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/Qwen__Qwen2.5-1.5B-annealing_continual-0.0002LR-8192CL-2GAS-2BS-1EPOCHS-0.9BETA1-0.95BETA2/",
+  "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_step13267

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f4b5b37446d5d6a79e85a543dc616b531a8701e7e7a53d275b8aaa67eda91543
+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 +1,1888 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500.0,
+  "global_step": 13267,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.003768749528906309,
+      "grad_norm": 0.35509032011032104,
+      "learning_rate": 9.999649547444612e-05,
+      "loss": 0.5094,
+      "step": 50
+    },
+    {
+      "epoch": 0.007537499057812618,
+      "grad_norm": 0.3374439477920532,
+      "learning_rate": 9.998598238905239e-05,
+      "loss": 0.4888,
+      "step": 100
+    },
+    {
+      "epoch": 0.011306248586718927,
+      "grad_norm": 0.3017200827598572,
+      "learning_rate": 9.996846221755392e-05,
+      "loss": 0.4871,
+      "step": 150
+    },
+    {
+      "epoch": 0.015074998115625236,
+      "grad_norm": 0.29322266578674316,
+      "learning_rate": 9.994393741594623e-05,
+      "loss": 0.4899,
+      "step": 200
+    },
+    {
+      "epoch": 0.018843747644531544,
+      "grad_norm": 0.3081373870372772,
+      "learning_rate": 9.99124114221411e-05,
+      "loss": 0.4896,
+      "step": 250
+    },
+    {
+      "epoch": 0.022612497173437853,
+      "grad_norm": 0.29016199707984924,
+      "learning_rate": 9.987388865548454e-05,
+      "loss": 0.4889,
+      "step": 300
+    },
+    {
+      "epoch": 0.026381246702344163,
+      "grad_norm": 0.265391081571579,
+      "learning_rate": 9.982837451613738e-05,
+      "loss": 0.4898,
+      "step": 350
+    },
+    {
+      "epoch": 0.030149996231250472,
+      "grad_norm": 0.27272671461105347,
+      "learning_rate": 9.977587538431816e-05,
+      "loss": 0.4894,
+      "step": 400
+    },
+    {
+      "epoch": 0.03391874576015678,
+      "grad_norm": 0.28726664185523987,
+      "learning_rate": 9.971639861940889e-05,
+      "loss": 0.4869,
+      "step": 450
+    },
+    {
+      "epoch": 0.03768749528906309,
+      "grad_norm": 0.28651759028434753,
+      "learning_rate": 9.964995255892323e-05,
+      "loss": 0.4912,
+      "step": 500
+    },
+    {
+      "epoch": 0.0414562448179694,
+      "grad_norm": 0.24553848803043365,
+      "learning_rate": 9.957654651733788e-05,
+      "loss": 0.4897,
+      "step": 550
+    },
+    {
+      "epoch": 0.045224994346875706,
+      "grad_norm": 0.24010591208934784,
+      "learning_rate": 9.949619078478677e-05,
+      "loss": 0.4866,
+      "step": 600
+    },
+    {
+      "epoch": 0.048993743875782016,
+      "grad_norm": 0.26084381341934204,
+      "learning_rate": 9.940889662561864e-05,
+      "loss": 0.4892,
+      "step": 650
+    },
+    {
+      "epoch": 0.052762493404688325,
+      "grad_norm": 0.2248304784297943,
+      "learning_rate": 9.931467627681792e-05,
+      "loss": 0.4849,
+      "step": 700
+    },
+    {
+      "epoch": 0.056531242933594635,
+      "grad_norm": 0.23623178899288177,
+      "learning_rate": 9.921354294628944e-05,
+      "loss": 0.4852,
+      "step": 750
+    },
+    {
+      "epoch": 0.060299992462500944,
+      "grad_norm": 0.23275640606880188,
+      "learning_rate": 9.910551081100684e-05,
+      "loss": 0.4855,
+      "step": 800
+    },
+    {
+      "epoch": 0.06406874199140725,
+      "grad_norm": 0.22925056517124176,
+      "learning_rate": 9.899059501502526e-05,
+      "loss": 0.4849,
+      "step": 850
+    },
+    {
+      "epoch": 0.06783749152031356,
+      "grad_norm": 0.23725946247577667,
+      "learning_rate": 9.886881166735846e-05,
+      "loss": 0.4839,
+      "step": 900
+    },
+    {
+      "epoch": 0.07160624104921987,
+      "grad_norm": 0.2293645143508911,
+      "learning_rate": 9.874017783972058e-05,
+      "loss": 0.486,
+      "step": 950
+    },
+    {
+      "epoch": 0.07537499057812617,
+      "grad_norm": 0.24602073431015015,
+      "learning_rate": 9.860471156413309e-05,
+      "loss": 0.4835,
+      "step": 1000
+    },
+    {
+      "epoch": 0.07914374010703248,
+      "grad_norm": 0.20568886399269104,
+      "learning_rate": 9.846243183039694e-05,
+      "loss": 0.4838,
+      "step": 1050
+    },
+    {
+      "epoch": 0.0829124896359388,
+      "grad_norm": 0.2165093570947647,
+      "learning_rate": 9.831335858343064e-05,
+      "loss": 0.4827,
+      "step": 1100
+    },
+    {
+      "epoch": 0.0866812391648451,
+      "grad_norm": 0.2424800992012024,
+      "learning_rate": 9.815751272047434e-05,
+      "loss": 0.4832,
+      "step": 1150
+    },
+    {
+      "epoch": 0.09044998869375141,
+      "grad_norm": 0.207057386636734,
+      "learning_rate": 9.79949160881604e-05,
+      "loss": 0.4809,
+      "step": 1200
+    },
+    {
+      "epoch": 0.09421873822265772,
+      "grad_norm": 0.21975122392177582,
+      "learning_rate": 9.782559147945094e-05,
+      "loss": 0.4827,
+      "step": 1250
+    },
+    {
+      "epoch": 0.09798748775156403,
+      "grad_norm": 0.21343478560447693,
+      "learning_rate": 9.76495626304427e-05,
+      "loss": 0.4812,
+      "step": 1300
+    },
+    {
+      "epoch": 0.10175623728047034,
+      "grad_norm": 0.20896418392658234,
+      "learning_rate": 9.746685421703961e-05,
+      "loss": 0.4792,
+      "step": 1350
+    },
+    {
+      "epoch": 0.10552498680937665,
+      "grad_norm": 0.2270091027021408,
+      "learning_rate": 9.727749185149388e-05,
+      "loss": 0.4795,
+      "step": 1400
+    },
+    {
+      "epoch": 0.10929373633828296,
+      "grad_norm": 0.2058868557214737,
+      "learning_rate": 9.708150207881543e-05,
+      "loss": 0.4794,
+      "step": 1450
+    },
+    {
+      "epoch": 0.11306248586718927,
+      "grad_norm": 0.19969668984413147,
+      "learning_rate": 9.687891237305096e-05,
+      "loss": 0.4803,
+      "step": 1500
+    },
+    {
+      "epoch": 0.11683123539609558,
+      "grad_norm": 0.19804421067237854,
+      "learning_rate": 9.666975113343246e-05,
+      "loss": 0.4782,
+      "step": 1550
