Upload folder using huggingface_hub

.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 +1,28 @@
+{
+  "</think>": 151668,
+  "</tool_call>": 151658,
+  "</tool_response>": 151666,
+  "<think>": 151667,
+  "<tool_call>": 151657,
+  "<tool_response>": 151665,
+  "<|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
+}

chat_template.jinja

@@ -0,0 +1,85 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}
+        {%- set ns.multi_step_tool = false %}
+        {%- set ns.last_query_index = index %}
+    {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set content = message.content %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is defined and message.reasoning_content is not none %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in message.content %}
+                {%- set content = message.content.split('</think>')[-1].lstrip('\n') %}
+                {%- set reasoning_content = message.content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {%- if loop.last or (not loop.last and reasoning_content) %}
+                {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
+            {%- else %}
+                {{- '<|im_start|>' + message.role + '\n' + content }}
+            {%- endif %}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}

config.json

@@ -0,0 +1,38 @@
+{
+  "architectures": [
+    "Qwen3MoeForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "decoder_sparse_step": 1,
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 2048,
+  "initializer_range": 0.02,
+  "intermediate_size": 6144,
+  "max_position_embeddings": 32768,
+  "max_window_layers": 48,
+  "mlp_only_layers": [],
+  "model_type": "qwen3_moe",
+  "moe_intermediate_size": 768,
+  "norm_topk_prob": true,
+  "num_attention_heads": 32,
+  "num_experts": 128,
+  "num_experts_per_tok": 8,
+  "num_hidden_layers": 48,
+  "num_key_value_heads": 4,
+  "output_router_logits": false,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000.0,
+  "router_aux_loss_coef": 0.001,
+  "sliding_window": null,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.52.0.dev0",
+  "use_cache": 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": 16384,
+  "transformers_version": "4.52.0.dev0"
+}

latest

@@ -0,0 +1 @@
+global_step157

model-00001-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5853cdc156356dd32339a7ec00fc8611a09863cc00fd878fef9aeac8a2b211ea
+size 4997184968

model-00002-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3e42a8999e80571ee58ea2418fbffa77fcde3368e4f7da5899113a2470d4b571
+size 4997741608

model-00003-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c743f6d68c1732ce568b702e99e04937b4fea9fac8c62dd34795353acebbe6b7
+size 4997742208

model-00004-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:109536e25a0f910c734b0877e40b1417210212cf06dddcbdb443cd1b0f3c5ecd
+size 4997743184

model-00005-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:816e9d12ab55bf6de0bd103e86777cabb0c3f95595595914addc9c80466816db
+size 4997743184

model-00006-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:127057c4733ccbe67901fc07f983bd2afa78904311a9a7a578a4bf2919119808
+size 4997743184

model-00007-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a47ba934c40b8bf938bbceef71d3373b0f68f3c81191613cac22a5af7335c3f8
+size 4997743184

model-00008-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ed41ebf29d81979e10fd2363eca660c97b04ab6e90a5be25530b1307112ff7a1
+size 4997743184

model-00009-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ba3ee0a9f2805e84f9298ccb662f9e68aae3b21cbc6c48e664b60db2a2251735
+size 4997743184

model-00010-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:33993d17fec74e240cef1246e46a85cc67078d630c7ac2d1d12bb59bc89d518b
+size 4997743184

model-00011-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:07ce2e07f1831a34f6c579b16b30afb8c9ab99f5643929aa4d2b97600a0c3df9
+size 4997743184

model-00012-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:88a9c49e20b0277b5c7a3a9df8a0b847674ece3564c6f8c4e23758499c2adf45
+size 4997743184

model-00013-of-00013.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6d9fafed28e4df542242d6daa582f0f06ecddffc4e4ca10f6ccf3ac6b1118957
+size 1094220288

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:aeb13307a71acd8fe81861d94ad54ab689df773318809eed3cbe794b4492dae4
+size 11422654

tokenizer_config.json

@@ -0,0 +1,240 @@
+{
+  "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
+    },
+    "151665": {
+      "content": "<tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151666": {
+      "content": "</tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151667": {
+      "content": "<think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151668": {
+      "content": "</think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "padding_side": "right",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

