366429540233066,
+      "grad_norm": 0.22012512385845184,
+      "learning_rate": 2.9181795939031088e-05,
+      "loss": 0.5041,
+      "step": 5750
+    },
+    {
+      "epoch": 0.6421789797104659,
+      "grad_norm": 0.2161373794078827,
+      "learning_rate": 2.839428340936966e-05,
+      "loss": 0.5071,
+      "step": 5800
+    },
+    {
+      "epoch": 0.647715005397625,
+      "grad_norm": 0.21077008545398712,
+      "learning_rate": 2.761330708667557e-05,
+      "loss": 0.5032,
+      "step": 5850
+    },
+    {
+      "epoch": 0.6532510310847842,
+      "grad_norm": 0.21698136627674103,
+      "learning_rate": 2.6839103233553692e-05,
+      "loss": 0.502,
+      "step": 5900
+    },
+    {
+      "epoch": 0.6587870567719434,
+      "grad_norm": 0.21964408457279205,
+      "learning_rate": 2.6071906063787154e-05,
+      "loss": 0.4997,
+      "step": 5950
+    },
+    {
+      "epoch": 0.6643230824591027,
+      "grad_norm": 0.21225056052207947,
+      "learning_rate": 2.5311947671482373e-05,
+      "loss": 0.5028,
+      "step": 6000
+    },
+    {
+      "epoch": 0.6698591081462618,
+      "grad_norm": 0.20255029201507568,
+      "learning_rate": 2.4559457960855564e-05,
+      "loss": 0.5004,
+      "step": 6050
+    },
+    {
+      "epoch": 0.675395133833421,
+      "grad_norm": 0.2110464870929718,
+      "learning_rate": 2.381466457668165e-05,
+      "loss": 0.5032,
+      "step": 6100
+    },
+    {
+      "epoch": 0.6809311595205801,
+      "grad_norm": 0.2102268934249878,
+      "learning_rate": 2.307779283542655e-05,
+      "loss": 0.4993,
+      "step": 6150
+    },
+    {
+      "epoch": 0.6864671852077394,
+      "grad_norm": 0.20706956088542938,
+      "learning_rate": 2.234906565708412e-05,
+      "loss": 0.4993,
+      "step": 6200
+    },
+    {
+      "epoch": 0.6920032108948986,
+      "grad_norm": 0.20903447270393372,
+      "learning_rate": 2.162870349773785e-05,
+      "loss": 0.5021,
+      "step": 6250
+    },
+    {
+      "epoch": 0.6975392365820577,
+      "grad_norm": 1.246344804763794,
+      "learning_rate": 2.0916924282868017e-05,
+      "loss": 0.5,
+      "step": 6300
+    },
+    {
+      "epoch": 0.7030752622692169,
+      "grad_norm": 0.217108353972435,
+      "learning_rate": 2.021394334142444e-05,
+      "loss": 0.501,
+      "step": 6350
+    },
+    {
+      "epoch": 0.7086112879563761,
+      "grad_norm": 0.21456560492515564,
+      "learning_rate": 1.9519973340684694e-05,
+      "loss": 0.5004,
+      "step": 6400
+    },
+    {
+      "epoch": 0.7141473136435353,
+      "grad_norm": 0.4793643355369568,
+      "learning_rate": 1.8835224221917575e-05,
+      "loss": 0.4963,
+      "step": 6450
+    },
+    {
+      "epoch": 0.7196833393306945,
+      "grad_norm": 0.20427371561527252,
+      "learning_rate": 1.8159903136871163e-05,
+      "loss": 0.4983,
+      "step": 6500
+    },
+    {
+      "epoch": 0.7252193650178537,
+      "grad_norm": 0.21231062710285187,
+      "learning_rate": 1.749421438510494e-05,
+      "loss": 0.4981,
+      "step": 6550
+    },
+    {
+      "epoch": 0.7307553907050128,
+      "grad_norm": 0.2009134292602539,
+      "learning_rate": 1.6838359352184576e-05,
+      "loss": 0.4962,
+      "step": 6600
+    },
+    {
+      "epoch": 0.7362914163921721,
+      "grad_norm": 0.21329347789287567,
+      "learning_rate": 1.6192536448758384e-05,
+      "loss": 0.4924,
+      "step": 6650
+    },
+    {
+      "epoch": 0.7418274420793313,
+      "grad_norm": 0.20288462936878204,
+      "learning_rate": 1.5556941050533636e-05,
+      "loss": 0.4933,
+      "step": 6700
+    },
+    {
+      "epoch": 0.7473634677664904,
+      "grad_norm": 0.20821629464626312,
+      "learning_rate": 1.4931765439171253e-05,
+      "loss": 0.4937,
+      "step": 6750
+    },
+    {
+      "epoch": 0.7528994934536496,
+      "grad_norm": 0.21015210449695587,
+      "learning_rate": 1.43171987441162e-05,
+      "loss": 0.4946,
+      "step": 6800
+    },
+    {
+      "epoch": 0.7584355191408089,
+      "grad_norm": 0.2095811516046524,
+      "learning_rate": 1.3713426885381659e-05,
+      "loss": 0.4902,
+      "step": 6850
+    },
+    {
+      "epoch": 0.763971544827968,
+      "grad_norm": 0.20578137040138245,
