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 +1,60 @@
+{
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 2048,
+  "initializer_range": 0.02,
+  "intermediate_size": 6144,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000,
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.55.0",
+  "use_cache": false,
+  "use_sliding_window": false,
+  "vocab_size": 151936
+}

generation_config.json

@@ -0,0 +1,6 @@
+{
+  "bos_token_id": 151643,
+  "eos_token_id": 151643,
+  "max_new_tokens": 2048,
+  "transformers_version": "4.55.0"
+}

global_step178/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f7ec42bb191e9ce883a21fff6b67770227a4abb56e720ff1b3e012e2b4a5a48e
+size 10323465953

global_step178/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6066b1a9277201d8466bcb773fc3a0f0a2b82466c8bf642c45d5274b20bf702b
+size 10323469089

global_step178/mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:993243f34ba0c4f8a0e8ab581f4cdb38cc6f3a7fa7d0d52f8de5b0db5049b6a2
+size 3441239781

latest

@@ -0,0 +1 @@
+global_step178

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:76cf3594574379da88852dd779b3fca790c333f27acea0a533c09916019c0a6a
+size 4063515640

rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:02ea5dcfd1b4a49b41b4fa01a8b24bba6186957162c3fd555ebff28620c7268b
+size 14917

rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c2a5af18bb5eae8b7fd6bdef66259014d98ba87ffb16d614bba38f2c32030798
+size 14917

scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c1471b13c3527644f2470e235b1883dca611fdf88e2d7aeb538870419cdea2e
+size 1465

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 +1,1294 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.335664335664336,
+  "eval_steps": 500,
+  "global_step": 180,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.018648018648018648,
+      "grad_norm": 7.488297462463379,
+      "learning_rate": 0.0,
+      "loss": 0.7208,
+      "step": 1
+    },
+    {
+      "epoch": 0.037296037296037296,
+      "grad_norm": 8.05743408203125,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.7936,
+      "step": 2
+    },
+    {
+      "epoch": 0.055944055944055944,
+      "grad_norm": 7.406175136566162,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.7104,
+      "step": 3
+    },
+    {
+      "epoch": 0.07459207459207459,
+      "grad_norm": 4.18705940246582,
+      "learning_rate": 1e-05,
+      "loss": 0.6796,
+      "step": 4
+    },
+    {
+      "epoch": 0.09324009324009325,
+      "grad_norm": 5.086103916168213,
+      "learning_rate": 9.99965389153533e-06,
+      "loss": 0.6597,
+      "step": 5
+    },
+    {
+      "epoch": 0.11188811188811189,
+      "grad_norm": 5.0644330978393555,
+      "learning_rate": 9.998615614057743e-06,
+      "loss": 0.6858,
+      "step": 6
+    },
+    {
+      "epoch": 0.13053613053613053,
+      "grad_norm": 5.087165355682373,
+      "learning_rate": 9.996885311309892e-06,
+      "loss": 0.6442,
+      "step": 7
+    },
+    {
+      "epoch": 0.14918414918414918,
+      "grad_norm": 4.19765043258667,
+      "learning_rate": 9.994463222840748e-06,
+      "loss": 0.5934,
+      "step": 8
+    },
+    {
+      "epoch": 0.16783216783216784,
+      "grad_norm": 3.387387752532959,
+      "learning_rate": 9.991349683972435e-06,
+      "loss": 0.607,
+      "step": 9
+    },
+    {
+      "epoch": 0.1864801864801865,
+      "grad_norm": 2.290091037750244,
+      "learning_rate": 9.987545125753818e-06,
+      "loss": 0.6387,
+      "step": 10
+    },
+    {
+      "epoch": 0.20512820512820512,
+      "grad_norm": 1.8914862871170044,
+      "learning_rate": 9.983050074900824e-06,
+      "loss": 0.5773,
+      "step": 11
+    },
+    {
+      "epoch": 0.22377622377622378,
+      "grad_norm": 1.6402865648269653,
+      "learning_rate": 9.977865153723508e-06,
+      "loss": 0.5736,
+      "step": 12
+    },
+    {
+      "epoch": 0.24242424242424243,
+      "grad_norm": 1.4222232103347778,
+      "learning_rate": 9.971991080039912e-06,
+      "loss": 0.531,
+      "step": 13
+    },
+    {
+      "epoch": 0.26107226107226106,
+      "grad_norm": 1.2749735116958618,
+      "learning_rate": 9.965428667076687e-06,
+      "loss": 0.5638,
+      "step": 14
+    },
+    {
+      "epoch": 0.27972027972027974,