trainer_state.json

@@ -0,0 +1,1133 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.20096,
+  "eval_steps": 500,
+  "global_step": 157,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00128,
+      "grad_norm": 2.8870697066158506,
+      "learning_rate": 0.0,
+      "loss": 0.8422,
+      "step": 1
+    },
+    {
+      "epoch": 0.00256,
+      "grad_norm": 2.88484389829891,
+      "learning_rate": 6.329113924050633e-07,
+      "loss": 0.8541,
+      "step": 2
+    },
+    {
+      "epoch": 0.00384,
+      "grad_norm": 2.858151965789657,
+      "learning_rate": 1.2658227848101265e-06,
+      "loss": 0.8376,
+      "step": 3
+    },
+    {
+      "epoch": 0.00512,
+      "grad_norm": 2.759628117182127,
+      "learning_rate": 1.8987341772151901e-06,
+      "loss": 0.8334,
+      "step": 4
+    },
+    {
+      "epoch": 0.0064,
+      "grad_norm": 2.796990062811218,
+      "learning_rate": 2.531645569620253e-06,
+      "loss": 0.8256,
+      "step": 5
+    },
+    {
+      "epoch": 0.00768,
+      "grad_norm": 2.5779298795445023,
+      "learning_rate": 3.1645569620253167e-06,
+      "loss": 0.8301,
+      "step": 6
+    },
+    {
+      "epoch": 0.00896,
+      "grad_norm": 2.182261607936066,
+      "learning_rate": 3.7974683544303802e-06,
+      "loss": 0.8156,
+      "step": 7
+    },
+    {
+      "epoch": 0.01024,
+      "grad_norm": 1.9615896152651355,
+      "learning_rate": 4.430379746835443e-06,
+      "loss": 0.7982,
+      "step": 8
+    },
+    {
+      "epoch": 0.01152,
+      "grad_norm": 1.452541644948315,
+      "learning_rate": 5.063291139240506e-06,
+      "loss": 0.7819,
+      "step": 9
+    },
+    {
+      "epoch": 0.0128,
+      "grad_norm": 1.4723286808630864,
+      "learning_rate": 5.69620253164557e-06,
+      "loss": 0.7906,
+      "step": 10
+    },
+    {
+      "epoch": 0.01408,
+      "grad_norm": 1.3529636617858944,
+      "learning_rate": 6.329113924050633e-06,
+      "loss": 0.7724,
+      "step": 11
+    },
+    {
+      "epoch": 0.01536,
+      "grad_norm": 1.960737179905222,
+      "learning_rate": 6.9620253164556965e-06,
+      "loss": 0.7495,
+      "step": 12
+    },
+    {
+      "epoch": 0.01664,
+      "grad_norm": 2.2349101055406337,
+      "learning_rate": 7.5949367088607605e-06,
+      "loss": 0.7581,
+      "step": 13
+    },
+    {
+      "epoch": 0.01792,
+      "grad_norm": 2.0897577150322477,
+      "learning_rate": 8.227848101265822e-06,
+      "loss": 0.7404,
+      "step": 14
+    },
+    {
+      "epoch": 0.0192,
+      "grad_norm": 1.8227218322635887,
+      "learning_rate": 8.860759493670886e-06,
+      "loss": 0.7382,
+      "step": 15
+    },
+    {
+      "epoch": 0.02048,
+      "grad_norm": 1.2099951464458898,
+      "learning_rate": 9.49367088607595e-06,
+      "loss": 0.7231,
+      "step": 16
+    },
+    {
+      "epoch": 0.02176,
+      "grad_norm": 1.2177037129914572,
+      "learning_rate": 1.0126582278481012e-05,
+      "loss": 0.7259,
+      "step": 17
+    },
+    {
+      "epoch": 0.02304,
+      "grad_norm": 1.1031346132830708,
+      "learning_rate": 1.0759493670886076e-05,
+      "loss": 0.7059,
+      "step": 18
+    },
+    {
+      "epoch": 0.02432,
+      "grad_norm": 0.9194779600801882,
+      "learning_rate": 1.139240506329114e-05,
+      "loss": 0.7137,
+      "step": 19
+    },
+    {
+      "epoch": 0.0256,
+      "grad_norm": 0.8679468005972053,
+      "learning_rate": 1.2025316455696203e-05,
+      "loss": 0.7036,
+      "step": 20
+    },
+    {
+      "epoch": 0.02688,
+      "grad_norm": 0.7227287276969042,
+      "learning_rate": 1.2658227848101267e-05,
+      "loss": 0.696,
+      "step": 21
+    },
+    {
+      "epoch": 0.02816,
+      "grad_norm": 0.7425882516811844,
+      "learning_rate": 1.3291139240506329e-05,
+      "loss": 0.6888,
+      "step": 22
+    },
+    {
+      "epoch": 0.02944,
+      "grad_norm": 0.7093793012252196,
+      "learning_rate": 1.3924050632911393e-05,
+      "loss": 0.6791,
+      "step": 23
+    },
+    {
+      "epoch": 0.03072,
+      "grad_norm": 0.6018215463147907,
+      "learning_rate": 1.4556962025316457e-05,
+      "loss": 0.6783,
+      "step": 24
+    },
+    {
+      "epoch": 0.032,
+      "grad_norm": 0.5846346732378257,
+      "learning_rate": 1.5189873417721521e-05,
+      "loss": 0.6811,
+      "step": 25
+    },
+    {
+      "epoch": 0.03328,
+      "grad_norm": 0.5855419788452784,
+      "learning_rate": 1.5822784810126583e-05,
+      "loss": 0.683,
+      "step": 26
+    },
+    {
+      "epoch": 0.03456,
+      "grad_norm": 0.5096689891724868,
+      "learning_rate": 1.6455696202531644e-05,
+      "loss": 0.6589,
+      "step": 27
+    },
+    {
+      "epoch": 0.03584,
+      "grad_norm": 0.4871170504081146,
+      "learning_rate": 1.7088607594936708e-05,
+      "loss": 0.6582,
+      "step": 28
+    },
+    {
+      "epoch": 0.03712,
+      "grad_norm": 0.4949600697144217,
+      "learning_rate": 1.7721518987341772e-05,
+      "loss": 0.669,
+      "step": 29
+    },
+    {
+      "epoch": 0.0384,
+      "grad_norm": 0.5082926031630941,
+      "learning_rate": 1.8354430379746836e-05,
+      "loss": 0.666,
+      "step": 30
+    },
+    {
+      "epoch": 0.03968,
+      "grad_norm": 0.49381475380567175,
+      "learning_rate": 1.89873417721519e-05,
+      "loss": 0.6556,
+      "step": 31
+    },
+    {
+      "epoch": 0.04096,