+      "learning_rate": 1.3120632517304204e-05,
+      "loss": 0.4942,
+      "step": 6900
+    },
+    {
+      "epoch": 0.7695075705151272,
+      "grad_norm": 0.20409756898880005,
+      "learning_rate": 1.2538994973286766e-05,
+      "loss": 0.4918,
+      "step": 6950
+    },
+    {
+      "epoch": 0.7750435962022864,
+      "grad_norm": 0.20583651959896088,
+      "learning_rate": 1.1968690211546323e-05,
+      "loss": 0.4894,
+      "step": 7000
+    },
+    {
+      "epoch": 0.7805796218894455,
+      "grad_norm": 0.2010050117969513,
+      "learning_rate": 1.1409890761882624e-05,
+      "loss": 0.4926,
+      "step": 7050
+    },
+    {
+      "epoch": 0.7861156475766048,
+      "grad_norm": 0.2147577702999115,
+      "learning_rate": 1.086276567348415e-05,
+      "loss": 0.4912,
+      "step": 7100
+    },
+    {
+      "epoch": 0.791651673263764,
+      "grad_norm": 0.21157945692539215,
+      "learning_rate": 1.0327480463787014e-05,
+      "loss": 0.4923,
+      "step": 7150
+    },
+    {
+      "epoch": 0.7971876989509231,
+      "grad_norm": 0.21193242073059082,
+      "learning_rate": 9.804197068402181e-06,
+      "loss": 0.4911,
+      "step": 7200
+    },
+    {
+      "epoch": 0.8027237246380823,
+      "grad_norm": 0.20365694165229797,
+      "learning_rate": 9.293073792126483e-06,
+      "loss": 0.4896,
+      "step": 7250
+    },
+    {
+      "epoch": 0.8082597503252416,
+      "grad_norm": 0.20320066809654236,
+      "learning_rate": 8.794265261051837e-06,
+      "loss": 0.4887,
+      "step": 7300
+    },
+    {
+      "epoch": 0.8137957760124007,
+      "grad_norm": 0.20599696040153503,
+      "learning_rate": 8.307922375787452e-06,
+      "loss": 0.4921,
+      "step": 7350
+    },
+    {
+      "epoch": 0.8193318016995599,
+      "grad_norm": 0.2039978951215744,
+      "learning_rate": 7.834192265809086e-06,
+      "loss": 0.4882,
+      "step": 7400
+    },
+    {
+      "epoch": 0.824867827386719,
+      "grad_norm": 0.22645381093025208,
+      "learning_rate": 7.3732182449491035e-06,
+      "loss": 0.4883,
+      "step": 7450
+    },
+    {
+      "epoch": 0.8304038530738783,
+      "grad_norm": 0.20420923829078674,
+      "learning_rate": 6.925139768040861e-06,
+      "loss": 0.4926,
+      "step": 7500
+    },
+    {
+      "epoch": 0.8359398787610375,
+      "grad_norm": 0.20432165265083313,
+      "learning_rate": 6.490092388730573e-06,
+      "loss": 0.4892,
+      "step": 7550
+    },
+    {
+      "epoch": 0.8414759044481966,
+      "grad_norm": 0.20835673809051514,
+      "learning_rate": 6.068207718469299e-06,
+      "loss": 0.4909,
+      "step": 7600
+    },
+    {
+      "epoch": 0.8470119301353558,
+      "grad_norm": 0.208354651927948,
+      "learning_rate": 5.659613386697599e-06,
+      "loss": 0.4858,
+      "step": 7650
+    },
+    {
+      "epoch": 0.852547955822515,
+      "grad_norm": 0.20932409167289734,
+      "learning_rate": 5.26443300223477e-06,
+      "loss": 0.4874,
+      "step": 7700
+    },
+    {
+      "epoch": 0.8580839815096742,
+      "grad_norm": 0.2009408324956894,
+      "learning_rate": 4.882786115884486e-06,
+      "loss": 0.4868,
+      "step": 7750
+    },
+    {
+      "epoch": 0.8636200071968334,
+      "grad_norm": 0.2020995020866394,
+      "learning_rate": 4.5147881842680155e-06,
+      "loss": 0.4869,
+      "step": 7800
+    },
+    {
+      "epoch": 0.8691560328839926,
+      "grad_norm": 0.2013290524482727,
+      "learning_rate": 4.1605505348960575e-06,
+      "loss": 0.4874,
+      "step": 7850
+    },
+    {
+      "epoch": 0.8746920585711517,
+      "grad_norm": 0.20548592507839203,
+      "learning_rate": 3.820180332489653e-06,
+      "loss": 0.4861,
+      "step": 7900
+    },
+    {
+      "epoch": 0.880228084258311,
+      "grad_norm": 0.2057260423898697,
+      "learning_rate": 3.493780546560588e-06,
+      "loss": 0.486,
+      "step": 7950
+    },
+    {
+      "epoch": 0.8857641099454702,
+      "grad_norm": 0.20100025832653046,
+      "learning_rate": 3.1814499202607375e-06,
+      "loss": 0.4897,
+      "step": 8000
+    },
+    {
+      "epoch": 0.8913001356326293,
+      "grad_norm": 0.1988091915845871,
+      "learning_rate": 2.8832829405101234e-06,