+      "grad_norm": 1.2392884492874146,
+      "learning_rate": 9.958178823356503e-06,
+      "loss": 0.5367,
+      "step": 15
+    },
+    {
+      "epoch": 0.29836829836829837,
+      "grad_norm": 1.188359022140503,
+      "learning_rate": 9.950242552572272e-06,
+      "loss": 0.5347,
+      "step": 16
+    },
+    {
+      "epoch": 0.317016317016317,
+      "grad_norm": 1.1442219018936157,
+      "learning_rate": 9.941620953448195e-06,
+      "loss": 0.5608,
+      "step": 17
+    },
+    {
+      "epoch": 0.3356643356643357,
+      "grad_norm": 1.0253201723098755,
+      "learning_rate": 9.932315219587641e-06,
+      "loss": 0.4761,
+      "step": 18
+    },
+    {
+      "epoch": 0.3543123543123543,
+      "grad_norm": 1.076796293258667,
+      "learning_rate": 9.922326639307918e-06,
+      "loss": 0.4992,
+      "step": 19
+    },
+    {
+      "epoch": 0.372960372960373,
+      "grad_norm": 1.1631619930267334,
+      "learning_rate": 9.911656595461899e-06,
+      "loss": 0.5297,
+      "step": 20
+    },
+    {
+      "epoch": 0.3916083916083916,
+      "grad_norm": 0.9999617338180542,
+      "learning_rate": 9.900306565246579e-06,
+      "loss": 0.497,
+      "step": 21
+    },
+    {
+      "epoch": 0.41025641025641024,
+      "grad_norm": 1.0188584327697754,
+      "learning_rate": 9.888278119998573e-06,
+      "loss": 0.5182,
+      "step": 22
+    },
+    {
+      "epoch": 0.4289044289044289,
+      "grad_norm": 0.9809191226959229,
+      "learning_rate": 9.875572924976568e-06,
+      "loss": 0.5251,
+      "step": 23
+    },
+    {
+      "epoch": 0.44755244755244755,
+      "grad_norm": 0.9418354630470276,
+      "learning_rate": 9.86219273913078e-06,
+      "loss": 0.5044,
+      "step": 24
+    },
+    {
+      "epoch": 0.4662004662004662,
+      "grad_norm": 0.9674189686775208,
+      "learning_rate": 9.848139414859441e-06,
+      "loss": 0.4687,
+      "step": 25
+    },
+    {
+      "epoch": 0.48484848484848486,
+      "grad_norm": 1.0557061433792114,
+      "learning_rate": 9.833414897752346e-06,
+      "loss": 0.4805,
+      "step": 26
+    },
+    {
+      "epoch": 0.5034965034965035,
+      "grad_norm": 1.0554925203323364,
+      "learning_rate": 9.818021226321502e-06,
+      "loss": 0.4917,
+      "step": 27
+    },
+    {
+      "epoch": 0.5221445221445221,
+      "grad_norm": 0.9118618369102478,
+      "learning_rate": 9.801960531718898e-06,
+      "loss": 0.5005,
+      "step": 28
+    },
+    {
+      "epoch": 0.5407925407925408,
+      "grad_norm": 0.9393364787101746,
+      "learning_rate": 9.785235037441473e-06,
+      "loss": 0.5315,
+      "step": 29
+    },
+    {
+      "epoch": 0.5594405594405595,
+      "grad_norm": 0.8810527920722961,
+      "learning_rate": 9.767847059023292e-06,
+      "loss": 0.4342,
+      "step": 30
+    },
+    {
+      "epoch": 0.578088578088578,
+      "grad_norm": 0.872603714466095,
+      "learning_rate": 9.749799003714954e-06,
+      "loss": 0.4746,
+      "step": 31
+    },
+    {
+      "epoch": 0.5967365967365967,
+      "grad_norm": 0.9307128190994263,
+      "learning_rate": 9.731093370150349e-06,
+      "loss": 0.4962,
+      "step": 32
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.962816596031189,
+      "learning_rate": 9.71173274800072e-06,
+      "loss": 0.4795,
+      "step": 33
+    },
+    {
+      "epoch": 0.634032634032634,
+      "grad_norm": 0.8412047028541565,
+      "learning_rate": 9.691719817616148e-06,
+      "loss": 0.4556,
+      "step": 34
+    },
+    {
+      "epoch": 0.6526806526806527,
+      "grad_norm": 0.933653712272644,
+      "learning_rate": 9.671057349654481e-06,
+      "loss": 0.4802,
+      "step": 35
+    },
+    {
+      "epoch": 0.6713286713286714,
+      "grad_norm": 0.9413991570472717,
+      "learning_rate": 9.649748204697741e-06,
+      "loss": 0.4592,
+      "step": 36
+    },
+    {
+      "epoch": 0.6899766899766899,
+      "grad_norm": 0.9195394515991211,
+      "learning_rate": 9.627795332856107e-06,
+      "loss": 0.4607,
+      "step": 37
+    },
+    {
+      "epoch": 0.7086247086247086,
+      "grad_norm": 0.8376944065093994,
+      "learning_rate": 9.605201773359485e-06,
+      "loss": 0.4468,
+      "step": 38
+    },
+    {
+      "epoch": 0.7272727272727273,