+    },
+    {
+      "epoch": 0.12059998492500189,
+      "grad_norm": 0.19650672376155853,
+      "learning_rate": 9.645404768039633e-05,
+      "loss": 0.4773,
+      "step": 1600
+    },
+    {
+      "epoch": 0.1243687344539082,
+      "grad_norm": 0.20196650922298431,
+      "learning_rate": 9.623183225147308e-05,
+      "loss": 0.4769,
+      "step": 1650
+    },
+    {
+      "epoch": 0.1281374839828145,
+      "grad_norm": 0.20083576440811157,
+      "learning_rate": 9.600313599704869e-05,
+      "loss": 0.4748,
+      "step": 1700
+    },
+    {
+      "epoch": 0.13190623351172082,
+      "grad_norm": 0.19036008417606354,
+      "learning_rate": 9.576799097599786e-05,
+      "loss": 0.4751,
+      "step": 1750
+    },
+    {
+      "epoch": 0.1356749830406271,
+      "grad_norm": 0.20416900515556335,
+      "learning_rate": 9.552643015118998e-05,
+      "loss": 0.4727,
+      "step": 1800
+    },
+    {
+      "epoch": 0.13944373256953344,
+      "grad_norm": 0.19550226628780365,
+      "learning_rate": 9.527848738486842e-05,
+      "loss": 0.4731,
+      "step": 1850
+    },
+    {
+      "epoch": 0.14321248209843973,
+      "grad_norm": 0.20287151634693146,
+      "learning_rate": 9.502419743390357e-05,
+      "loss": 0.4745,
+      "step": 1900
+    },
+    {
+      "epoch": 0.14698123162734605,
+      "grad_norm": 0.18741615116596222,
+      "learning_rate": 9.476359594492068e-05,
+      "loss": 0.4734,
+      "step": 1950
+    },
+    {
+      "epoch": 0.15074998115625235,
+      "grad_norm": 0.20605124533176422,
+      "learning_rate": 9.449671944930288e-05,
+      "loss": 0.4732,
+      "step": 2000
+    },
+    {
+      "epoch": 0.15451873068515867,
+      "grad_norm": 0.20259861648082733,
+      "learning_rate": 9.422360535807009e-05,
+      "loss": 0.4745,
+      "step": 2050
+    },
+    {
+      "epoch": 0.15828748021406497,
+      "grad_norm": 0.19558943808078766,
+      "learning_rate": 9.394429195663478e-05,
+      "loss": 0.4723,
+      "step": 2100
+    },
+    {
+      "epoch": 0.1620562297429713,
+      "grad_norm": 0.20177054405212402,
+      "learning_rate": 9.365881839943508e-05,
+      "loss": 0.4699,
+      "step": 2150
+    },
+    {
+      "epoch": 0.1658249792718776,
+      "grad_norm": 0.20023804903030396,
+      "learning_rate": 9.336722470444604e-05,
+      "loss": 0.4719,
+      "step": 2200
+    },
+    {
+      "epoch": 0.1695937288007839,
+      "grad_norm": 0.19571995735168457,
+      "learning_rate": 9.306955174756985e-05,
+      "loss": 0.4708,
+      "step": 2250
+    },
+    {
+      "epoch": 0.1733624783296902,
+      "grad_norm": 0.18980449438095093,
+      "learning_rate": 9.27658412569059e-05,
+      "loss": 0.4697,
+      "step": 2300
+    },
+    {
+      "epoch": 0.17713122785859653,
+      "grad_norm": 0.18121857941150665,
+      "learning_rate": 9.24561358069012e-05,
+      "loss": 0.4692,
+      "step": 2350
+    },
+    {
+      "epoch": 0.18089997738750282,
+      "grad_norm": 0.18635448813438416,
+      "learning_rate": 9.214047881238233e-05,
+      "loss": 0.4682,
+      "step": 2400
+    },
+    {
+      "epoch": 0.18466872691640915,
+      "grad_norm": 0.18292276561260223,
+      "learning_rate": 9.181891452246937e-05,
+      "loss": 0.4717,
+      "step": 2450
+    },
+    {
+      "epoch": 0.18843747644531544,
+      "grad_norm": 0.4070293605327606,
+      "learning_rate": 9.149148801437321e-05,
+      "loss": 0.4685,
+      "step": 2500
+    },
+    {
+      "epoch": 0.19220622597422174,
+      "grad_norm": 0.19017393887043,
+      "learning_rate": 9.115824518707644e-05,
+      "loss": 0.4675,
+      "step": 2550
+    },
+    {
+      "epoch": 0.19597497550312806,
+      "grad_norm": 0.2028086632490158,
+      "learning_rate": 9.08192327548992e-05,
+      "loss": 0.4668,
+      "step": 2600
+    },
+    {
+      "epoch": 0.19974372503203436,
+      "grad_norm": 0.18879903852939606,
+      "learning_rate": 9.047449824095075e-05,
+      "loss": 0.466,
+      "step": 2650
+    },
+    {
+      "epoch": 0.20351247456094068,
+      "grad_norm": 0.18708941340446472,
+      "learning_rate": 9.012408997046766e-05,
+      "loss": 0.467,
+      "step": 2700
+    },
+    {
+      "epoch": 0.20728122408984698,
+      "grad_norm": 0.18148259818553925,
+      "learning_rate": 8.976805706403942e-05,
+      "loss": 0.4657,
+      "step": 2750
+    },
+    {
+      "epoch": 0.2110499736187533,
+      "grad_norm": 0.18493063747882843,
+      "learning_rate": 8.94064494307228e-05,
+      "loss": 0.4638,
+      "step": 2800
+    },
+    {
+      "epoch": 0.2148187231476596,
+      "grad_norm": 0.18034948408603668,
+      "learning_rate": 8.903931776104545e-05,
+      "loss": 0.4624,
+      "step": 2850
+    },
+    {
+      "epoch": 0.21858747267656592,
+      "grad_norm": 0.18979419767856598,
+      "learning_rate": 8.866671351990007e-05,
+      "loss": 0.4629,
+      "step": 2900
+    },
+    {
+      "epoch": 0.22235622220547221,
+      "grad_norm": 0.18408875167369843,
+      "learning_rate": 8.82886889393301e-05,
+      "loss": 0.4638,
+      "step": 2950
+    },
+    {
+      "epoch": 0.22612497173437854,
+      "grad_norm": 0.17015992105007172,
+      "learning_rate": 8.790529701120759e-05,
+      "loss": 0.4608,
+      "step": 3000
+    },
+    {
+      "epoch": 0.22989372126328483,
+      "grad_norm": 0.17827536165714264,
+      "learning_rate": 8.751659147980493e-05,
+      "loss": 0.4635,
+      "step": 3050
+    },
+    {
+      "epoch": 0.23366247079219116,
+      "grad_norm": 0.1894233673810959,
+      "learning_rate": 8.712262683426082e-05,
+      "loss": 0.4593,
+      "step": 3100
+    },
+    {
+      "epoch": 0.23743122032109745,
+      "grad_norm": 0.19764114916324615,
+      "learning_rate": 8.672345830094199e-05,
+      "loss": 0.4622,
+      "step": 3150
+    },
+    {
+      "epoch": 0.24119996985000378,
+      "grad_norm": 0.18290351331233978,
+      "learning_rate": 8.631914183570143e-05,
+      "loss": 0.4608,
+      "step": 3200
+    },
+    {
+      "epoch": 0.24496871937891007,
+      "grad_norm": 0.18013353645801544,
+      "learning_rate": 8.590973411603452e-05,