+      "loss": 0.4853,
+      "step": 8050
+    },
+    {
+      "epoch": 0.8968361613197885,
+      "grad_norm": 0.2371445745229721,
+      "learning_rate": 2.5993698094125496e-06,
+      "loss": 0.482,
+      "step": 8100
+    },
+    {
+      "epoch": 0.9023721870069477,
+      "grad_norm": 0.39132463932037354,
+      "learning_rate": 2.329796416967445e-06,
+      "loss": 0.488,
+      "step": 8150
+    },
+    {
+      "epoch": 0.9079082126941069,
+      "grad_norm": 0.204506054520607,
+      "learning_rate": 2.074644315086244e-06,
+      "loss": 0.486,
+      "step": 8200
+    },
+    {
+      "epoch": 0.9134442383812661,
+      "grad_norm": 0.20363277196884155,
+      "learning_rate": 1.8339906929211338e-06,
+      "loss": 0.4861,
+      "step": 8250
+    },
+    {
+      "epoch": 0.9189802640684253,
+      "grad_norm": 0.20154082775115967,
+      "learning_rate": 1.607908353513632e-06,
+      "loss": 0.4868,
+      "step": 8300
+    },
+    {
+      "epoch": 0.9245162897555844,
+      "grad_norm": 0.19429387152194977,
+      "learning_rate": 1.3964656917700491e-06,
+      "loss": 0.4842,
+      "step": 8350
+    },
+    {
+      "epoch": 0.9300523154427437,
+      "grad_norm": 0.2060394138097763,
+      "learning_rate": 1.1997266737705294e-06,
+      "loss": 0.4851,
+      "step": 8400
+    },
+    {
+      "epoch": 0.9355883411299029,
+      "grad_norm": 0.19958224892616272,
+      "learning_rate": 1.017750817417873e-06,
+      "loss": 0.4812,
+      "step": 8450
+    },
+    {
+      "epoch": 0.941124366817062,
+      "grad_norm": 0.19916601479053497,
+      "learning_rate": 8.50593174432096e-07,
+      "loss": 0.4836,
+      "step": 8500
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 9031,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.7611267462793265e+19,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cf146d6399bd68012f9046fdacf96ae2259b76d7299e5939ff6de3d97e69ad93
+size 6968

zero_to_fp32.py

@@ -0,0 +1,674 @@
+#!/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 json
+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)
+
+        if BUFFER_NAMES not in state_dict:
+            raise ValueError(f"{file} is not a model state checkpoint")
+        buffer_names = state_dict[BUFFER_NAMES]
+        if debug:
+            print("Found buffers:", buffer_names)
+
+        # recover just the buffers while restoring them to fp32 if they were saved in fp16
+        buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
+        param_shapes = state_dict[PARAM_SHAPES]
+
+        # collect parameters that are included in param_shapes
+        param_names = []
+        for s in param_shapes:
+            for name in s.keys():
+                param_names.append(name)
+
+        # update with frozen parameters
+        frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
+        if frozen_param_shapes is not None:
+            if debug:
+                print(f"Found frozen_param_shapes: {frozen_param_shapes}")
+            param_names += list(frozen_param_shapes.keys())
+
+        # handle shared params
+        shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
+
+        ds_version = state_dict.get(DS_VERSION, None)
+
+        frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
+
+        z_model_state = zero_model_state(buffers=buffers,
+                                         param_shapes=param_shapes,
+                                         shared_params=shared_params,
+                                         ds_version=ds_version,
+                                         frozen_param_shapes=frozen_param_shapes,
+                                         frozen_param_fragments=frozen_param_fragments)
+        zero_model_states.append(z_model_state)
+
+    return zero_model_states
+
+
+def parse_optim_states(files, ds_checkpoint_dir):
+    total_files = len(files)
+    state_dicts = []
+    for f in files:
+        state_dict = torch.load(f, map_location=device)
+        # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
+        # and also handle the case where it was already removed by another helper script
+        state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
+        state_dicts.append(state_dict)