+      "grad_norm": 0.9352995753288269,
+      "learning_rate": 9.581970654136752e-06,
+      "loss": 0.4958,
+      "step": 39
+    },
+    {
+      "epoch": 0.745920745920746,
+      "grad_norm": 0.8960230946540833,
+      "learning_rate": 9.55810519138271e-06,
+      "loss": 0.4782,
+      "step": 40
+    },
+    {
+      "epoch": 0.7645687645687645,
+      "grad_norm": 0.8604126572608948,
+      "learning_rate": 9.533608689112827e-06,
+      "loss": 0.4522,
+      "step": 41
+    },
+    {
+      "epoch": 0.7832167832167832,
+      "grad_norm": 1.0348273515701294,
+      "learning_rate": 9.508484538705823e-06,
+      "loss": 0.4831,
+      "step": 42
+    },
+    {
+      "epoch": 0.8018648018648019,
+      "grad_norm": 0.9544129371643066,
+      "learning_rate": 9.482736218434144e-06,
+      "loss": 0.4314,
+      "step": 43
+    },
+    {
+      "epoch": 0.8205128205128205,
+      "grad_norm": 0.9038442969322205,
+      "learning_rate": 9.45636729298243e-06,
+      "loss": 0.4567,
+      "step": 44
+    },
+    {
+      "epoch": 0.8391608391608392,
+      "grad_norm": 0.9116093516349792,
+      "learning_rate": 9.429381412954e-06,
+      "loss": 0.4364,
+      "step": 45
+    },
+    {
+      "epoch": 0.8578088578088578,
+      "grad_norm": 0.9651602506637573,
+      "learning_rate": 9.401782314365458e-06,
+      "loss": 0.4591,
+      "step": 46
+    },
+    {
+      "epoch": 0.8764568764568764,
+      "grad_norm": 0.8423922657966614,
+      "learning_rate": 9.37357381812946e-06,
+      "loss": 0.4307,
+      "step": 47
+    },
+    {
+      "epoch": 0.8951048951048951,
+      "grad_norm": 0.9258006811141968,
+      "learning_rate": 9.344759829525734e-06,
+      "loss": 0.4627,
+      "step": 48
+    },
+    {
+      "epoch": 0.9137529137529138,
+      "grad_norm": 0.9286342263221741,
+      "learning_rate": 9.315344337660422e-06,
+      "loss": 0.4559,
+      "step": 49
+    },
+    {
+      "epoch": 0.9324009324009324,
+      "grad_norm": 0.9417174458503723,
+      "learning_rate": 9.285331414913816e-06,
+      "loss": 0.4853,
+      "step": 50
+    },
+    {
+      "epoch": 0.951048951048951,
+      "grad_norm": 0.9240500330924988,
+      "learning_rate": 9.254725216376562e-06,
+      "loss": 0.4485,
+      "step": 51
+    },
+    {
+      "epoch": 0.9696969696969697,
+      "grad_norm": 0.955512523651123,
+      "learning_rate": 9.223529979274411e-06,
+      "loss": 0.4888,
+      "step": 52
+    },
+    {
+      "epoch": 0.9883449883449883,
+      "grad_norm": 0.9273045659065247,
+      "learning_rate": 9.191750022381613e-06,
+      "loss": 0.4718,
+      "step": 53
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.9273045659065247,
+      "learning_rate": 9.159389745423003e-06,
+      "loss": 0.4304,
+      "step": 54
+    },
+    {
+      "epoch": 1.0186480186480187,
+      "grad_norm": 1.3491257429122925,
+      "learning_rate": 9.126453628464889e-06,
+      "loss": 0.416,
+      "step": 55
+    },
+    {
+      "epoch": 1.0372960372960374,
+      "grad_norm": 0.9643077254295349,
+      "learning_rate": 9.09294623129482e-06,
+      "loss": 0.4462,
+      "step": 56
+    },
+    {
+      "epoch": 1.055944055944056,
+      "grad_norm": 1.1403921842575073,
+      "learning_rate": 9.058872192790314e-06,
+      "loss": 0.4276,
+      "step": 57
+    },
+    {
+      "epoch": 1.0745920745920745,
+      "grad_norm": 0.892738401889801,
+      "learning_rate": 9.02423623027663e-06,
+      "loss": 0.3873,
+      "step": 58
+    },
+    {
+      "epoch": 1.0932400932400932,
+      "grad_norm": 0.963935136795044,
+      "learning_rate": 8.98904313887369e-06,
+      "loss": 0.4257,
+      "step": 59
+    },
+    {
+      "epoch": 1.1118881118881119,
+      "grad_norm": 0.8977372646331787,
+      "learning_rate": 8.953297790832231e-06,
+      "loss": 0.4004,
+      "step": 60
+    },
+    {
+      "epoch": 1.1305361305361306,
+      "grad_norm": 0.9680702090263367,
+      "learning_rate": 8.917005134859263e-06,
+      "loss": 0.3976,
+      "step": 61
+    },
+    {
+      "epoch": 1.1491841491841492,
+      "grad_norm": 0.9117997884750366,
+      "learning_rate": 8.88017019543296e-06,
+      "loss": 0.3653,
+      "step": 62
+    },
+    {
+      "epoch": 1.167832167832168,
+      "grad_norm": 0.9713727235794067,