+      "grad_norm": 0.4265624784331274,
+      "learning_rate": 1.962025316455696e-05,
+      "loss": 0.646,
+      "step": 32
+    },
+    {
+      "epoch": 0.04224,
+      "grad_norm": 0.39190416547723717,
+      "learning_rate": 2.0253164556962025e-05,
+      "loss": 0.6473,
+      "step": 33
+    },
+    {
+      "epoch": 0.04352,
+      "grad_norm": 0.4631353399929371,
+      "learning_rate": 2.088607594936709e-05,
+      "loss": 0.6441,
+      "step": 34
+    },
+    {
+      "epoch": 0.0448,
+      "grad_norm": 0.3928335126997034,
+      "learning_rate": 2.1518987341772153e-05,
+      "loss": 0.6352,
+      "step": 35
+    },
+    {
+      "epoch": 0.04608,
+      "grad_norm": 0.36295027582313966,
+      "learning_rate": 2.2151898734177217e-05,
+      "loss": 0.6333,
+      "step": 36
+    },
+    {
+      "epoch": 0.04736,
+      "grad_norm": 0.35026852064181846,
+      "learning_rate": 2.278481012658228e-05,
+      "loss": 0.6399,
+      "step": 37
+    },
+    {
+      "epoch": 0.04864,
+      "grad_norm": 0.39778614916835536,
+      "learning_rate": 2.341772151898734e-05,
+      "loss": 0.6298,
+      "step": 38
+    },
+    {
+      "epoch": 0.04992,
+      "grad_norm": 0.33278348666417684,
+      "learning_rate": 2.4050632911392405e-05,
+      "loss": 0.6301,
+      "step": 39
+    },
+    {
+      "epoch": 0.0512,
+      "grad_norm": 0.31444068712551376,
+      "learning_rate": 2.468354430379747e-05,
+      "loss": 0.6263,
+      "step": 40
+    },
+    {
+      "epoch": 0.05248,
+      "grad_norm": 0.36059728676958264,
+      "learning_rate": 2.5316455696202533e-05,
+      "loss": 0.6458,
+      "step": 41
+    },
+    {
+      "epoch": 0.05376,
+      "grad_norm": 0.3916144552301749,
+      "learning_rate": 2.5949367088607597e-05,
+      "loss": 0.6331,
+      "step": 42
+    },
+    {
+      "epoch": 0.05504,
+      "grad_norm": 0.32338566356420756,
+      "learning_rate": 2.6582278481012658e-05,
+      "loss": 0.6332,
+      "step": 43
+    },
+    {
+      "epoch": 0.05632,
+      "grad_norm": 0.33704233729853356,
+      "learning_rate": 2.7215189873417722e-05,
+      "loss": 0.6348,
+      "step": 44
+    },
+    {
+      "epoch": 0.0576,
+      "grad_norm": 0.36015399213900634,
+      "learning_rate": 2.7848101265822786e-05,
+      "loss": 0.6392,
+      "step": 45
+    },
+    {
+      "epoch": 0.05888,
+      "grad_norm": 0.31471331803021757,
+      "learning_rate": 2.848101265822785e-05,
+      "loss": 0.6272,
+      "step": 46
+    },
+    {
+      "epoch": 0.06016,
+      "grad_norm": 0.3225170654156012,
+      "learning_rate": 2.9113924050632914e-05,
+      "loss": 0.635,
+      "step": 47
+    },
+    {
+      "epoch": 0.06144,
+      "grad_norm": 0.3064473735810606,
+      "learning_rate": 2.9746835443037974e-05,
+      "loss": 0.6284,
+      "step": 48
+    },
+    {
+      "epoch": 0.06272,
+      "grad_norm": 0.3038289969291092,
+      "learning_rate": 3.0379746835443042e-05,
+      "loss": 0.6149,
+      "step": 49
+    },
+    {
+      "epoch": 0.064,
+      "grad_norm": 0.3226803690164346,
+      "learning_rate": 3.10126582278481e-05,
+      "loss": 0.626,
+      "step": 50
+    },
+    {
+      "epoch": 0.06528,
+      "grad_norm": 0.3096398144524693,
+      "learning_rate": 3.1645569620253167e-05,
+      "loss": 0.621,
+      "step": 51
+    },
+    {
+      "epoch": 0.06656,
+      "grad_norm": 0.2754757429130796,
+      "learning_rate": 3.227848101265823e-05,
+      "loss": 0.6185,
+      "step": 52
+    },
+    {
+      "epoch": 0.06784,
+      "grad_norm": 0.3262507218160328,
+      "learning_rate": 3.291139240506329e-05,
+      "loss": 0.6171,
+      "step": 53
+    },
+    {
+      "epoch": 0.06912,
+      "grad_norm": 0.34971068352090656,
+      "learning_rate": 3.354430379746836e-05,
+      "loss": 0.616,
+      "step": 54
+    },
+    {
+      "epoch": 0.0704,
+      "grad_norm": 0.2841621281043231,
+      "learning_rate": 3.4177215189873416e-05,
+      "loss": 0.5995,
+      "step": 55
+    },
+    {
+      "epoch": 0.07168,
+      "grad_norm": 0.4003223636484448,
+      "learning_rate": 3.4810126582278487e-05,
+      "loss": 0.6169,
+      "step": 56
+    },
+    {
+      "epoch": 0.07296,
+      "grad_norm": 0.31868860231705426,
+      "learning_rate": 3.5443037974683544e-05,
+      "loss": 0.6077,
+      "step": 57
+    },
+    {
+      "epoch": 0.07424,
+      "grad_norm": 0.3960425782005289,
+      "learning_rate": 3.607594936708861e-05,
+      "loss": 0.6164,
+      "step": 58
+    },
+    {
+      "epoch": 0.07552,
+      "grad_norm": 0.363865574596696,
+      "learning_rate": 3.670886075949367e-05,
+      "loss": 0.6118,
+      "step": 59
+    },
+    {
+      "epoch": 0.0768,
+      "grad_norm": 0.33961478774466697,
+      "learning_rate": 3.7341772151898736e-05,
+      "loss": 0.6137,
+      "step": 60
+    },
+    {
+      "epoch": 0.07808,
+      "grad_norm": 0.4212164741206082,
+      "learning_rate": 3.79746835443038e-05,
+      "loss": 0.6275,
+      "step": 61
+    },
+    {
+      "epoch": 0.07936,
+      "grad_norm": 0.29878729710395663,
+      "learning_rate": 3.8607594936708864e-05,
+      "loss": 0.6084,
+      "step": 62
+    },
+    {
+      "epoch": 0.08064,
+      "grad_norm": 0.36745026817379894,
+      "learning_rate": 3.924050632911392e-05,
+      "loss": 0.607,
+      "step": 63
+    },
+    {
+      "epoch": 0.08192,