+      "loss": 0.4601,
+      "step": 3250
+    },
+    {
+      "epoch": 0.2487374689078164,
+      "grad_norm": 0.17729552090168,
+      "learning_rate": 8.549529253313386e-05,
+      "loss": 0.4611,
+      "step": 3300
+    },
+    {
+      "epoch": 0.2525062184367227,
+      "grad_norm": 0.1892414540052414,
+      "learning_rate": 8.507587518384421e-05,
+      "loss": 0.4583,
+      "step": 3350
+    },
+    {
+      "epoch": 0.256274967965629,
+      "grad_norm": 0.17193005979061127,
+      "learning_rate": 8.465154086251828e-05,
+      "loss": 0.4572,
+      "step": 3400
+    },
+    {
+      "epoch": 0.2600437174945353,
+      "grad_norm": 0.18148685991764069,
+      "learning_rate": 8.422234905277495e-05,
+      "loss": 0.4583,
+      "step": 3450
+    },
+    {
+      "epoch": 0.26381246702344163,
+      "grad_norm": 0.19143982231616974,
+      "learning_rate": 8.378835991916083e-05,
+      "loss": 0.4582,
+      "step": 3500
+    },
+    {
+      "epoch": 0.26758121655234796,
+      "grad_norm": 0.18079186975955963,
+      "learning_rate": 8.334963429871627e-05,
+      "loss": 0.4599,
+      "step": 3550
+    },
+    {
+      "epoch": 0.2713499660812542,
+      "grad_norm": 0.17887386679649353,
+      "learning_rate": 8.290623369244721e-05,
+      "loss": 0.4574,
+      "step": 3600
+    },
+    {
+      "epoch": 0.27511871561016055,
+      "grad_norm": 0.17481209337711334,
+      "learning_rate": 8.245822025670384e-05,
+      "loss": 0.4588,
+      "step": 3650
+    },
+    {
+      "epoch": 0.27888746513906687,
+      "grad_norm": 0.17591702938079834,
+      "learning_rate": 8.200565679446753e-05,
+      "loss": 0.4543,
+      "step": 3700
+    },
+    {
+      "epoch": 0.2826562146679732,
+      "grad_norm": 0.17434370517730713,
+      "learning_rate": 8.154860674654698e-05,
+      "loss": 0.4552,
+      "step": 3750
+    },
+    {
+      "epoch": 0.28642496419687946,
+      "grad_norm": 0.17741286754608154,
+      "learning_rate": 8.108713418268514e-05,
+      "loss": 0.4551,
+      "step": 3800
+    },
+    {
+      "epoch": 0.2901937137257858,
+      "grad_norm": 0.1794031709432602,
+      "learning_rate": 8.062130379257764e-05,
+      "loss": 0.4557,
+      "step": 3850
+    },
+    {
+      "epoch": 0.2939624632546921,
+      "grad_norm": 0.17624689638614655,
+      "learning_rate": 8.015118087680477e-05,
+      "loss": 0.4558,
+      "step": 3900
+    },
+    {
+      "epoch": 0.2977312127835984,
+      "grad_norm": 0.173648402094841,
+      "learning_rate": 7.96768313376774e-05,
+      "loss": 0.4519,
+      "step": 3950
+    },
+    {
+      "epoch": 0.3014999623125047,
+      "grad_norm": 0.17087939381599426,
+      "learning_rate": 7.919832166999874e-05,
+      "loss": 0.454,
+      "step": 4000
+    },
+    {
+      "epoch": 0.305268711841411,
+      "grad_norm": 0.1744805872440338,
+      "learning_rate": 7.871571895174316e-05,
+      "loss": 0.4511,
+      "step": 4050
+    },
+    {
+      "epoch": 0.30903746137031735,
+      "grad_norm": 0.1831275224685669,
+      "learning_rate": 7.822909083465298e-05,
+      "loss": 0.4537,
+      "step": 4100
+    },
+    {
+      "epoch": 0.3128062108992236,
+      "grad_norm": 0.17621232569217682,
+      "learning_rate": 7.773850553475508e-05,
+      "loss": 0.4506,
+      "step": 4150
+    },
+    {
+      "epoch": 0.31657496042812994,
+      "grad_norm": 0.1809280514717102,
+      "learning_rate": 7.724403182279823e-05,
+      "loss": 0.4537,
+      "step": 4200
+    },
+    {
+      "epoch": 0.32034370995703626,
+      "grad_norm": 0.18568743765354156,
+      "learning_rate": 7.674573901461282e-05,
+      "loss": 0.4484,
+      "step": 4250
+    },
+    {
+      "epoch": 0.3241124594859426,
+      "grad_norm": 0.17346200346946716,
+      "learning_rate": 7.624369696139402e-05,
+      "loss": 0.4492,
+      "step": 4300
+    },
+    {
+      "epoch": 0.32788120901484885,
+      "grad_norm": 0.16987943649291992,
+      "learning_rate": 7.573797603991004e-05,
+      "loss": 0.4511,
+      "step": 4350
+    },
+    {
+      "epoch": 0.3316499585437552,
+      "grad_norm": 0.1740700751543045,
+      "learning_rate": 7.522864714263655e-05,
+      "loss": 0.4504,
+      "step": 4400
+    },
+    {
+      "epoch": 0.3354187080726615,
+      "grad_norm": 0.18099980056285858,
+      "learning_rate": 7.471578166781899e-05,
+      "loss": 0.4509,
+      "step": 4450
+    },
+    {
+      "epoch": 0.3391874576015678,
+      "grad_norm": 0.1742471605539322,
+      "learning_rate": 7.419945150946386e-05,
+      "loss": 0.4482,
+      "step": 4500
+    },
+    {
+      "epoch": 0.3429562071304741,
+      "grad_norm": 0.17314079403877258,
+      "learning_rate": 7.367972904726055e-05,
+      "loss": 0.4497,
+      "step": 4550
+    },
+    {
+      "epoch": 0.3467249566593804,
+      "grad_norm": 0.1672036498785019,
+      "learning_rate": 7.3156687136435e-05,
+      "loss": 0.4476,
+      "step": 4600
+    },
+    {
+      "epoch": 0.35049370618828674,
+      "grad_norm": 0.1716027557849884,
+      "learning_rate": 7.26303990975369e-05,
+      "loss": 0.4484,
+      "step": 4650
+    },
+    {
+      "epoch": 0.35426245571719306,
+      "grad_norm": 0.16599993407726288,
+      "learning_rate": 7.210093870616155e-05,
+      "loss": 0.4478,
+      "step": 4700
+    },
+    {
+      "epoch": 0.3580312052460993,
+      "grad_norm": 0.16980785131454468,
+      "learning_rate": 7.156838018260776e-05,
+      "loss": 0.4468,
+      "step": 4750
+    },
+    {
+      "epoch": 0.36179995477500565,
+      "grad_norm": 0.17415867745876312,
+      "learning_rate": 7.103279818147371e-05,
+      "loss": 0.4444,
+      "step": 4800
+    },
+    {
+      "epoch": 0.365568704303912,
+      "grad_norm": 0.17735563218593597,
+      "learning_rate": 7.049426778119179e-05,
+      "loss": 0.4454,
+      "step": 4850
+    },
+    {
+      "epoch": 0.3693374538328183,
+      "grad_norm": 0.1772989183664322,
+      "learning_rate": 6.995286447350397e-05,
+      "loss": 0.4456,
+      "step": 4900
+    },
+    {
+      "epoch": 0.37310620336172456,
+      "grad_norm": 0.17225749790668488,