+      "learning_rate": 8.842798072107055e-06,
+      "loss": 0.4242,
+      "step": 63
+    },
+    {
+      "epoch": 1.1864801864801864,
+      "grad_norm": 0.9422354698181152,
+      "learning_rate": 8.804893938804839e-06,
+      "loss": 0.3848,
+      "step": 64
+    },
+    {
+      "epoch": 1.205128205128205,
+      "grad_norm": 1.0252125263214111,
+      "learning_rate": 8.766463043102864e-06,
+      "loss": 0.4362,
+      "step": 65
+    },
+    {
+      "epoch": 1.2237762237762237,
+      "grad_norm": 0.9968674182891846,
+      "learning_rate": 8.727510705504453e-06,
+      "loss": 0.402,
+      "step": 66
+    },
+    {
+      "epoch": 1.2424242424242424,
+      "grad_norm": 0.9503087401390076,
+      "learning_rate": 8.688042318703111e-06,
+      "loss": 0.4062,
+      "step": 67
+    },
+    {
+      "epoch": 1.2610722610722611,
+      "grad_norm": 1.0598119497299194,
+      "learning_rate": 8.648063346835943e-06,
+      "loss": 0.4133,
+      "step": 68
+    },
+    {
+      "epoch": 1.2797202797202798,
+      "grad_norm": 0.9262954592704773,
+      "learning_rate": 8.607579324727175e-06,
+      "loss": 0.4194,
+      "step": 69
+    },
+    {
+      "epoch": 1.2983682983682985,
+      "grad_norm": 1.0740933418273926,
+      "learning_rate": 8.566595857121902e-06,
+      "loss": 0.3986,
+      "step": 70
+    },
+    {
+      "epoch": 1.317016317016317,
+      "grad_norm": 1.044493317604065,
+      "learning_rate": 8.525118617910144e-06,
+      "loss": 0.3729,
+      "step": 71
+    },
+    {
+      "epoch": 1.3356643356643356,
+      "grad_norm": 0.8950116634368896,
+      "learning_rate": 8.483153349341336e-06,
+      "loss": 0.4125,
+      "step": 72
+    },
+    {
+      "epoch": 1.3543123543123543,
+      "grad_norm": 0.9413709044456482,
+      "learning_rate": 8.440705861229344e-06,
+      "loss": 0.3907,
+      "step": 73
+    },
+    {
+      "epoch": 1.372960372960373,
+      "grad_norm": 0.9134018421173096,
+      "learning_rate": 8.397782030148147e-06,
+      "loss": 0.3976,
+      "step": 74
+    },
+    {
+      "epoch": 1.3916083916083917,
+      "grad_norm": 0.8930859565734863,
+      "learning_rate": 8.354387798618254e-06,
+      "loss": 0.4085,
+      "step": 75
+    },
+    {
+      "epoch": 1.4102564102564101,
+      "grad_norm": 0.9354174733161926,
+      "learning_rate": 8.310529174284004e-06,
+      "loss": 0.3956,
+      "step": 76
+    },
+    {
+      "epoch": 1.428904428904429,
+      "grad_norm": 1.0586148500442505,
+      "learning_rate": 8.266212229081846e-06,
+      "loss": 0.4193,
+      "step": 77
+    },
+    {
+      "epoch": 1.4475524475524475,
+      "grad_norm": 1.0139036178588867,
+      "learning_rate": 8.221443098399733e-06,
+      "loss": 0.4102,
+      "step": 78
+    },
+    {
+      "epoch": 1.4662004662004662,
+      "grad_norm": 0.8833545446395874,
+      "learning_rate": 8.176227980227693e-06,
+      "loss": 0.3745,
+      "step": 79
+    },
+    {
+      "epoch": 1.4848484848484849,
+      "grad_norm": 0.9952216744422913,
+      "learning_rate": 8.130573134299782e-06,
+      "loss": 0.4088,
+      "step": 80
+    },
+    {
+      "epoch": 1.5034965034965035,
+      "grad_norm": 1.010685920715332,
+      "learning_rate": 8.084484881227449e-06,
+      "loss": 0.4057,
+      "step": 81
+    },
+    {
+      "epoch": 1.5221445221445222,
+      "grad_norm": 0.9328898191452026,
+      "learning_rate": 8.037969601624495e-06,
+      "loss": 0.3595,
+      "step": 82
+    },
+    {
+      "epoch": 1.5407925407925407,
+      "grad_norm": 0.9624993205070496,
+      "learning_rate": 7.99103373522373e-06,
+      "loss": 0.3785,
+      "step": 83
+    },
+    {
+      "epoch": 1.5594405594405596,
+      "grad_norm": 1.0064465999603271,
+      "learning_rate": 7.943683779985412e-06,
+      "loss": 0.3818,
+      "step": 84
+    },
+    {
+      "epoch": 1.578088578088578,
+      "grad_norm": 0.9292221069335938,
+      "learning_rate": 7.895926291197667e-06,
+      "loss": 0.3807,
+      "step": 85
+    },
+    {
+      "epoch": 1.5967365967365967,
+      "grad_norm": 0.9069374799728394,
+      "learning_rate": 7.847767880568944e-06,
+      "loss": 0.39,
+      "step": 86
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.9943258762359619,