+
+    if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
+        raise ValueError(f"{files[0]} is not a zero checkpoint")
+    zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
+    world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
+
+    # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
+    # parameters can be different from data parallelism for non-expert parameters. So we can just
+    # use the max of the partition_count to get the dp world_size.
+
+    if type(world_size) is list:
+        world_size = max(world_size)
+
+    if world_size != total_files:
+        raise ValueError(
+            f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
+            "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
+        )
+
+    # the groups are named differently in each stage
+    if zero_stage <= 2:
+        fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
+    elif zero_stage == 3:
+        fp32_groups_key = FP32_FLAT_GROUPS
+    else:
+        raise ValueError(f"unknown zero stage {zero_stage}")
+
+    if zero_stage <= 2:
+        fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
+    elif zero_stage == 3:
+        # if there is more than one param group, there will be multiple flattened tensors - one
+        # flattened tensor per group - for simplicity merge them into a single tensor
+        #
+        # XXX: could make the script more memory efficient for when there are multiple groups - it
+        # will require matching the sub-lists of param_shapes for each param group flattened tensor
+
+        fp32_flat_groups = [
+            torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
+        ]
+
+    return zero_stage, world_size, fp32_flat_groups
+
+
+def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters):
+    """
+    Returns fp32 state_dict reconstructed from ds checkpoint
+
+    Args:
+        - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
+
+    """
+    print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
+
+    optim_files = get_optim_files(ds_checkpoint_dir)
+    zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
+    print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
+
+    model_files = get_model_state_files(ds_checkpoint_dir)
+
+    zero_model_states = parse_model_states(model_files)
+    print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
+
+    if zero_stage <= 2:
+        return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+    elif zero_stage == 3:
+        return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+
+
+def _zero2_merge_frozen_params(state_dict, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+    frozen_param_fragments = zero_model_states[0].frozen_param_fragments
+
+    if debug:
+        num_elem = sum(s.numel() for s in frozen_param_shapes.values())
+        print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        state_dict[name] = frozen_param_fragments[name]
+
+        if debug:
+            print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _has_callable(obj, fn):
+    attr = getattr(obj, fn, None)
+    return callable(attr)
+
+
+def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+
+    # Reconstruction protocol:
+    #
+    # XXX: document this
+
+    if debug:
+        for i in range(world_size):
+            for j in range(len(fp32_flat_groups[0])):
+                print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
+
+    # XXX: memory usage doubles here (zero2)
+    num_param_groups = len(fp32_flat_groups[0])
+    merged_single_partition_of_fp32_groups = []
+    for i in range(num_param_groups):
+        merged_partitions = [sd[i] for sd in fp32_flat_groups]
+        full_single_fp32_vector = torch.cat(merged_partitions, 0)
+        merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
+    avail_numel = sum(
+        [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
+
+    if debug:
+        wanted_params = sum([len(shapes) for shapes in param_shapes])
+        wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
+        # not asserting if there is a mismatch due to possible padding
+        print(f"Have {avail_numel} numels to process.")