+      "learning_rate": 6.940866415287931e-05,
+      "loss": 0.4453,
+      "step": 4950
+    },
+    {
+      "epoch": 0.3768749528906309,
+      "grad_norm": 0.16734232008457184,
+      "learning_rate": 6.886174310587501e-05,
+      "loss": 0.4429,
+      "step": 5000
+    },
+    {
+      "epoch": 0.3806437024195372,
+      "grad_norm": 0.17711064219474792,
+      "learning_rate": 6.831217800044252e-05,
+      "loss": 0.4455,
+      "step": 5050
+    },
+    {
+      "epoch": 0.3844124519484435,
+      "grad_norm": 0.16528938710689545,
+      "learning_rate": 6.776004587518001e-05,
+      "loss": 0.4452,
+      "step": 5100
+    },
+    {
+      "epoch": 0.3881812014773498,
+      "grad_norm": 0.16722093522548676,
+      "learning_rate": 6.720542412853319e-05,
+      "loss": 0.4427,
+      "step": 5150
+    },
+    {
+      "epoch": 0.3919499510062561,
+      "grad_norm": 0.16517098248004913,
+      "learning_rate": 6.66483905079454e-05,
+      "loss": 0.4424,
+      "step": 5200
+    },
+    {
+      "epoch": 0.39571870053516245,
+      "grad_norm": 0.16755063831806183,
+      "learning_rate": 6.608902309895895e-05,
+      "loss": 0.4405,
+      "step": 5250
+    },
+    {
+      "epoch": 0.3994874500640687,
+      "grad_norm": 0.1688978523015976,
+      "learning_rate": 6.552740031426902e-05,
+      "loss": 0.437,
+      "step": 5300
+    },
+    {
+      "epoch": 0.40325619959297504,
+      "grad_norm": 0.16191639006137848,
+      "learning_rate": 6.496360088273161e-05,
+      "loss": 0.4405,
+      "step": 5350
+    },
+    {
+      "epoch": 0.40702494912188136,
+      "grad_norm": 0.1776248961687088,
+      "learning_rate": 6.439770383832732e-05,
+      "loss": 0.4405,
+      "step": 5400
+    },
+    {
+      "epoch": 0.4107936986507877,
+      "grad_norm": 0.16206714510917664,
+      "learning_rate": 6.382978850908226e-05,
+      "loss": 0.44,
+      "step": 5450
+    },
+    {
+      "epoch": 0.41456244817969395,
+      "grad_norm": 0.16774949431419373,
+      "learning_rate": 6.325993450594782e-05,
+      "loss": 0.4405,
+      "step": 5500
+    },
+    {
+      "epoch": 0.4183311977086003,
+      "grad_norm": 0.16804030537605286,
+      "learning_rate": 6.26882217116406e-05,
+      "loss": 0.4386,
+      "step": 5550
+    },
+    {
+      "epoch": 0.4220999472375066,
+      "grad_norm": 0.16452039778232574,
+      "learning_rate": 6.211473026944452e-05,
+      "loss": 0.4369,
+      "step": 5600
+    },
+    {
+      "epoch": 0.4258686967664129,
+      "grad_norm": 0.15764504671096802,
+      "learning_rate": 6.153954057197612e-05,
+      "loss": 0.438,
+      "step": 5650
+    },
+    {
+      "epoch": 0.4296374462953192,
+      "grad_norm": 0.16407234966754913,
+      "learning_rate": 6.0962733249915135e-05,
+      "loss": 0.4366,
+      "step": 5700
+    },
+    {
+      "epoch": 0.4334061958242255,
+      "grad_norm": 0.16679194569587708,
+      "learning_rate": 6.038438916070155e-05,
+      "loss": 0.4381,
+      "step": 5750
+    },
+    {
+      "epoch": 0.43717494535313184,
+      "grad_norm": 0.16508112847805023,
+      "learning_rate": 5.9804589377200946e-05,
+      "loss": 0.4369,
+      "step": 5800
+    },
+    {
+      "epoch": 0.44094369488203816,
+      "grad_norm": 0.16879412531852722,
+      "learning_rate": 5.922341517633965e-05,
+      "loss": 0.4382,
+      "step": 5850
+    },
+    {
+      "epoch": 0.44471244441094443,
+      "grad_norm": 0.16117092967033386,
+      "learning_rate": 5.864094802771115e-05,
+      "loss": 0.4348,
+      "step": 5900
+    },
+    {
+      "epoch": 0.44848119393985075,
+      "grad_norm": 0.1632978767156601,
+      "learning_rate": 5.8057269582155735e-05,
+      "loss": 0.4371,
+      "step": 5950
+    },
+    {
+      "epoch": 0.4522499434687571,
+      "grad_norm": 0.16430360078811646,
+      "learning_rate": 5.7472461660314504e-05,
+      "loss": 0.435,
+      "step": 6000
+    },
+    {
+      "epoch": 0.4560186929976634,
+      "grad_norm": 0.16830819845199585,
+      "learning_rate": 5.6886606241159714e-05,
+      "loss": 0.4337,
+      "step": 6050
+    },
+    {
+      "epoch": 0.45978744252656967,
+      "grad_norm": 0.16006672382354736,
+      "learning_rate": 5.6299785450502853e-05,
+      "loss": 0.4336,
+      "step": 6100
+    },
+    {
+      "epoch": 0.463556192055476,
+      "grad_norm": 0.16786810755729675,
+      "learning_rate": 5.571208154948218e-05,
+      "loss": 0.4335,
+      "step": 6150
+    },
+    {
+      "epoch": 0.4673249415843823,
+      "grad_norm": 0.16502316296100616,
+      "learning_rate": 5.5123576923031253e-05,
+      "loss": 0.433,
+      "step": 6200
+    },
+    {
+      "epoch": 0.47109369111328864,
+      "grad_norm": 0.16036230325698853,
+      "learning_rate": 5.453435406833017e-05,
+      "loss": 0.4296,
+      "step": 6250
+    },
+    {
+      "epoch": 0.4748624406421949,
+      "grad_norm": 0.16277125477790833,
+      "learning_rate": 5.3944495583240987e-05,
+      "loss": 0.4349,
+      "step": 6300
+    },
+    {
+      "epoch": 0.47863119017110123,
+      "grad_norm": 0.1639643758535385,
+      "learning_rate": 5.3354084154729034e-05,
+      "loss": 0.4311,
+      "step": 6350
+    },
+    {
+      "epoch": 0.48239993970000755,
+      "grad_norm": 0.1611129492521286,
+      "learning_rate": 5.276320254727187e-05,
+      "loss": 0.4315,
+      "step": 6400
+    },
+    {
+      "epoch": 0.4861686892289138,
+      "grad_norm": 0.1649327427148819,
+      "learning_rate": 5.217193359125724e-05,
+      "loss": 0.433,
+      "step": 6450
+    },
+    {
+      "epoch": 0.48993743875782014,
+      "grad_norm": 0.16595204174518585,
+      "learning_rate": 5.15803601713717e-05,
+      "loss": 0.432,
+      "step": 6500
+    },
+    {
+      "epoch": 0.49370618828672647,
+      "grad_norm": 0.16455797851085663,
+      "learning_rate": 5.0988565214981976e-05,
+      "loss": 0.4291,
+      "step": 6550
+    },
+    {
+      "epoch": 0.4974749378156328,
+      "grad_norm": 0.16371013224124908,
+      "learning_rate": 5.0396631680509945e-05,
+      "loss": 0.4299,
+      "step": 6600
+    },
+    {
+      "epoch": 0.5012436873445391,