+      "learning_rate": 7.799215215312667e-06,
+      "loss": 0.3736,
+      "step": 87
+    },
+    {
+      "epoch": 1.6340326340326339,
+      "grad_norm": 0.9510675668716431,
+      "learning_rate": 7.750275017224208e-06,
+      "loss": 0.4185,
+      "step": 88
+    },
+    {
+      "epoch": 1.6526806526806528,
+      "grad_norm": 0.9599855542182922,
+      "learning_rate": 7.700954061750295e-06,
+      "loss": 0.4017,
+      "step": 89
+    },
+    {
+      "epoch": 1.6713286713286712,
+      "grad_norm": 0.9646207094192505,
+      "learning_rate": 7.651259177050996e-06,
+      "loss": 0.3945,
+      "step": 90
+    },
+    {
+      "epoch": 1.68997668997669,
+      "grad_norm": 0.9255053400993347,
+      "learning_rate": 7.601197243054411e-06,
+      "loss": 0.3993,
+      "step": 91
+    },
+    {
+      "epoch": 1.7086247086247086,
+      "grad_norm": 0.9360446929931641,
+      "learning_rate": 7.5507751905041885e-06,
+      "loss": 0.3778,
+      "step": 92
+    },
+    {
+      "epoch": 1.7272727272727273,
+      "grad_norm": 0.9245442748069763,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.3958,
+      "step": 93
+    },
+    {
+      "epoch": 1.745920745920746,
+      "grad_norm": 0.9772454500198364,
+      "learning_rate": 7.4488787010311425e-06,
+      "loss": 0.3915,
+      "step": 94
+    },
+    {
+      "epoch": 1.7645687645687644,
+      "grad_norm": 0.8778578639030457,
+      "learning_rate": 7.3974183710033334e-06,
+      "loss": 0.401,
+      "step": 95
+    },
+    {
+      "epoch": 1.7832167832167833,
+      "grad_norm": 0.9100750684738159,
+      "learning_rate": 7.345626134258897e-06,
+      "loss": 0.3991,
+      "step": 96
+    },
+    {
+      "epoch": 1.8018648018648018,
+      "grad_norm": 0.9568976759910583,
+      "learning_rate": 7.293509161090453e-06,
+      "loss": 0.3738,
+      "step": 97
+    },
+    {
+      "epoch": 1.8205128205128205,
+      "grad_norm": 0.9305320382118225,
+      "learning_rate": 7.241074666748228e-06,
+      "loss": 0.4006,
+      "step": 98
+    },
+    {
+      "epoch": 1.8391608391608392,
+      "grad_norm": 0.9657385349273682,
+      "learning_rate": 7.188329910441154e-06,
+      "loss": 0.395,
+      "step": 99
+    },
+    {
+      "epoch": 1.8578088578088578,
+      "grad_norm": 0.9285731315612793,
+      "learning_rate": 7.135282194331881e-06,
+      "loss": 0.3822,
+      "step": 100
+    },
+    {
+      "epoch": 1.8764568764568765,
+      "grad_norm": 0.9422906041145325,
+      "learning_rate": 7.0819388625258385e-06,
+      "loss": 0.3991,
+      "step": 101
+    },
+    {
+      "epoch": 1.895104895104895,
+      "grad_norm": 0.9327917695045471,
+      "learning_rate": 7.028307300054499e-06,
+      "loss": 0.3854,
+      "step": 102
+    },
+    {
+      "epoch": 1.913752913752914,
+      "grad_norm": 0.9129431247711182,
+      "learning_rate": 6.974394931852957e-06,
+      "loss": 0.3897,
+      "step": 103
+    },
+    {
+      "epoch": 1.9324009324009324,
+      "grad_norm": 1.037550449371338,
+      "learning_rate": 6.920209221732007e-06,
+      "loss": 0.4183,
+      "step": 104
+    },
+    {
+      "epoch": 1.951048951048951,
+      "grad_norm": 0.969743549823761,
+      "learning_rate": 6.865757671344827e-06,
+      "loss": 0.4034,
+      "step": 105
+    },
+    {
+      "epoch": 1.9696969696969697,
+      "grad_norm": 0.9196340441703796,
+      "learning_rate": 6.811047819148413e-06,
+      "loss": 0.3844,
+      "step": 106
+    },
+    {
+      "epoch": 1.9883449883449882,
+      "grad_norm": 0.9086804389953613,
+      "learning_rate": 6.756087239359948e-06,
+      "loss": 0.4033,
+      "step": 107
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.103885531425476,
+      "learning_rate": 6.700883540908185e-06,
+      "loss": 0.362,
+      "step": 108
+    },
+    {
+      "epoch": 2.0186480186480185,
+      "grad_norm": 1.1509166955947876,
+      "learning_rate": 6.64544436638005e-06,
+      "loss": 0.3565,
+      "step": 109
+    },
+    {
+      "epoch": 2.0372960372960374,
+      "grad_norm": 1.0585626363754272,
+      "learning_rate": 6.589777390962575e-06,
+      "loss": 0.3591,
+      "step": 110
+    },
+    {
+      "epoch": 2.055944055944056,
+      "grad_norm": 1.006718397140503,