+      "grad_norm": 0.38983571508393644,
+      "learning_rate": 3.987341772151899e-05,
+      "loss": 0.6176,
+      "step": 64
+    },
+    {
+      "epoch": 0.0832,
+      "grad_norm": 0.37337392917475115,
+      "learning_rate": 4.050632911392405e-05,
+      "loss": 0.6184,
+      "step": 65
+    },
+    {
+      "epoch": 0.08448,
+      "grad_norm": 0.3668068115925863,
+      "learning_rate": 4.113924050632912e-05,
+      "loss": 0.6194,
+      "step": 66
+    },
+    {
+      "epoch": 0.08576,
+      "grad_norm": 0.36138503055306903,
+      "learning_rate": 4.177215189873418e-05,
+      "loss": 0.6077,
+      "step": 67
+    },
+    {
+      "epoch": 0.08704,
+      "grad_norm": 0.43361127462043814,
+      "learning_rate": 4.240506329113924e-05,
+      "loss": 0.6147,
+      "step": 68
+    },
+    {
+      "epoch": 0.08832,
+      "grad_norm": 0.33520423726109644,
+      "learning_rate": 4.3037974683544305e-05,
+      "loss": 0.6118,
+      "step": 69
+    },
+    {
+      "epoch": 0.0896,
+      "grad_norm": 0.4381154362148859,
+      "learning_rate": 4.367088607594937e-05,
+      "loss": 0.6031,
+      "step": 70
+    },
+    {
+      "epoch": 0.09088,
+      "grad_norm": 0.3717345864324632,
+      "learning_rate": 4.430379746835443e-05,
+      "loss": 0.6031,
+      "step": 71
+    },
+    {
+      "epoch": 0.09216,
+      "grad_norm": 0.4861728465398392,
+      "learning_rate": 4.49367088607595e-05,
+      "loss": 0.6006,
+      "step": 72
+    },
+    {
+      "epoch": 0.09344,
+      "grad_norm": 0.3264992939190504,
+      "learning_rate": 4.556962025316456e-05,
+      "loss": 0.6151,
+      "step": 73
+    },
+    {
+      "epoch": 0.09472,
+      "grad_norm": 0.4319794925001871,
+      "learning_rate": 4.6202531645569625e-05,
+      "loss": 0.6058,
+      "step": 74
+    },
+    {
+      "epoch": 0.096,
+      "grad_norm": 0.4616345840492333,
+      "learning_rate": 4.683544303797468e-05,
+      "loss": 0.5967,
+      "step": 75
+    },
+    {
+      "epoch": 0.09728,
+      "grad_norm": 0.4405721152587957,
+      "learning_rate": 4.7468354430379746e-05,
+      "loss": 0.6002,
+      "step": 76
+    },
+    {
+      "epoch": 0.09856,
+      "grad_norm": 0.5122605377853799,
+      "learning_rate": 4.810126582278481e-05,
+      "loss": 0.6076,
+      "step": 77
+    },
+    {
+      "epoch": 0.09984,
+      "grad_norm": 0.45313870340097556,
+      "learning_rate": 4.8734177215189874e-05,
+      "loss": 0.6074,
+      "step": 78
+    },
+    {
+      "epoch": 0.10112,
+      "grad_norm": 0.4340044755876676,
+      "learning_rate": 4.936708860759494e-05,
+      "loss": 0.606,
+      "step": 79
+    },
+    {
+      "epoch": 0.1024,
+      "grad_norm": 0.4987172862476422,
+      "learning_rate": 5e-05,
+      "loss": 0.6158,
+      "step": 80
+    },
+    {
+      "epoch": 0.10368,
+      "grad_norm": 0.6226880208665108,
+      "learning_rate": 4.999974965737065e-05,
+      "loss": 0.621,
+      "step": 81
+    },
+    {
+      "epoch": 0.10496,
+      "grad_norm": 0.5448293131914782,
+      "learning_rate": 4.999899863449631e-05,
+      "loss": 0.6014,
+      "step": 82
+    },
+    {
+      "epoch": 0.10624,
+      "grad_norm": 0.3427022601926917,
+      "learning_rate": 4.999774694641803e-05,
+      "loss": 0.6198,
+      "step": 83
+    },
+    {
+      "epoch": 0.10752,
+      "grad_norm": 0.5005152113593655,
+      "learning_rate": 4.999599461820387e-05,
+      "loss": 0.6054,
+      "step": 84
+    },
+    {
+      "epoch": 0.1088,
+      "grad_norm": 0.5702968806820528,
+      "learning_rate": 4.999374168494844e-05,
+      "loss": 0.6069,
+      "step": 85
+    },
+    {
+      "epoch": 0.11008,
+      "grad_norm": 0.4671310661706222,
+      "learning_rate": 4.999098819177214e-05,
+      "loss": 0.6017,
+      "step": 86
+    },
+    {
+      "epoch": 0.11136,
+      "grad_norm": 0.46081768174689064,
+      "learning_rate": 4.9987734193820324e-05,
+      "loss": 0.5988,
+      "step": 87
+    },
+    {
+      "epoch": 0.11264,
+      "grad_norm": 0.5448729856183013,
+      "learning_rate": 4.9983979756262136e-05,
+      "loss": 0.6181,
+      "step": 88
+    },
+    {
+      "epoch": 0.11392,
+      "grad_norm": 0.5095775592779056,
+      "learning_rate": 4.9979724954289244e-05,
+      "loss": 0.608,
+      "step": 89
+    },
+    {
+      "epoch": 0.1152,
+      "grad_norm": 0.41119162739543413,
+      "learning_rate": 4.997496987311431e-05,
+      "loss": 0.5979,
+      "step": 90
+    },
+    {
+      "epoch": 0.11648,
+      "grad_norm": 0.45501958535738946,
+      "learning_rate": 4.996971460796929e-05,
+      "loss": 0.6019,
+      "step": 91
+    },
+    {
+      "epoch": 0.11776,
+      "grad_norm": 0.4287172104360816,
+      "learning_rate": 4.9963959264103544e-05,
+      "loss": 0.5955,
+      "step": 92
+    },
+    {
+      "epoch": 0.11904,
+      "grad_norm": 0.409872269342458,
+      "learning_rate": 4.995770395678171e-05,
+      "loss": 0.5927,
+      "step": 93
+    },
+    {
+      "epoch": 0.12032,
+      "grad_norm": 0.4304173966206036,
+      "learning_rate": 4.995094881128138e-05,
+      "loss": 0.5967,
+      "step": 94
+    },
+    {
+      "epoch": 0.1216,
+      "grad_norm": 0.4229799776298517,
+      "learning_rate": 4.994369396289063e-05,
+      "loss": 0.6084,
+      "step": 95
+    },
+    {
+      "epoch": 0.12288,