+      "grad_norm": 0.16786278784275055,
+      "learning_rate": 4.9804642545803524e-05,
+      "loss": 0.43,
+      "step": 6650
+    },
+    {
+      "epoch": 0.5050124368734454,
+      "grad_norm": 0.16633006930351257,
+      "learning_rate": 4.9212680796504704e-05,
+      "loss": 0.4289,
+      "step": 6700
+    },
+    {
+      "epoch": 0.5087811864023517,
+      "grad_norm": 0.1592586487531662,
+      "learning_rate": 4.8620829414416615e-05,
+      "loss": 0.4296,
+      "step": 6750
+    },
+    {
+      "epoch": 0.512549935931258,
+      "grad_norm": 0.16653411090373993,
+      "learning_rate": 4.8029171365870926e-05,
+      "loss": 0.4282,
+      "step": 6800
+    },
+    {
+      "epoch": 0.5163186854601644,
+      "grad_norm": 0.16303293406963348,
+      "learning_rate": 4.743778959009766e-05,
+      "loss": 0.4267,
+      "step": 6850
+    },
+    {
+      "epoch": 0.5200874349890706,
+      "grad_norm": 0.1592382937669754,
+      "learning_rate": 4.684676698759864e-05,
+      "loss": 0.4268,
+      "step": 6900
+    },
+    {
+      "epoch": 0.5238561845179769,
+      "grad_norm": 0.15816909074783325,
+      "learning_rate": 4.62561864085264e-05,
+      "loss": 0.4261,
+      "step": 6950
+    },
+    {
+      "epoch": 0.5276249340468833,
+      "grad_norm": 0.1695041060447693,
+      "learning_rate": 4.566613064107015e-05,
+      "loss": 0.427,
+      "step": 7000
+    },
+    {
+      "epoch": 0.5313936835757895,
+      "grad_norm": 0.16515901684761047,
+      "learning_rate": 4.507668239985055e-05,
+      "loss": 0.4263,
+      "step": 7050
+    },
+    {
+      "epoch": 0.5351624331046959,
+      "grad_norm": 0.15738603472709656,
+      "learning_rate": 4.448792431432451e-05,
+      "loss": 0.4277,
+      "step": 7100
+    },
+    {
+      "epoch": 0.5389311826336022,
+      "grad_norm": 0.17032016813755035,
+      "learning_rate": 4.389993891720232e-05,
+      "loss": 0.4262,
+      "step": 7150
+    },
+    {
+      "epoch": 0.5426999321625084,
+      "grad_norm": 0.1652156412601471,
+      "learning_rate": 4.3312808632877924e-05,
+      "loss": 0.4228,
+      "step": 7200
+    },
+    {
+      "epoch": 0.5464686816914148,
+      "grad_norm": 0.161549910902977,
+      "learning_rate": 4.27266157658747e-05,
+      "loss": 0.4231,
+      "step": 7250
+    },
+    {
+      "epoch": 0.5502374312203211,
+      "grad_norm": 0.15818439424037933,
+      "learning_rate": 4.214144248930797e-05,
+      "loss": 0.4238,
+      "step": 7300
+    },
+    {
+      "epoch": 0.5540061807492274,
+      "grad_norm": 0.15689703822135925,
+      "learning_rate": 4.155737083336575e-05,
+      "loss": 0.4242,
+      "step": 7350
+    },
+    {
+      "epoch": 0.5577749302781337,
+      "grad_norm": 0.16835862398147583,
+      "learning_rate": 4.097448267380979e-05,
+      "loss": 0.4246,
+      "step": 7400
+    },
+    {
+      "epoch": 0.56154367980704,
+      "grad_norm": 0.162080317735672,
+      "learning_rate": 4.03928597204981e-05,
+      "loss": 0.4204,
+      "step": 7450
+    },
+    {
+      "epoch": 0.5653124293359464,
+      "grad_norm": 0.16594427824020386,
+      "learning_rate": 3.9812583505930786e-05,
+      "loss": 0.4236,
+      "step": 7500
+    },
+    {
+      "epoch": 0.5690811788648527,
+      "grad_norm": 0.1567797213792801,
+      "learning_rate": 3.923373537382074e-05,
+      "loss": 0.422,
+      "step": 7550
+    },
+    {
+      "epoch": 0.5728499283937589,
+      "grad_norm": 0.1608632504940033,
+      "learning_rate": 3.86563964676908e-05,
+      "loss": 0.4213,
+      "step": 7600
+    },
+    {
+      "epoch": 0.5766186779226653,
+      "grad_norm": 0.16220742464065552,
+      "learning_rate": 3.808064771949893e-05,
+      "loss": 0.4208,
+      "step": 7650
+    },
+    {
+      "epoch": 0.5803874274515716,
+      "grad_norm": 0.16180914640426636,
+      "learning_rate": 3.75065698382932e-05,
+      "loss": 0.4213,
+      "step": 7700
+    },
+    {
+      "epoch": 0.5841561769804778,
+      "grad_norm": 0.17086252570152283,
+      "learning_rate": 3.693424329889776e-05,
+      "loss": 0.4209,
+      "step": 7750
+    },
+    {
+      "epoch": 0.5879249265093842,
+      "grad_norm": 0.1601138710975647,
+      "learning_rate": 3.636374833063191e-05,
+      "loss": 0.4206,
+      "step": 7800
+    },
+    {
+      "epoch": 0.5916936760382905,
+      "grad_norm": 0.15639857947826385,
+      "learning_rate": 3.579516490606346e-05,
+      "loss": 0.4191,
+      "step": 7850
+    },
+    {
+      "epoch": 0.5954624255671968,
+      "grad_norm": 0.16204357147216797,
+      "learning_rate": 3.522857272979804e-05,
+      "loss": 0.4185,
+      "step": 7900
+    },
+    {
+      "epoch": 0.5992311750961031,
+      "grad_norm": 0.1710115373134613,
+      "learning_rate": 3.4664051227306026e-05,
+      "loss": 0.4178,
+      "step": 7950
+    },
+    {
+      "epoch": 0.6029999246250094,
+      "grad_norm": 0.15990346670150757,
+      "learning_rate": 3.4101679533788734e-05,
+      "loss": 0.4161,
+      "step": 8000
+    },
+    {
+      "epoch": 0.6067686741539158,
+      "grad_norm": 0.15846975147724152,
+      "learning_rate": 3.354153648308492e-05,
+      "loss": 0.4168,
+      "step": 8050
+    },
+    {
+      "epoch": 0.610537423682822,
+      "grad_norm": 0.15800924599170685,
+      "learning_rate": 3.298370059662004e-05,
+      "loss": 0.4165,
+      "step": 8100
+    },
+    {
+      "epoch": 0.6143061732117283,
+      "grad_norm": 0.1673530787229538,
+      "learning_rate": 3.2428250072398846e-05,
+      "loss": 0.4164,
+      "step": 8150
+    },
+    {
+      "epoch": 0.6180749227406347,
+      "grad_norm": 0.16620007157325745,
+      "learning_rate": 3.187526277404355e-05,
+      "loss": 0.4193,
+      "step": 8200
+    },
+    {
+      "epoch": 0.621843672269541,
+      "grad_norm": 0.1582447588443756,
+      "learning_rate": 3.1324816219878903e-05,
+      "loss": 0.416,
+      "step": 8250
+    },
+    {
+      "epoch": 0.6256124217984472,
+      "grad_norm": 0.16663286089897156,
+      "learning_rate": 3.077698757206552e-05,
+      "loss": 0.4172,
+      "step": 8300
+    },