+      "learning_rate": 6.53389032138032e-06,
+      "loss": 0.3444,
+      "step": 111
+    },
+    {
+      "epoch": 2.0745920745920747,
+      "grad_norm": 1.0916168689727783,
+      "learning_rate": 6.477790894828422e-06,
+      "loss": 0.3576,
+      "step": 112
+    },
+    {
+      "epoch": 2.093240093240093,
+      "grad_norm": 0.9819603562355042,
+      "learning_rate": 6.421486877901436e-06,
+      "loss": 0.3127,
+      "step": 113
+    },
+    {
+      "epoch": 2.111888111888112,
+      "grad_norm": 0.8397829532623291,
+      "learning_rate": 6.364986065518106e-06,
+      "loss": 0.328,
+      "step": 114
+    },
+    {
+      "epoch": 2.1305361305361306,
+      "grad_norm": 0.9791805744171143,
+      "learning_rate": 6.308296279842204e-06,
+      "loss": 0.3349,
+      "step": 115
+    },
+    {
+      "epoch": 2.149184149184149,
+      "grad_norm": 1.006253957748413,
+      "learning_rate": 6.2514253691996e-06,
+      "loss": 0.3562,
+      "step": 116
+    },
+    {
+      "epoch": 2.167832167832168,
+      "grad_norm": 0.8682781457901001,
+      "learning_rate": 6.194381206991723e-06,
+      "loss": 0.3372,
+      "step": 117
+    },
+    {
+      "epoch": 2.1864801864801864,
+      "grad_norm": 0.9617028832435608,
+      "learning_rate": 6.1371716906055336e-06,
+      "loss": 0.3477,
+      "step": 118
+    },
+    {
+      "epoch": 2.2051282051282053,
+      "grad_norm": 1.0304099321365356,
+      "learning_rate": 6.079804740320181e-06,
+      "loss": 0.3587,
+      "step": 119
+    },
+    {
+      "epoch": 2.2237762237762237,
+      "grad_norm": 1.041025996208191,
+      "learning_rate": 6.022288298210502e-06,
+      "loss": 0.3306,
+      "step": 120
+    },
+    {
+      "epoch": 2.242424242424242,
+      "grad_norm": 0.9497987031936646,
+      "learning_rate": 5.964630327047485e-06,
+      "loss": 0.3547,
+      "step": 121
+    },
+    {
+      "epoch": 2.261072261072261,
+      "grad_norm": 0.8545536398887634,
+      "learning_rate": 5.906838809195879e-06,
+      "loss": 0.3485,
+      "step": 122
+    },
+    {
+      "epoch": 2.2797202797202796,
+      "grad_norm": 0.9179214835166931,
+      "learning_rate": 5.848921745509094e-06,
+      "loss": 0.3398,
+      "step": 123
+    },
+    {
+      "epoch": 2.2983682983682985,
+      "grad_norm": 1.0032718181610107,
+      "learning_rate": 5.790887154221521e-06,
+      "loss": 0.3732,
+      "step": 124
+    },
+    {
+      "epoch": 2.317016317016317,
+      "grad_norm": 0.9405228495597839,
+      "learning_rate": 5.7327430698384775e-06,
+      "loss": 0.3429,
+      "step": 125
+    },
+    {
+      "epoch": 2.335664335664336,
+      "grad_norm": 0.8877108693122864,
+      "learning_rate": 5.674497542023875e-06,
+      "loss": 0.3179,
+      "step": 126
+    },
+    {
+      "epoch": 2.3543123543123543,
+      "grad_norm": 0.912688672542572,
+      "learning_rate": 5.616158634485793e-06,
+      "loss": 0.3615,
+      "step": 127
+    },
+    {
+      "epoch": 2.3729603729603728,
+      "grad_norm": 0.9756841659545898,
+      "learning_rate": 5.557734423860122e-06,
+      "loss": 0.3476,
+      "step": 128
+    },
+    {
+      "epoch": 2.3916083916083917,
+      "grad_norm": 0.9159253239631653,
+      "learning_rate": 5.499232998592399e-06,
+      "loss": 0.3288,
+      "step": 129
+    },
+    {
+      "epoch": 2.41025641025641,
+      "grad_norm": 0.9156895279884338,
+      "learning_rate": 5.44066245781801e-06,
+      "loss": 0.3376,
+      "step": 130
+    },
+    {
+      "epoch": 2.428904428904429,
+      "grad_norm": 0.8652094006538391,
+      "learning_rate": 5.382030910240936e-06,
+      "loss": 0.3407,
+      "step": 131
+    },
+    {
+      "epoch": 2.4475524475524475,
+      "grad_norm": 0.9103012681007385,
+      "learning_rate": 5.3233464730111426e-06,
+      "loss": 0.3556,
+      "step": 132
+    },
+    {
+      "epoch": 2.4662004662004664,
+      "grad_norm": 0.908830463886261,
+      "learning_rate": 5.2646172706008154e-06,
+      "loss": 0.3662,
+      "step": 133
+    },
+    {
+      "epoch": 2.484848484848485,
+      "grad_norm": 0.9576376080513,
+      "learning_rate": 5.20585143367959e-06,
+      "loss": 0.3059,
+      "step": 134
+    },
+    {
+      "epoch": 2.5034965034965033,