+      "grad_norm": 0.4509596954971553,
+      "learning_rate": 4.9935939556905295e-05,
+      "loss": 0.6134,
+      "step": 96
+    },
+    {
+      "epoch": 0.12416,
+      "grad_norm": 0.467661146414229,
+      "learning_rate": 4.992768574862603e-05,
+      "loss": 0.5986,
+      "step": 97
+    },
+    {
+      "epoch": 0.12544,
+      "grad_norm": 0.42432875998240194,
+      "learning_rate": 4.9918932703355256e-05,
+      "loss": 0.6028,
+      "step": 98
+    },
+    {
+      "epoch": 0.12672,
+      "grad_norm": 0.43479377184835605,
+      "learning_rate": 4.990968059639379e-05,
+      "loss": 0.5942,
+      "step": 99
+    },
+    {
+      "epoch": 0.128,
+      "grad_norm": 0.3680676685801686,
+      "learning_rate": 4.989992961303738e-05,
+      "loss": 0.5994,
+      "step": 100
+    },
+    {
+      "epoch": 0.12928,
+      "grad_norm": 0.3956815409903461,
+      "learning_rate": 4.9889679948572974e-05,
+      "loss": 0.5871,
+      "step": 101
+    },
+    {
+      "epoch": 0.13056,
+      "grad_norm": 0.34354949934586104,
+      "learning_rate": 4.98789318082748e-05,
+      "loss": 0.5873,
+      "step": 102
+    },
+    {
+      "epoch": 0.13184,
+      "grad_norm": 0.3608260963951222,
+      "learning_rate": 4.986768540740028e-05,
+      "loss": 0.5883,
+      "step": 103
+    },
+    {
+      "epoch": 0.13312,
+      "grad_norm": 0.3937004101078116,
+      "learning_rate": 4.98559409711857e-05,
+      "loss": 0.6029,
+      "step": 104
+    },
+    {
+      "epoch": 0.1344,
+      "grad_norm": 0.3401718481532899,
+      "learning_rate": 4.9843698734841705e-05,
+      "loss": 0.5983,
+      "step": 105
+    },
+    {
+      "epoch": 0.13568,
+      "grad_norm": 0.4371868869288284,
+      "learning_rate": 4.983095894354858e-05,
+      "loss": 0.5866,
+      "step": 106
+    },
+    {
+      "epoch": 0.13696,
+      "grad_norm": 0.3722813571279646,
+      "learning_rate": 4.981772185245135e-05,
+      "loss": 0.5954,
+      "step": 107
+    },
+    {
+      "epoch": 0.13824,
+      "grad_norm": 0.36493596395606354,
+      "learning_rate": 4.980398772665468e-05,
+      "loss": 0.5806,
+      "step": 108
+    },
+    {
+      "epoch": 0.13952,
+      "grad_norm": 0.43678937522389644,
+      "learning_rate": 4.9789756841217546e-05,
+      "loss": 0.595,
+      "step": 109
+    },
+    {
+      "epoch": 0.1408,
+      "grad_norm": 0.34968596729530604,
+      "learning_rate": 4.977502948114772e-05,
+      "loss": 0.5999,
+      "step": 110
+    },
+    {
+      "epoch": 0.14208,
+      "grad_norm": 0.4035249077012057,
+      "learning_rate": 4.9759805941396075e-05,
+      "loss": 0.582,
+      "step": 111
+    },
+    {
+      "epoch": 0.14336,
+      "grad_norm": 0.3396387531525401,
+      "learning_rate": 4.974408652685072e-05,
+      "loss": 0.5912,
+      "step": 112
+    },
+    {
+      "epoch": 0.14464,
+      "grad_norm": 0.3888124435581031,
+      "learning_rate": 4.9727871552330794e-05,
+      "loss": 0.5994,
+      "step": 113
+    },
+    {
+      "epoch": 0.14592,
+      "grad_norm": 0.3487289265208422,
+      "learning_rate": 4.971116134258025e-05,
+      "loss": 0.598,
+      "step": 114
+    },
+    {
+      "epoch": 0.1472,
+      "grad_norm": 0.34084258932596606,
+      "learning_rate": 4.969395623226133e-05,
+      "loss": 0.5965,
+      "step": 115
+    },
+    {
+      "epoch": 0.14848,
+      "grad_norm": 0.33211872605390524,
+      "learning_rate": 4.967625656594782e-05,
+      "loss": 0.5984,
+      "step": 116
+    },
+    {
+      "epoch": 0.14976,
+      "grad_norm": 0.31055192632357626,
+      "learning_rate": 4.9658062698118213e-05,
+      "loss": 0.593,
+      "step": 117
+    },
+    {
+      "epoch": 0.15104,
+      "grad_norm": 0.35790400007793166,
+      "learning_rate": 4.963937499314857e-05,
+      "loss": 0.6035,
+      "step": 118
+    },
+    {
+      "epoch": 0.15232,
+      "grad_norm": 0.31118450185510343,
+      "learning_rate": 4.962019382530521e-05,
+      "loss": 0.5811,
+      "step": 119
+    },
+    {
+      "epoch": 0.1536,
+      "grad_norm": 0.3326176465041298,
+      "learning_rate": 4.960051957873725e-05,
+      "loss": 0.581,
+      "step": 120
+    },
+    {
+      "epoch": 0.15488,
+      "grad_norm": 0.30210249377153575,
+      "learning_rate": 4.958035264746893e-05,
+      "loss": 0.5837,
+      "step": 121
+    },
+    {
+      "epoch": 0.15616,
+      "grad_norm": 0.3480385124671555,
+      "learning_rate": 4.955969343539162e-05,
+      "loss": 0.5768,
+      "step": 122
+    },
+    {
+      "epoch": 0.15744,
+      "grad_norm": 0.3003392569743352,
+      "learning_rate": 4.9538542356255866e-05,
+      "loss": 0.5938,
+      "step": 123
+    },
+    {
+      "epoch": 0.15872,
+      "grad_norm": 0.32082565179488104,
+      "learning_rate": 4.9516899833663e-05,
+      "loss": 0.5948,
+      "step": 124
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3564349708048278,
+      "learning_rate": 4.949476630105669e-05,
+      "loss": 0.595,
+      "step": 125
+    },
+    {
+      "epoch": 0.16128,
+      "grad_norm": 0.32049541972124757,
+      "learning_rate": 4.94721422017143e-05,
+      "loss": 0.5838,
+      "step": 126
+    },
+    {
+      "epoch": 0.16256,
+      "grad_norm": 0.3317680882353993,
+      "learning_rate": 4.944902798873794e-05,
+      "loss": 0.5952,
+      "step": 127
+    },
+    {