+    {
+      "epoch": 0.6293811713273536,
+      "grad_norm": 0.16307072341442108,
+      "learning_rate": 3.0231853625783163e-05,
+      "loss": 0.4145,
+      "step": 8350
+    },
+    {
+      "epoch": 0.6331499208562599,
+      "grad_norm": 0.1569572240114212,
+      "learning_rate": 2.9689490798465698e-05,
+      "loss": 0.4146,
+      "step": 8400
+    },
+    {
+      "epoch": 0.6369186703851663,
+      "grad_norm": 0.1609562188386917,
+      "learning_rate": 2.9149975119088596e-05,
+      "loss": 0.4146,
+      "step": 8450
+    },
+    {
+      "epoch": 0.6406874199140725,
+      "grad_norm": 0.15904489159584045,
+      "learning_rate": 2.8613382217511265e-05,
+      "loss": 0.4125,
+      "step": 8500
+    },
+    {
+      "epoch": 0.6444561694429788,
+      "grad_norm": 0.16079629957675934,
+      "learning_rate": 2.807978731387516e-05,
+      "loss": 0.4151,
+      "step": 8550
+    },
+    {
+      "epoch": 0.6482249189718852,
+      "grad_norm": 0.16790254414081573,
+      "learning_rate": 2.754926520805925e-05,
+      "loss": 0.4141,
+      "step": 8600
+    },
+    {
+      "epoch": 0.6519936685007914,
+      "grad_norm": 0.16292473673820496,
+      "learning_rate": 2.702189026919465e-05,
+      "loss": 0.4143,
+      "step": 8650
+    },
+    {
+      "epoch": 0.6557624180296977,
+      "grad_norm": 0.16475141048431396,
+      "learning_rate": 2.6497736425239315e-05,
+      "loss": 0.4129,
+      "step": 8700
+    },
+    {
+      "epoch": 0.6595311675586041,
+      "grad_norm": 0.1627720594406128,
+      "learning_rate": 2.597687715261484e-05,
+      "loss": 0.4127,
+      "step": 8750
+    },
+    {
+      "epoch": 0.6632999170875103,
+      "grad_norm": 0.1639036387205124,
+      "learning_rate": 2.5459385465906517e-05,
+      "loss": 0.4121,
+      "step": 8800
+    },
+    {
+      "epoch": 0.6670686666164167,
+      "grad_norm": 0.16833463311195374,
+      "learning_rate": 2.4945333907627892e-05,
+      "loss": 0.4129,
+      "step": 8850
+    },
+    {
+      "epoch": 0.670837416145323,
+      "grad_norm": 0.15790612995624542,
+      "learning_rate": 2.443479453805189e-05,
+      "loss": 0.4098,
+      "step": 8900
+    },
+    {
+      "epoch": 0.6746061656742293,
+      "grad_norm": 0.16387607157230377,
+      "learning_rate": 2.392783892510917e-05,
+      "loss": 0.411,
+      "step": 8950
+    },
+    {
+      "epoch": 0.6783749152031356,
+      "grad_norm": 0.16562491655349731,
+      "learning_rate": 2.3424538134355715e-05,
+      "loss": 0.4122,
+      "step": 9000
+    },
+    {
+      "epoch": 0.6821436647320419,
+      "grad_norm": 0.15516149997711182,
+      "learning_rate": 2.2924962719010874e-05,
+      "loss": 0.4112,
+      "step": 9050
+    },
+    {
+      "epoch": 0.6859124142609482,
+      "grad_norm": 0.16322891414165497,
+      "learning_rate": 2.242918271006698e-05,
+      "loss": 0.4109,
+      "step": 9100
+    },
+    {
+      "epoch": 0.6896811637898546,
+      "grad_norm": 0.16053235530853271,
+      "learning_rate": 2.193726760647245e-05,
+      "loss": 0.4088,
+      "step": 9150
+    },
+    {
+      "epoch": 0.6934499133187608,
+      "grad_norm": 0.1649434119462967,
+      "learning_rate": 2.1449286365389342e-05,
+      "loss": 0.4103,
+      "step": 9200
+    },
+    {
+      "epoch": 0.6972186628476671,
+      "grad_norm": 0.1640276312828064,
+      "learning_rate": 2.0965307392526818e-05,
+      "loss": 0.409,
+      "step": 9250
+    },
+    {
+      "epoch": 0.7009874123765735,
+      "grad_norm": 0.1616964489221573,
+      "learning_rate": 2.048539853255197e-05,
+      "loss": 0.4105,
+      "step": 9300
+    },
+    {
+      "epoch": 0.7047561619054797,
+      "grad_norm": 0.1595754474401474,
+      "learning_rate": 2.0009627059579372e-05,
+      "loss": 0.4108,
+      "step": 9350
+    },
+    {
+      "epoch": 0.7085249114343861,
+      "grad_norm": 0.1613272726535797,
+      "learning_rate": 1.953805966774037e-05,
+      "loss": 0.4114,
+      "step": 9400
+    },
+    {
+      "epoch": 0.7122936609632924,
+      "grad_norm": 0.15698370337486267,
+      "learning_rate": 1.9070762461834018e-05,
+      "loss": 0.4087,
+      "step": 9450
+    },
+    {
+      "epoch": 0.7160624104921987,
+      "grad_norm": 0.17342697083950043,
+      "learning_rate": 1.8607800948060266e-05,
+      "loss": 0.4072,
+      "step": 9500
+    },
+    {
+      "epoch": 0.719831160021105,
+      "grad_norm": 0.16098545491695404,
+      "learning_rate": 1.8149240024837315e-05,
+      "loss": 0.4071,
+      "step": 9550
+    },
+    {
+      "epoch": 0.7235999095500113,
+      "grad_norm": 0.16427302360534668,
+      "learning_rate": 1.7695143973704143e-05,
+      "loss": 0.4067,
+      "step": 9600
+    },
+    {
+      "epoch": 0.7273686590789176,
+      "grad_norm": 0.1618986576795578,
+      "learning_rate": 1.7245576450309316e-05,
+      "loss": 0.4081,
+      "step": 9650
+    },
+    {
+      "epoch": 0.731137408607824,
+      "grad_norm": 0.15955495834350586,
+      "learning_rate": 1.6800600475487826e-05,
+      "loss": 0.4085,
+      "step": 9700
+    },
+    {
+      "epoch": 0.7349061581367302,
+      "grad_norm": 0.159365713596344,
+      "learning_rate": 1.6360278426426624e-05,
+      "loss": 0.4069,
+      "step": 9750
+    },
+    {
+      "epoch": 0.7386749076656366,
+      "grad_norm": 0.1574493795633316,
+      "learning_rate": 1.5924672027920663e-05,
+      "loss": 0.4058,
+      "step": 9800
+    },
+    {
+      "epoch": 0.7424436571945429,
+      "grad_norm": 0.16413567960262299,
+      "learning_rate": 1.5493842343720104e-05,
+      "loss": 0.4047,
+      "step": 9850
+    },
+    {
+      "epoch": 0.7462124067234491,
+      "grad_norm": 0.1662568747997284,
+      "learning_rate": 1.5067849767970488e-05,
+      "loss": 0.4046,
+      "step": 9900
+    },
+    {
+      "epoch": 0.7499811562523555,
+      "grad_norm": 0.16023589670658112,
+      "learning_rate": 1.4646754016746483e-05,
+      "loss": 0.4072,
+      "step": 9950
+    },
+    {
+      "epoch": 0.7537499057812618,
+      "grad_norm": 0.16238714754581451,
+      "learning_rate": 1.4230614119680957e-05,