+      "grad_norm": 0.9205601215362549,
+      "learning_rate": 5.147057097988898e-06,
+      "loss": 0.3764,
+      "step": 135
+    },
+    {
+      "epoch": 2.5221445221445222,
+      "grad_norm": 0.908481240272522,
+      "learning_rate": 5.088242403215644e-06,
+      "loss": 0.3423,
+      "step": 136
+    },
+    {
+      "epoch": 2.5407925407925407,
+      "grad_norm": 0.9143463373184204,
+      "learning_rate": 5.029415491865311e-06,
+      "loss": 0.3214,
+      "step": 137
+    },
+    {
+      "epoch": 2.5594405594405596,
+      "grad_norm": 0.8553743362426758,
+      "learning_rate": 4.97058450813469e-06,
+      "loss": 0.3548,
+      "step": 138
+    },
+    {
+      "epoch": 2.578088578088578,
+      "grad_norm": 0.8994094133377075,
+      "learning_rate": 4.911757596784358e-06,
+      "loss": 0.3312,
+      "step": 139
+    },
+    {
+      "epoch": 2.596736596736597,
+      "grad_norm": 0.9915500283241272,
+      "learning_rate": 4.8529429020111035e-06,
+      "loss": 0.35,
+      "step": 140
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.8941128849983215,
+      "learning_rate": 4.794148566320412e-06,
+      "loss": 0.3301,
+      "step": 141
+    },
+    {
+      "epoch": 2.634032634032634,
+      "grad_norm": 0.9031572341918945,
+      "learning_rate": 4.7353827293991845e-06,
+      "loss": 0.3517,
+      "step": 142
+    },
+    {
+      "epoch": 2.652680652680653,
+      "grad_norm": 0.9451303482055664,
+      "learning_rate": 4.676653526988858e-06,
+      "loss": 0.3308,
+      "step": 143
+    },
+    {
+      "epoch": 2.6713286713286712,
+      "grad_norm": 1.0032352209091187,
+      "learning_rate": 4.617969089759066e-06,
+      "loss": 0.3402,
+      "step": 144
+    },
+    {
+      "epoch": 2.6899766899766897,
+      "grad_norm": 0.9340781569480896,
+      "learning_rate": 4.559337542181993e-06,
+      "loss": 0.3365,
+      "step": 145
+    },
+    {
+      "epoch": 2.7086247086247086,
+      "grad_norm": 0.8977580070495605,
+      "learning_rate": 4.500767001407604e-06,
+      "loss": 0.3551,
+      "step": 146
+    },
+    {
+      "epoch": 2.7272727272727275,
+      "grad_norm": 0.9564929008483887,
+      "learning_rate": 4.4422655761398785e-06,
+      "loss": 0.3388,
+      "step": 147
+    },
+    {
+      "epoch": 2.745920745920746,
+      "grad_norm": 0.9190077185630798,
+      "learning_rate": 4.383841365514208e-06,
+      "loss": 0.3107,
+      "step": 148
+    },
+    {
+      "epoch": 2.7645687645687644,
+      "grad_norm": 0.9031611680984497,
+      "learning_rate": 4.325502457976126e-06,
+      "loss": 0.3262,
+      "step": 149
+    },
+    {
+      "epoch": 2.7832167832167833,
+      "grad_norm": 0.8627392649650574,
+      "learning_rate": 4.267256930161523e-06,
+      "loss": 0.334,
+      "step": 150
+    },
+    {
+      "epoch": 2.801864801864802,
+      "grad_norm": 0.9746893644332886,
+      "learning_rate": 4.209112845778481e-06,
+      "loss": 0.3201,
+      "step": 151
+    },
+    {
+      "epoch": 2.8205128205128203,
+      "grad_norm": 0.8949517011642456,
+      "learning_rate": 4.151078254490908e-06,
+      "loss": 0.3329,
+      "step": 152
+    },
+    {
+      "epoch": 2.839160839160839,
+      "grad_norm": 0.9285176992416382,
+      "learning_rate": 4.09316119080412e-06,
+      "loss": 0.3193,
+      "step": 153
+    },
+    {
+      "epoch": 2.857808857808858,
+      "grad_norm": 0.8917948603630066,
+      "learning_rate": 4.035369672952516e-06,
+      "loss": 0.3403,
+      "step": 154
+    },
+    {
+      "epoch": 2.8764568764568765,
+      "grad_norm": 0.8647080659866333,
+      "learning_rate": 3.977711701789499e-06,
+      "loss": 0.3415,
+      "step": 155
+    },
+    {
+      "epoch": 2.895104895104895,
+      "grad_norm": 0.9055953025817871,
+      "learning_rate": 3.920195259679822e-06,
+      "loss": 0.3399,
+      "step": 156
+    },
+    {
+      "epoch": 2.913752913752914,
+      "grad_norm": 0.8970502614974976,
+      "learning_rate": 3.862828309394469e-06,
+      "loss": 0.3518,
+      "step": 157
+    },
+    {
+      "epoch": 2.9324009324009324,
+      "grad_norm": 0.8647226691246033,
+      "learning_rate": 3.805618793008279e-06,
+      "loss": 0.3481,
+      "step": 158
+    },
+    {
+      "epoch": 2.951048951048951,