+      "epoch": 0.16384,
+      "grad_norm": 0.3381465061198974,
+      "learning_rate": 4.942542412504543e-05,
+      "loss": 0.6004,
+      "step": 128
+    },
+    {
+      "epoch": 0.16512,
+      "grad_norm": 0.38351657127693595,
+      "learning_rate": 4.940133108336105e-05,
+      "loss": 0.6014,
+      "step": 129
+    },
+    {
+      "epoch": 0.1664,
+      "grad_norm": 0.3276142738951724,
+      "learning_rate": 4.9376749346206006e-05,
+      "loss": 0.5853,
+      "step": 130
+    },
+    {
+      "epoch": 0.16768,
+      "grad_norm": 0.37146400882939534,
+      "learning_rate": 4.935167940588887e-05,
+      "loss": 0.5995,
+      "step": 131
+    },
+    {
+      "epoch": 0.16896,
+      "grad_norm": 0.32804274509201087,
+      "learning_rate": 4.9326121764495596e-05,
+      "loss": 0.5955,
+      "step": 132
+    },
+    {
+      "epoch": 0.17024,
+      "grad_norm": 0.3344845806030499,
+      "learning_rate": 4.9300076933879574e-05,
+      "loss": 0.5818,
+      "step": 133
+    },
+    {
+      "epoch": 0.17152,
+      "grad_norm": 0.3479572078392269,
+      "learning_rate": 4.92735454356513e-05,
+      "loss": 0.5941,
+      "step": 134
+    },
+    {
+      "epoch": 0.1728,
+      "grad_norm": 0.34868252062960353,
+      "learning_rate": 4.924652780116799e-05,
+      "loss": 0.5898,
+      "step": 135
+    },
+    {
+      "epoch": 0.17408,
+      "grad_norm": 0.35674279058993497,
+      "learning_rate": 4.921902457152289e-05,
+      "loss": 0.5899,
+      "step": 136
+    },
+    {
+      "epoch": 0.17536,
+      "grad_norm": 0.3672614416380493,
+      "learning_rate": 4.9191036297534454e-05,
+      "loss": 0.585,
+      "step": 137
+    },
+    {
+      "epoch": 0.17664,
+      "grad_norm": 0.4039478601084677,
+      "learning_rate": 4.916256353973535e-05,
+      "loss": 0.5994,
+      "step": 138
+    },
+    {
+      "epoch": 0.17792,
+      "grad_norm": 0.3428958061155067,
+      "learning_rate": 4.913360686836117e-05,
+      "loss": 0.575,
+      "step": 139
+    },
+    {
+      "epoch": 0.1792,
+      "grad_norm": 0.4024960602256603,
+      "learning_rate": 4.910416686333906e-05,
+      "loss": 0.5913,
+      "step": 140
+    },
+    {
+      "epoch": 0.18048,
+      "grad_norm": 0.31040065034832104,
+      "learning_rate": 4.907424411427608e-05,
+      "loss": 0.5761,
+      "step": 141
+    },
+    {
+      "epoch": 0.18176,
+      "grad_norm": 0.359237099401051,
+      "learning_rate": 4.90438392204474e-05,
+      "loss": 0.5885,
+      "step": 142
+    },
+    {
+      "epoch": 0.18304,
+      "grad_norm": 0.3357545415879296,
+      "learning_rate": 4.901295279078431e-05,
+      "loss": 0.5907,
+      "step": 143
+    },
+    {
+      "epoch": 0.18432,
+      "grad_norm": 0.2846403022642179,
+      "learning_rate": 4.898158544386201e-05,
+      "loss": 0.5886,
+      "step": 144
+    },
+    {
+      "epoch": 0.1856,
+      "grad_norm": 0.3636245125193307,
+      "learning_rate": 4.894973780788722e-05,
+      "loss": 0.5816,
+      "step": 145
+    },
+    {
+      "epoch": 0.18688,
+      "grad_norm": 0.25440894793562924,
+      "learning_rate": 4.8917410520685635e-05,
+      "loss": 0.576,
+      "step": 146
+    },
+    {
+      "epoch": 0.18816,
+      "grad_norm": 0.3380189678855273,
+      "learning_rate": 4.888460422968908e-05,
+      "loss": 0.5931,
+      "step": 147
+    },
+    {
+      "epoch": 0.18944,
+      "grad_norm": 0.3096794617975588,
+      "learning_rate": 4.885131959192262e-05,
+      "loss": 0.5829,
+      "step": 148
+    },
+    {
+      "epoch": 0.19072,
+      "grad_norm": 0.280174710159943,
+      "learning_rate": 4.881755727399134e-05,
+      "loss": 0.5794,
+      "step": 149
+    },
+    {
+      "epoch": 0.192,
+      "grad_norm": 0.31769340776297994,
+      "learning_rate": 4.878331795206705e-05,
+      "loss": 0.5729,
+      "step": 150
+    },
+    {
+      "epoch": 0.19328,
+      "grad_norm": 0.31671973855902796,
+      "learning_rate": 4.8748602311874694e-05,
+      "loss": 0.5905,
+      "step": 151
+    },
+    {
+      "epoch": 0.19456,
+      "grad_norm": 0.32614211009906474,
+      "learning_rate": 4.8713411048678635e-05,
+      "loss": 0.5855,
+      "step": 152
+    },
+    {
+      "epoch": 0.19584,
+      "grad_norm": 0.29921149443441614,
+      "learning_rate": 4.8677744867268764e-05,
+      "loss": 0.5779,
+      "step": 153
+    },
+    {
+      "epoch": 0.19712,
+      "grad_norm": 0.3558339409344647,
+      "learning_rate": 4.8641604481946314e-05,
+      "loss": 0.5892,
+      "step": 154
+    },
+    {
+      "epoch": 0.1984,
+      "grad_norm": 0.285079025062,
+      "learning_rate": 4.8604990616509616e-05,
+      "loss": 0.5912,
+      "step": 155
+    },
+    {
+      "epoch": 0.19968,
+      "grad_norm": 0.32189736402098207,
+      "learning_rate": 4.856790400423958e-05,
+      "loss": 0.5881,
+      "step": 156
+    },
+    {
+      "epoch": 0.20096,
+      "grad_norm": 0.3293153125716864,
+      "learning_rate": 4.8530345387885004e-05,
+      "loss": 0.5679,
+      "step": 157
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 781,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 157,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 194598775488512.0,
+  "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 sha256:17d4ae4b6b6214515a5dd9c49ec78773756cbcc25fe748eaf748b8059d9427af
+size 8081