+      "loss": 0.4072,
+      "step": 10000
+    },
+    {
+      "epoch": 0.757518655310168,
+      "grad_norm": 0.16267696022987366,
+      "learning_rate": 1.3819488411690018e-05,
+      "loss": 0.4056,
+      "step": 10050
+    },
+    {
+      "epoch": 0.7612874048390744,
+      "grad_norm": 0.1596570760011673,
+      "learning_rate": 1.3413434524795631e-05,
+      "loss": 0.4049,
+      "step": 10100
+    },
+    {
+      "epoch": 0.7650561543679807,
+      "grad_norm": 0.17054887115955353,
+      "learning_rate": 1.3012509380046745e-05,
+      "loss": 0.4032,
+      "step": 10150
+    },
+    {
+      "epoch": 0.768824903896887,
+      "grad_norm": 0.15980112552642822,
+      "learning_rate": 1.2616769179539944e-05,
+      "loss": 0.405,
+      "step": 10200
+    },
+    {
+      "epoch": 0.7725936534257933,
+      "grad_norm": 0.16071230173110962,
+      "learning_rate": 1.222626939854103e-05,
+      "loss": 0.4027,
+      "step": 10250
+    },
+    {
+      "epoch": 0.7763624029546996,
+      "grad_norm": 0.15903015434741974,
+      "learning_rate": 1.1841064777708483e-05,
+      "loss": 0.4043,
+      "step": 10300
+    },
+    {
+      "epoch": 0.780131152483606,
+      "grad_norm": 0.16393496096134186,
+      "learning_rate": 1.1461209315419758e-05,
+      "loss": 0.4009,
+      "step": 10350
+    },
+    {
+      "epoch": 0.7838999020125123,
+      "grad_norm": 0.16215142607688904,
+      "learning_rate": 1.1086756260201859e-05,
+      "loss": 0.4032,
+      "step": 10400
+    },
+    {
+      "epoch": 0.7876686515414185,
+      "grad_norm": 0.16436554491519928,
+      "learning_rate": 1.0717758103266805e-05,
+      "loss": 0.4035,
+      "step": 10450
+    },
+    {
+      "epoch": 0.7914374010703249,
+      "grad_norm": 0.16526173055171967,
+      "learning_rate": 1.0354266571153399e-05,
+      "loss": 0.4023,
+      "step": 10500
+    },
+    {
+      "epoch": 0.7952061505992312,
+      "grad_norm": 0.1610834300518036,
+      "learning_rate": 9.996332618476172e-06,
+      "loss": 0.4031,
+      "step": 10550
+    },
+    {
+      "epoch": 0.7989749001281374,
+      "grad_norm": 0.1577410101890564,
+      "learning_rate": 9.644006420782476e-06,
+      "loss": 0.4006,
+      "step": 10600
+    },
+    {
+      "epoch": 0.8027436496570438,
+      "grad_norm": 0.15618577599525452,
+      "learning_rate": 9.29733736751881e-06,
+      "loss": 0.4037,
+      "step": 10650
+    },
+    {
+      "epoch": 0.8065123991859501,
+      "grad_norm": 0.15943607687950134,
+      "learning_rate": 8.956374055107442e-06,
+      "loss": 0.4026,
+      "step": 10700
+    },
+    {
+      "epoch": 0.8102811487148565,
+      "grad_norm": 0.1561410278081894,
+      "learning_rate": 8.621164280134004e-06,
+      "loss": 0.4021,
+      "step": 10750
+    },
+    {
+      "epoch": 0.8140498982437627,
+      "grad_norm": 0.166486918926239,
+      "learning_rate": 8.291755032647402e-06,
+      "loss": 0.4017,
+      "step": 10800
+    },
+    {
+      "epoch": 0.817818647772669,
+      "grad_norm": 0.16270950436592102,
+      "learning_rate": 7.96819248957265e-06,
+      "loss": 0.4019,
+      "step": 10850
+    },
+    {
+      "epoch": 0.8215873973015754,
+      "grad_norm": 0.1590346395969391,
+      "learning_rate": 7.650522008237754e-06,
+      "loss": 0.4014,
+      "step": 10900
+    },
+    {
+      "epoch": 0.8253561468304816,
+      "grad_norm": 0.15762105584144592,
+      "learning_rate": 7.338788120015522e-06,
+      "loss": 0.4005,
+      "step": 10950
+    },
+    {
+      "epoch": 0.8291248963593879,
+      "grad_norm": 0.1578063815832138,
+      "learning_rate": 7.033034524081023e-06,
+      "loss": 0.4008,
+      "step": 11000
+    },
+    {
+      "epoch": 0.8328936458882943,
+      "grad_norm": 0.1597341001033783,
+      "learning_rate": 6.733304081285874e-06,
+      "loss": 0.4005,
+      "step": 11050
+    },
+    {
+      "epoch": 0.8366623954172006,
+      "grad_norm": 0.16171535849571228,
+      "learning_rate": 6.439638808149923e-06,
+      "loss": 0.4018,
+      "step": 11100
+    },
+    {
+      "epoch": 0.8404311449461069,
+      "grad_norm": 0.16193453967571259,
+      "learning_rate": 6.152079870971311e-06,
+      "loss": 0.3993,
+      "step": 11150
+    },
+    {
+      "epoch": 0.8441998944750132,
+      "grad_norm": 0.15683096647262573,
+      "learning_rate": 5.870667580055805e-06,
+      "loss": 0.4014,
+      "step": 11200
+    },
+    {
+      "epoch": 0.8479686440039195,
+      "grad_norm": 0.16238045692443848,
+      "learning_rate": 5.595441384065986e-06,
+      "loss": 0.402,
+      "step": 11250
+    },
+    {
+      "epoch": 0.8517373935328258,
+      "grad_norm": 0.15774257481098175,
+      "learning_rate": 5.3264398644913114e-06,
+      "loss": 0.4,
+      "step": 11300
+    },
+    {
+      "epoch": 0.8555061430617321,
+      "grad_norm": 0.15233269333839417,
+      "learning_rate": 5.063700730239784e-06,
+      "loss": 0.3994,
+      "step": 11350
+    },
+    {
+      "epoch": 0.8592748925906384,
+      "grad_norm": 0.16699600219726562,
+      "learning_rate": 4.807260812351793e-06,
+      "loss": 0.399,
+      "step": 11400
+    },
+    {
+      "epoch": 0.8630436421195448,
+      "grad_norm": 0.1617075353860855,
+      "learning_rate": 4.557156058837137e-06,
+      "loss": 0.3988,
+      "step": 11450
+    },
+    {
+      "epoch": 0.866812391648451,
+      "grad_norm": 0.1602969914674759,
+      "learning_rate": 4.31342152963583e-06,
+      "loss": 0.3999,
+      "step": 11500
+    },
+    {
+      "epoch": 0.8705811411773573,
+      "grad_norm": 0.15863758325576782,
+      "learning_rate": 4.076091391703302e-06,
+      "loss": 0.3999,
+      "step": 11550
+    },
+    {
+      "epoch": 0.8743498907062637,
+      "grad_norm": 0.1571153998374939,
+      "learning_rate": 3.845198914220871e-06,
+      "loss": 0.3984,
+      "step": 11600
+    },
+    {
+      "epoch": 0.87811864023517,
+      "grad_norm": 0.15634584426879883,
+      "learning_rate": 3.6207764639320462e-06,
+      "loss": 0.3989,
+      "step": 11650
+    },
+    {
+      "epoch": 0.8818873897640763,
+      "grad_norm": 0.16058678925037384,