+      "grad_norm": 0.8951221108436584,
+      "learning_rate": 3.7485746308004013e-06,
+      "loss": 0.3267,
+      "step": 159
+    },
+    {
+      "epoch": 2.9696969696969697,
+      "grad_norm": 0.8731971979141235,
+      "learning_rate": 3.6917037201577977e-06,
+      "loss": 0.341,
+      "step": 160
+    },
+    {
+      "epoch": 2.988344988344988,
+      "grad_norm": 0.977042019367218,
+      "learning_rate": 3.635013934481895e-06,
+      "loss": 0.3723,
+      "step": 161
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.977042019367218,
+      "learning_rate": 3.578513122098566e-06,
+      "loss": 0.3482,
+      "step": 162
+    },
+    {
+      "epoch": 3.0186480186480185,
+      "grad_norm": 1.377078652381897,
+      "learning_rate": 3.5222091051715803e-06,
+      "loss": 0.3105,
+      "step": 163
+    },
+    {
+      "epoch": 3.0372960372960374,
+      "grad_norm": 0.8943535089492798,
+      "learning_rate": 3.466109678619681e-06,
+      "loss": 0.2782,
+      "step": 164
+    },
+    {
+      "epoch": 3.055944055944056,
+      "grad_norm": 0.9158867001533508,
+      "learning_rate": 3.4102226090374246e-06,
+      "loss": 0.3328,
+      "step": 165
+    },
+    {
+      "epoch": 3.0745920745920747,
+      "grad_norm": 0.9390725493431091,
+      "learning_rate": 3.35455563361995e-06,
+      "loss": 0.2927,
+      "step": 166
+    },
+    {
+      "epoch": 3.093240093240093,
+      "grad_norm": 0.9263814091682434,
+      "learning_rate": 3.2991164590918162e-06,
+      "loss": 0.3117,
+      "step": 167
+    },
+    {
+      "epoch": 3.111888111888112,
+      "grad_norm": 0.8674337267875671,
+      "learning_rate": 3.2439127606400546e-06,
+      "loss": 0.3022,
+      "step": 168
+    },
+    {
+      "epoch": 3.1305361305361306,
+      "grad_norm": 0.8507941365242004,
+      "learning_rate": 3.1889521808515888e-06,
+      "loss": 0.3395,
+      "step": 169
+    },
+    {
+      "epoch": 3.149184149184149,
+      "grad_norm": 0.9532745480537415,
+      "learning_rate": 3.1342423286551756e-06,
+      "loss": 0.3149,
+      "step": 170
+    },
+    {
+      "epoch": 3.167832167832168,
+      "grad_norm": 0.8734713792800903,
+      "learning_rate": 3.0797907782679944e-06,
+      "loss": 0.3105,
+      "step": 171
+    },
+    {
+      "epoch": 3.1864801864801864,
+      "grad_norm": 0.9466208815574646,
+      "learning_rate": 3.0256050681470446e-06,
+      "loss": 0.3472,
+      "step": 172
+    },
+    {
+      "epoch": 3.2051282051282053,
+      "grad_norm": 0.9010168313980103,
+      "learning_rate": 2.971692699945502e-06,
+      "loss": 0.3112,
+      "step": 173
+    },
+    {
+      "epoch": 3.2237762237762237,
+      "grad_norm": 0.8842659592628479,
+      "learning_rate": 2.9180611374741623e-06,
+      "loss": 0.3103,
+      "step": 174
+    },
+    {
+      "epoch": 3.242424242424242,
+      "grad_norm": 0.872593879699707,
+      "learning_rate": 2.8647178056681197e-06,
+      "loss": 0.3133,
+      "step": 175
+    },
+    {
+      "epoch": 3.261072261072261,
+      "grad_norm": 0.9134573340415955,
+      "learning_rate": 2.8116700895588473e-06,
+      "loss": 0.3046,
+      "step": 176
+    },
+    {
+      "epoch": 3.2797202797202796,
+      "grad_norm": 0.9120718240737915,
+      "learning_rate": 2.7589253332517736e-06,
+      "loss": 0.3136,
+      "step": 177
+    },
+    {
+      "epoch": 3.2983682983682985,
+      "grad_norm": 0.913482129573822,
+      "learning_rate": 2.706490838909547e-06,
+      "loss": 0.3047,
+      "step": 178
+    },
+    {
+      "epoch": 3.317016317016317,
+      "grad_norm": 0.8746190071105957,
+      "learning_rate": 2.6543738657411033e-06,
+      "loss": 0.3018,
+      "step": 179
+    },
+    {
+      "epoch": 3.335664335664336,
+      "grad_norm": 0.8129086494445801,
+      "learning_rate": 2.6025816289966703e-06,
+      "loss": 0.3319,
+      "step": 180
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 270,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 20,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.4536587564023808e+17,
+  "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:c5ace74e64028baa71aeaac444cae3a499b5b051f0bbc007ced55b209c59a5fa
+size 8145

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)