zero_to_fp32.py

@@ -0,0 +1,760 @@
+#!/usr/bin/env python
+
+# Copyright (c) Microsoft Corporation.
+# SPDX-License-Identifier: Apache-2.0
+
+# DeepSpeed Team
+
+# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
+# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
+# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
+# application.
+#
+# example:
+#   python zero_to_fp32.py . output_dir/
+#   or
+#   python zero_to_fp32.py . output_dir/ --safe_serialization
+
+import argparse
+import torch
+import glob
+import math
+import os
+import re
+import gc
+import json
+import numpy as np
+from tqdm import tqdm
+from collections import OrderedDict
+from dataclasses import dataclass
+
+# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
+# DeepSpeed data structures it has to be available in the current python environment.
+from deepspeed.utils import logger
+from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
+                                            FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
+                                            FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
+
+
+@dataclass
+class zero_model_state:
+    buffers: dict()
+    param_shapes: dict()
+    shared_params: list
+    ds_version: int
+    frozen_param_shapes: dict()
+    frozen_param_fragments: dict()
+
+
+debug = 0
+
+# load to cpu
+device = torch.device('cpu')
+
+
+def atoi(text):
+    return int(text) if text.isdigit() else text
+
+
+def natural_keys(text):
+    '''
+    alist.sort(key=natural_keys) sorts in human order
+    http://nedbatchelder.com/blog/200712/human_sorting.html
+    (See Toothy's implementation in the comments)
+    '''
+    return [atoi(c) for c in re.split(r'(\d+)', text)]
+
+
+def get_model_state_file(checkpoint_dir, zero_stage):
+    if not os.path.isdir(checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
+
+    # there should be only one file
+    if zero_stage <= 2:
+        file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
+    elif zero_stage == 3:
+        file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
+
+    if not os.path.exists(file):
+        raise FileNotFoundError(f"can't find model states file at '{file}'")
+
+    return file
+
+
+def get_checkpoint_files(checkpoint_dir, glob_pattern):
+    # XXX: need to test that this simple glob rule works for multi-node setup too
+    ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
+
+    if len(ckpt_files) == 0:
+        raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
+
+    return ckpt_files
+
+
+def get_optim_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
+
+
+def get_model_state_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
+
+
+def parse_model_states(files):
+    zero_model_states = []
+    for file in files:
+        state_dict = torch.load(file, map_location=device, weights_only=False)
+
+        if BUFFER_NAMES not in state_dict:
+            raise ValueError(f"{file} is not a model state checkpoint")
+        buffer_names = state_dict[BUFFER_NAMES]
+        if debug:
+            print("Found buffers:", buffer_names)
+
+        # recover just the buffers while restoring them to fp32 if they were saved in fp16
+        buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
+        param_shapes = state_dict[PARAM_SHAPES]
+
+        # collect parameters that are included in param_shapes
+        param_names = []
+        for s in param_shapes:
+            for name in s.keys():
+                param_names.append(name)
+
+        # update with frozen parameters
+        frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
+        if frozen_param_shapes is not None:
+            if debug:
+                print(f"Found frozen_param_shapes: {frozen_param_shapes}")
+            param_names += list(frozen_param_shapes.keys())
+
+        # handle shared params
+        shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
+
+        ds_version = state_dict.get(DS_VERSION, None)
+
+        frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
+
+        z_model_state = zero_model_state(buffers=buffers,
+                                         param_shapes=param_shapes,
+                                         shared_params=shared_params,
+                                         ds_version=ds_version,
+                                         frozen_param_shapes=frozen_param_shapes,
+                                         frozen_param_fragments=frozen_param_fragments)
+        zero_model_states.append(z_model_state)
+
+    return zero_model_states
+
+
+def parse_optim_states(files, ds_checkpoint_dir):
+    total_files = len(files)
+    state_dicts = []
+    for f in tqdm(files, desc='Loading checkpoint shards'):
+        state_dict = torch.load(f, map_location=device, mmap=True, weights_only=False)
+        # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
+        # and also handle the case where it was already removed by another helper script
+        state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
+        state_dicts.append(state_dict)
+
+    if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
+        raise ValueError(f"{files[0]} is not a zero checkpoint")
+    zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
+    world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
+
+    # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
+    # parameters can be different from data parallelism for non-expert parameters. So we can just
+    # use the max of the partition_count to get the dp world_size.
+
+    if type(world_size) is list:
+        world_size = max(world_size)
+
+    if world_size != total_files:
+        raise ValueError(
+            f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
+            "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
+        )
+
+    # the groups are named differently in each stage
+    if zero_stage <= 2:
+        fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
+    elif zero_stage == 3:
+        fp32_groups_key = FP32_FLAT_GROUPS
+    else:
+        raise ValueError(f"unknown zero stage {zero_stage}")
+
+    fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
+    return zero_stage, world_size, fp32_flat_groups
+
+
+def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters):
+    """
+    Returns fp32 state_dict reconstructed from ds checkpoint
+
+    Args:
+        - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
+
+    """
+    print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
+
+    optim_files = get_optim_files(ds_checkpoint_dir)
+    zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
+    print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
+
+    model_files = get_model_state_files(ds_checkpoint_dir)
+
+    zero_model_states = parse_model_states(model_files)
+    print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
+
+    if zero_stage <= 2:
+        return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+    elif zero_stage == 3:
+        return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+
+
+def _zero2_merge_frozen_params(state_dict, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+    frozen_param_fragments = zero_model_states[0].frozen_param_fragments
+
+    if debug:
+        num_elem = sum(s.numel() for s in frozen_param_shapes.values())
+        print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        state_dict[name] = frozen_param_fragments[name]
+
+        if debug:
+            print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _has_callable(obj, fn):
+    attr = getattr(obj, fn, None)
+    return callable(attr)
+
+
+def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+
+    # Reconstruction protocol:
+    #
+    # XXX: document this
+
+    if debug:
+        for i in range(world_size):
+            for j in range(len(fp32_flat_groups[0])):
+                print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
+
+    # XXX: memory usage doubles here (zero2)
+    num_param_groups = len(fp32_flat_groups[0])
+    merged_single_partition_of_fp32_groups = []
+    for i in range(num_param_groups):
+        merged_partitions = [sd[i] for sd in fp32_flat_groups]
+        full_single_fp32_vector = torch.cat(merged_partitions, 0)
+        merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
+    avail_numel = sum(
+        [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
+
+    if debug:
+        wanted_params = sum([len(shapes) for shapes in param_shapes])
+        wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
+        # not asserting if there is a mismatch due to possible padding
+        print(f"Have {avail_numel} numels to process.")
+        print(f"Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    total_numel = 0
+    total_params = 0
+    for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
+        offset = 0
+        avail_numel = full_single_fp32_vector.numel()
+        for name, shape in shapes.items():
+
+            unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
+            total_numel += unpartitioned_numel
+            total_params += 1
+
+            if debug:
+                print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+            state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
+            offset += unpartitioned_numel
+
+        # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
+        # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
+        # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
+        # live optimizer object, so we are checking that the numbers are within the right range
+        align_to = 2 * world_size
+
+        def zero2_align(x):
+            return align_to * math.ceil(x / align_to)
+
+        if debug:
+            print(f"original offset={offset}, avail_numel={avail_numel}")
+
+        offset = zero2_align(offset)
+        avail_numel = zero2_align(avail_numel)
+
+        if debug:
+            print(f"aligned  offset={offset}, avail_numel={avail_numel}")
+
+        # Sanity check
+        if offset != avail_numel:
+            raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero2_merge_frozen_params(state_dict, zero_model_states)
+
+    _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def zero3_partitioned_param_info(unpartitioned_numel, world_size):
+    remainder = unpartitioned_numel % world_size
+    padding_numel = (world_size - remainder) if remainder else 0
+    partitioned_numel = math.ceil(unpartitioned_numel / world_size)
+    return partitioned_numel, padding_numel
+
+
+def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    if debug:
+        for i in range(world_size):
+            num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
+            print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in zero_model_states[0].frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
+        state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
+
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+class GatheredTensor:
+    """
+    A pseudo tensor that collects partitioned weights.
+    It is more memory efficient when there are multiple groups.
+    """
+
+    def __init__(self, flat_groups, flat_groups_offset, offset, partitioned_numel, shape):
+        self.flat_groups = flat_groups
+        self.flat_groups_offset = flat_groups_offset
+        self.offset = offset
+        self.partitioned_numel = partitioned_numel
+        self.shape = shape
+        self.dtype = self.flat_groups[0][0].dtype
+
+    def contiguous(self):
+        """
+        Merge partitioned weights from flat_groups into a single tensor.
+        """
+        end_idx = self.offset + self.partitioned_numel
+        world_size = len(self.flat_groups)
+        pad_flat_param_chunks = []
+
+        for rank_i in range(world_size):
+            # for each rank, we need to collect weights from related group/groups
+            flat_groups_at_rank_i = self.flat_groups[rank_i]
+            start_group_id = None
+            end_group_id = None
+            for group_id in range(len(self.flat_groups_offset)):
+                if self.flat_groups_offset[group_id] <= self.offset < self.flat_groups_offset[group_id + 1]:
+                    start_group_id = group_id
+                if self.flat_groups_offset[group_id] < end_idx <= self.flat_groups_offset[group_id + 1]:
+                    end_group_id = group_id
+                    break
+            # collect weights from related group/groups
+            for group_id in range(start_group_id, end_group_id + 1):