+      "learning_rate": 3.4028555006052953e-06,
+      "loss": 0.4015,
+      "step": 11700
+    },
+    {
+      "epoch": 0.8856561392929826,
+      "grad_norm": 0.15872234106063843,
+      "learning_rate": 3.191466572624019e-06,
+      "loss": 0.3979,
+      "step": 11750
+    },
+    {
+      "epoch": 0.8894248888218889,
+      "grad_norm": 0.16169828176498413,
+      "learning_rate": 2.986639312704209e-06,
+      "loss": 0.3984,
+      "step": 11800
+    },
+    {
+      "epoch": 0.8931936383507952,
+      "grad_norm": 0.16279493272304535,
+      "learning_rate": 2.788402433740517e-06,
+      "loss": 0.3982,
+      "step": 11850
+    },
+    {
+      "epoch": 0.8969623878797015,
+      "grad_norm": 0.1638520359992981,
+      "learning_rate": 2.596783724781282e-06,
+      "loss": 0.4002,
+      "step": 11900
+    },
+    {
+      "epoch": 0.9007311374086078,
+      "grad_norm": 0.15517061948776245,
+      "learning_rate": 2.4118100471329787e-06,
+      "loss": 0.3974,
+      "step": 11950
+    },
+    {
+      "epoch": 0.9044998869375142,
+      "grad_norm": 0.1604815125465393,
+      "learning_rate": 2.2335073305948086e-06,
+      "loss": 0.3992,
+      "step": 12000
+    },
+    {
+      "epoch": 0.9082686364664204,
+      "grad_norm": 0.157390296459198,
+      "learning_rate": 2.0619005698238437e-06,
+      "loss": 0.3989,
+      "step": 12050
+    },
+    {
+      "epoch": 0.9120373859953268,
+      "grad_norm": 0.15608523786067963,
+      "learning_rate": 1.8970138208311949e-06,
+      "loss": 0.3971,
+      "step": 12100
+    },
+    {
+      "epoch": 0.9158061355242331,
+      "grad_norm": 0.15673068165779114,
+      "learning_rate": 1.7388701976099041e-06,
+      "loss": 0.3994,
+      "step": 12150
+    },
+    {
+      "epoch": 0.9195748850531393,
+      "grad_norm": 0.1587488353252411,
+      "learning_rate": 1.5874918688946972e-06,
+      "loss": 0.3985,
+      "step": 12200
+    },
+    {
+      "epoch": 0.9233436345820457,
+      "grad_norm": 0.16035687923431396,
+      "learning_rate": 1.4429000550544414e-06,
+      "loss": 0.399,
+      "step": 12250
+    },
+    {
+      "epoch": 0.927112384110952,
+      "grad_norm": 0.15816493332386017,
+      "learning_rate": 1.305115025117387e-06,
+      "loss": 0.4,
+      "step": 12300
+    },
+    {
+      "epoch": 0.9308811336398582,
+      "grad_norm": 0.16530562937259674,
+      "learning_rate": 1.1741560939298791e-06,
+      "loss": 0.3995,
+      "step": 12350
+    },
+    {
+      "epoch": 0.9346498831687646,
+      "grad_norm": 0.1594778597354889,
+      "learning_rate": 1.0500416194487384e-06,
+      "loss": 0.3997,
+      "step": 12400
+    },
+    {
+      "epoch": 0.9384186326976709,
+      "grad_norm": 0.15754447877407074,
+      "learning_rate": 9.327890001678719e-07,
+      "loss": 0.3972,
+      "step": 12450
+    },
+    {
+      "epoch": 0.9421873822265773,
+      "grad_norm": 0.15905898809432983,
+      "learning_rate": 8.224146726792947e-07,
+      "loss": 0.3972,
+      "step": 12500
+    },
+    {
+      "epoch": 0.9459561317554835,
+      "grad_norm": 0.15947633981704712,
+      "learning_rate": 7.189341093690627e-07,
+      "loss": 0.3964,
+      "step": 12550
+    },
+    {
+      "epoch": 0.9497248812843898,
+      "grad_norm": 0.16022710502147675,
+      "learning_rate": 6.223618162483014e-07,
+      "loss": 0.3993,
+      "step": 12600
+    },
+    {
+      "epoch": 0.9534936308132962,
+      "grad_norm": 0.16380661725997925,
+      "learning_rate": 5.327113309197828e-07,
+      "loss": 0.4,
+      "step": 12650
+    },
+    {
+      "epoch": 0.9572623803422025,
+      "grad_norm": 0.15692880749702454,
+      "learning_rate": 4.4999522068017164e-07,
+      "loss": 0.3982,
+      "step": 12700
+    },
+    {
+      "epoch": 0.9610311298711087,
+      "grad_norm": 0.16599752008914948,
+      "learning_rate": 3.7422508075835583e-07,
+      "loss": 0.397,
+      "step": 12750
+    },
+    {
+      "epoch": 0.9647998794000151,
+      "grad_norm": 0.15927733480930328,
+      "learning_rate": 3.05411532689992e-07,
+      "loss": 0.396,
+      "step": 12800
+    },
+    {
+      "epoch": 0.9685686289289214,
+      "grad_norm": 0.15940117835998535,
+      "learning_rate": 2.435642228285906e-07,
+      "loss": 0.3983,
+      "step": 12850
+    },
+    {
+      "epoch": 0.9723373784578276,
+      "grad_norm": 0.16387763619422913,
+      "learning_rate": 1.886918209932642e-07,
+      "loss": 0.398,
+      "step": 12900
+    },
+    {
+      "epoch": 0.976106127986734,
+      "grad_norm": 0.15747365355491638,
+      "learning_rate": 1.4080201925338322e-07,
+      "loss": 0.3978,
+      "step": 12950
+    },
+    {
+      "epoch": 0.9798748775156403,
+      "grad_norm": 0.1579546183347702,
+      "learning_rate": 9.99015308503215e-08,
+      "loss": 0.4005,
+      "step": 13000
+    },
+    {
+      "epoch": 0.9836436270445467,
+      "grad_norm": 0.15818338096141815,
+      "learning_rate": 6.599608925633715e-08,
+      "loss": 0.3978,
+      "step": 13050
+    },
+    {
+      "epoch": 0.9874123765734529,
+      "grad_norm": 0.15863798558712006,
+      "learning_rate": 3.909044737089307e-08,
+      "loss": 0.3991,
+      "step": 13100
+    },
+    {
+      "epoch": 0.9911811261023592,
+      "grad_norm": 0.1556527018547058,
+      "learning_rate": 1.9188376854373246e-08,
+      "loss": 0.3985,
+      "step": 13150
+    },
+    {
+      "epoch": 0.9949498756312656,
+      "grad_norm": 0.16225971281528473,
+      "learning_rate": 6.292667599366864e-09,
+      "loss": 0.3979,
+      "step": 13200
+    },
+    {
+      "epoch": 0.9987186251601718,
+      "grad_norm": 0.16236484050750732,
+      "learning_rate": 4.0512733956998837e-10,
+      "loss": 0.3953,
+      "step": 13250
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 13267,
+  "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": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6.892725356142474e+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 sha256:0085c343f2206a794f40a3e085ec366c9dc3ce02d1c86bbca36834ab3eb500ea
+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)