+        print(f"Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    total_numel = 0
+    total_params = 0
+    for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
+        offset = 0
+        avail_numel = full_single_fp32_vector.numel()
+        for name, shape in shapes.items():
+
+            unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
+            total_numel += unpartitioned_numel
+            total_params += 1
+
+            if debug:
+                print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+            state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
+            offset += unpartitioned_numel
+
+        # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
+        # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
+        # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
+        # live optimizer object, so we are checking that the numbers are within the right range
+        align_to = 2 * world_size
+
+        def zero2_align(x):
+            return align_to * math.ceil(x / align_to)
+
+        if debug:
+            print(f"original offset={offset}, avail_numel={avail_numel}")
+
+        offset = zero2_align(offset)
+        avail_numel = zero2_align(avail_numel)
+
+        if debug:
+            print(f"aligned  offset={offset}, avail_numel={avail_numel}")
+
+        # Sanity check
+        if offset != avail_numel:
+            raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero2_merge_frozen_params(state_dict, zero_model_states)
+
+    _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def zero3_partitioned_param_info(unpartitioned_numel, world_size):
+    remainder = unpartitioned_numel % world_size
+    padding_numel = (world_size - remainder) if remainder else 0
+    partitioned_numel = math.ceil(unpartitioned_numel / world_size)
+    return partitioned_numel, padding_numel
+
+
+def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    if debug:
+        for i in range(world_size):
+            num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
+            print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in zero_model_states[0].frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
+        state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
+
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+    avail_numel = fp32_flat_groups[0].numel() * world_size
+    # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
+    # param, re-consolidating each param, while dealing with padding if any
+
+    # merge list of dicts, preserving order
+    param_shapes = {k: v for d in param_shapes for k, v in d.items()}
+
+    if debug:
+        for i in range(world_size):
+            print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
+
+        wanted_params = len(param_shapes)
+        wanted_numel = sum(shape.numel() for shape in param_shapes.values())
+        # not asserting if there is a mismatch due to possible padding
+        avail_numel = fp32_flat_groups[0].numel() * world_size
+        print(f"Trainable params: Have {avail_numel} numels to process.")
+        print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    offset = 0
+    total_numel = 0
+    total_params = 0
+    for name, shape in 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}"
+            )
+
+        # XXX: memory usage doubles here
+        state_dict[name] = torch.cat(
+            tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
+            0).narrow(0, 0, unpartitioned_numel).view(shape)
+        offset += partitioned_numel
+
+    offset *= world_size
+
+    # Sanity check
+    if offset != avail_numel:
+        raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
+
+    _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
+    ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
+    via a model hub.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+
+    Returns:
+        - pytorch ``state_dict``
+
+    Note: this approach may not work if your application doesn't have sufficient free CPU memory and
+    you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
+    the checkpoint.
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+        # do the training and checkpoint saving
+        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
+        model = model.cpu() # move to cpu
+        model.load_state_dict(state_dict)
+        # submit to model hub or save the model to share with others
+
+    In this example the ``model`` will no longer be usable in the deepspeed context of the same
+    application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
+
+    """
+    if tag is None:
+        latest_path = os.path.join(checkpoint_dir, 'latest')
+        if os.path.isfile(latest_path):
+            with open(latest_path, 'r') as fd:
+                tag = fd.read().strip()
+        else:
+            raise ValueError(f"Unable to find 'latest' file at {latest_path}")
+
+    ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
+
+    if not os.path.isdir(ds_checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
+
+    return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
+
+
+def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir,
+                                               output_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)
+
+    # 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")
+        state_dict_split = split_torch_state_dict_into_shards(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
+    filename_to_tensors = state_dict_split.filename_to_tensors.items()
+    for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"):
+        shard = {tensor: state_dict[tensor].contiguous() for tensor in tensors}
+        output_path = os.path.join(output_dir, shard_file)
+        if safe_serialization:
+            save_file(shard, output_path, metadata={"format": "pt"})
+        else:
+            torch.save(shard, output_path)
+
+    # 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)