+                flat_tensor = flat_groups_at_rank_i[group_id]
+                start_offset = self.offset - self.flat_groups_offset[group_id]
+                end_offset = min(end_idx, self.flat_groups_offset[group_id + 1]) - self.flat_groups_offset[group_id]
+                pad_flat_param_chunks.append(flat_tensor[start_offset:end_offset])
+
+        # collect weights from all ranks
+        pad_flat_param = torch.cat(pad_flat_param_chunks, dim=0)
+        param = pad_flat_param[:self.shape.numel()].view(self.shape).contiguous()
+        return param
+
+
+def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+    avail_numel = sum([flat_group.numel() for flat_group in fp32_flat_groups[0]]) * world_size
+
+    # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
+    # param, re-consolidating each param, while dealing with padding if any
+
+    # merge list of dicts, preserving order
+    param_shapes = {k: v for d in param_shapes for k, v in d.items()}
+
+    if debug:
+        for i in range(world_size):
+            print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
+
+        wanted_params = len(param_shapes)
+        wanted_numel = sum(shape.numel() for shape in param_shapes.values())
+        # not asserting if there is a mismatch due to possible padding
+        avail_numel = fp32_flat_groups[0].numel() * world_size
+        print(f"Trainable params: Have {avail_numel} numels to process.")
+        print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    offset = 0
+    total_numel = 0
+    total_params = 0
+    flat_groups_offset = [0] + list(np.cumsum([flat_tensor.numel() for flat_tensor in fp32_flat_groups[0]]))
+    for name, shape in tqdm(param_shapes.items(), desc='Gathering sharded weights'):
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+        total_params += 1
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+        # memory efficient tensor
+        tensor = GatheredTensor(fp32_flat_groups, flat_groups_offset, offset, partitioned_numel, shape)
+        state_dict[name] = tensor
+        offset += partitioned_numel
+
+    offset *= world_size
+
+    # Sanity check
+    if offset != avail_numel:
+        raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
+
+    _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def to_torch_tensor(state_dict, return_empty_tensor=False):
+    """
+    Convert state_dict of GatheredTensor to torch tensor
+    """
+    torch_state_dict = {}
+    converted_tensors = {}
+    for name, tensor in state_dict.items():
+        tensor_id = id(tensor)
+        if tensor_id in converted_tensors:  # shared tensors
+            shared_tensor = torch_state_dict[converted_tensors[tensor_id]]
+            torch_state_dict[name] = shared_tensor
+        else:
+            converted_tensors[tensor_id] = name
+            if return_empty_tensor:
+                torch_state_dict[name] = torch.empty(tensor.shape, dtype=tensor.dtype)
+            else:
+                torch_state_dict[name] = tensor.contiguous()
+    return torch_state_dict
+
+
+def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir,
+                                             tag=None,
+                                             exclude_frozen_parameters=False,
+                                             lazy_mode=False):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
+    ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
+    via a model hub.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+        - ``lazy_mode``: get state_dict in lazy mode. It returns a dict of pesduo tensor instead of torch tensor, which is more memory efficient.
+          Convert the pesduo tensor to torch tensor by ``.contiguous()``
+
+    Returns:
+        - pytorch ``state_dict``
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+        # do the training and checkpoint saving
+        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
+        model = model.cpu() # move to cpu
+        model.load_state_dict(state_dict)
+        # submit to model hub or save the model to share with others
+
+    In this example the ``model`` will no longer be usable in the deepspeed context of the same
+    application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
+
+    Note: the above usage may not work if your application doesn't have sufficient free CPU memory.
+    You may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
+    the checkpoint. Or you can load state_dict in lazy mode ::
+
+        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, lazy_mode=True) # not on cpu
+        for name, lazy_tensor in state_dict.item():
+            tensor = lazy_tensor.contiguous()  # to cpu
+            print(name, tensor)
+            # del tensor to release memory if it no longer in use
+    """
+    if tag is None:
+        latest_path = os.path.join(checkpoint_dir, 'latest')
+        if os.path.isfile(latest_path):
+            with open(latest_path, 'r') as fd:
+                tag = fd.read().strip()
+        else:
+            raise ValueError(f"Unable to find 'latest' file at {latest_path}")
+
+    ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
+
+    if not os.path.isdir(ds_checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
+
+    state_dict = _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
+    if lazy_mode:
+        return state_dict
+    else:
+        return to_torch_tensor(state_dict)
+
+
+def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir,
+                                               output_dir,
+                                               max_shard_size="5GB",
+                                               safe_serialization=False,
+                                               tag=None,
+                                               exclude_frozen_parameters=False):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
+    loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``output_dir``: directory to the pytorch fp32 state_dict output files
+        - ``max_shard_size``: the maximum size for a checkpoint before being sharded, default value is 5GB
+        - ``safe_serialization``:  whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+    """
+
+    # Dependency pre-check
+    if safe_serialization:
+        try:
+            from safetensors.torch import save_file
+        except ImportError:
+            print('If you want to use `safe_serialization`, please `pip install safetensors`')
+            raise
+    if max_shard_size is not None:
+        try:
+            from huggingface_hub import split_torch_state_dict_into_shards
+        except ImportError:
+            print('If you want to use `max_shard_size`, please `pip install huggingface_hub`')
+            raise
+
+    # Convert zero checkpoint to state_dict
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir,
+                                                          tag,
+                                                          exclude_frozen_parameters,
+                                                          lazy_mode=True)
+
+    # Shard the model if it is too big.
+    weights_name = "model.safetensors" if safe_serialization else "pytorch_model.bin"
+    if max_shard_size is not None:
+        filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors")
+        # an memory-efficient approach for sharding
+        empty_state_dict = to_torch_tensor(state_dict, return_empty_tensor=True)
+        state_dict_split = split_torch_state_dict_into_shards(empty_state_dict,
+                                                              filename_pattern=filename_pattern,
+                                                              max_shard_size=max_shard_size)
+    else:
+        from collections import namedtuple
+        StateDictSplit = namedtuple("StateDictSplit", ["is_sharded", "filename_to_tensors"])
+        state_dict_split = StateDictSplit(is_sharded=False,
+                                          filename_to_tensors={weights_name: list(state_dict.keys())})
+
+    # Save the model by shard
+    os.makedirs(output_dir, exist_ok=True)
+    filename_to_tensors = state_dict_split.filename_to_tensors.items()
+    for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"):
+        shard_state_dict = {tensor_name: state_dict[tensor_name] for tensor_name in tensors}
+        shard_state_dict = to_torch_tensor(shard_state_dict)
+        output_path = os.path.join(output_dir, shard_file)
+        if safe_serialization:
+            save_file(shard_state_dict, output_path, metadata={"format": "pt"})
+        else:
+            torch.save(shard_state_dict, output_path)
+        # release the memory of current shard
+        for tensor_name in list(shard_state_dict.keys()):
+            del state_dict[tensor_name]
+            del shard_state_dict[tensor_name]
+        del shard_state_dict
+        gc.collect()
+
+    # Save index if sharded
+    if state_dict_split.is_sharded:
+        index = {
+            "metadata": state_dict_split.metadata,
+            "weight_map": state_dict_split.tensor_to_filename,
+        }
+        save_index_file = "model.safetensors.index.json" if safe_serialization else "pytorch_model.bin.index.json"
+        save_index_file = os.path.join(output_dir, save_index_file)
+        with open(save_index_file, "w", encoding="utf-8") as f:
+            content = json.dumps(index, indent=2, sort_keys=True) + "\n"
+            f.write(content)
+
+
+def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
+    """
+    1. Put the provided model to cpu
+    2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
+    3. Load it into the provided model
+
+    Args:
+        - ``model``: the model object to update
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+
+    Returns:
+        - ``model`: modified model
+
+    Make sure you have plenty of CPU memory available before you call this function. If you don't
+    have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
+    conveniently placed for you in the checkpoint folder.
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
+        model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
+        # submit to model hub or save the model to share with others
+
+    Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
+    of the same application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    """
+    logger.info(f"Extracting fp32 weights")
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
+
+    logger.info(f"Overwriting model with fp32 weights")
+    model = model.cpu()
+    model.load_state_dict(state_dict, strict=False)
+
+    return model
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("checkpoint_dir",
+                        type=str,
+                        help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
+    parser.add_argument("output_dir",
+                        type=str,
+                        help="directory to the pytorch fp32 state_dict output files"
+                        "(e.g. path/checkpoint-12-output/)")
+    parser.add_argument(
+        "--max_shard_size",
+        type=str,
+        default="5GB",
+        help="The maximum size for a checkpoint before being sharded. Checkpoints shard will then be each of size"
+        "lower than this size. If expressed as a string, needs to be digits followed by a unit (like `5MB`"
+        "We default it to 5GB in order for models to be able to run easily on free-tier google colab instances"
+        "without CPU OOM issues.")
+    parser.add_argument(
+        "--safe_serialization",
+        default=False,
+        action='store_true',
+        help="Whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).")
+    parser.add_argument("-t",
+                        "--tag",
+                        type=str,
+                        default=None,
+                        help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
+    parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters")
+    parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
+    args = parser.parse_args()
+
+    debug = args.debug
+
+    convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir,
+                                               args.output_dir,
+                                               max_shard_size=args.max_shard_size,
+                                               safe_serialization=args.safe_serialization,
+                                               tag=args.tag,
+                                               exclude_frozen_parameters=args.exclude_frozen_parameters)