Upload 13 files

added_tokens.json

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

config.json

@@ -0,0 +1,105 @@
+{
+  "architectures": [
+    "TinyLlavaForConditionalGeneration"
+  ],
+  "cache_dir": null,
+  "connector_type": "mlp2x_gelu",
+  "hidden_size": 1536,
+  "ignore_index": -100,
+  "image_aspect_ratio": "square",
+  "image_token_index": -200,
+  "llm_model_name_or_path": "Qwen/Qwen2.5-1.5B",
+  "model_type": "tinyllava",
+  "num_queries": 128,
+  "num_resampler_layers": 3,
+  "pad_token": "<|endoftext|>",
+  "pad_token_id": 151643,
+  "resampler_hidden_size": 768,
+  "text_config": {
+    "_name_or_path": "Qwen/Qwen2.5-1.5B",
+    "architectures": [
+      "Qwen2ForCausalLM"
+    ],
+    "bos_token_id": 151643,
+    "eos_token_id": 151643,
+    "hidden_size": 1536,
+    "intermediate_size": 8960,
+    "max_position_embeddings": 131072,
+    "max_window_layers": 28,
+    "model_type": "qwen2",
+    "num_attention_heads": 12,
+    "num_hidden_layers": 28,
+    "num_key_value_heads": 2,
+    "rope_theta": 1000000.0,
+    "sliding_window": 131072,
+    "tie_word_embeddings": true,
+    "torch_dtype": "float16",
+    "use_mrope": false,
+    "use_sliding_window": false,
+    "vocab_size": 151936
+  },
+  "tokenizer_model_max_length": 2048,
+  "tokenizer_name_or_path": "Qwen/Qwen2.5-1.5B",
+  "tokenizer_padding_side": "right",
+  "tokenizer_use_fast": false,
+  "torch_dtype": "float16",
+  "transformers_version": "4.39.3",
+  "tune_type_connector": "full",
+  "tune_type_llm": "frozen",
+  "tune_type_vision_tower": "frozen",
+  "tune_vision_tower_from_layer": 0,
+  "use_cache": false,
+  "vision_config": {
+    "_name_or_path": "facebook/dinov2-small",
+    "apply_layernorm": true,
+    "architectures": [
+      "Dinov2Model"
+    ],
+    "attention_probs_dropout_prob": 0.0,
+    "drop_path_rate": 0.0,
+    "hidden_act": "gelu",
+    "hidden_dropout_prob": 0.0,
+    "hidden_size": 384,
+    "image_size": 518,
+    "layer_norm_eps": 1e-06,
+    "layerscale_value": 1.0,
+    "mlp_ratio": 4,
+    "model_name_or_path": "facebook/dinov2-small",
+    "model_name_or_path2": "",
+    "model_type": "dinov2",
+    "num_attention_heads": 6,
+    "num_hidden_layers": 12,
+    "out_features": [
+      "stage12"
+    ],
+    "out_indices": [
+      12
+    ],
+    "patch_size": 14,
+    "qkv_bias": true,
+    "reshape_hidden_states": true,
+    "stage_names": [
+      "stem",
+      "stage1",
+      "stage2",
+      "stage3",
+      "stage4",
+      "stage5",
+      "stage6",
+      "stage7",
+      "stage8",
+      "stage9",
+      "stage10",
+      "stage11",
+      "stage12"
+    ],
+    "torch_dtype": "float32",
+    "use_swiglu_ffn": false
+  },
+  "vision_feature_layer": -2,
+  "vision_feature_select_strategy": "patch",
+  "vision_hidden_size": 384,
+  "vision_model_name_or_path": "facebook/dinov2-small",
+  "vision_model_name_or_path2": "",
+  "vocab_size": 151936
+}

generation_config.json

@@ -0,0 +1,7 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 151643,
+  "eos_token_id": 151643,
+  "transformers_version": "4.39.3",
+  "use_cache": false
+}

latest

@@ -0,0 +1 @@
+global_step80

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8d87d1a32bf7179a3b738b3a5c3ec3d66caa41014f9b08a0dbcbf726dd209315
+size 3137520392

rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:835f869ea325fd6edf27b48b589309fb66641cb92b45f2fc13d1bb6e8814106c
+size 14244

scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fa0d8c142fdbc61db482a38a31179e68dec0153118296e3d5a05758374076a8
+size 1064

special_tokens_map.json

@@ -0,0 +1,32 @@
+{
+  "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": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": "<|endoftext|>"
+}

tokenizer_config.json

@@ -0,0 +1,208 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "chat_template": "{%- if tools %}\n    {{- '<|im_start|>system\\n' }}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- messages[0]['content'] }}\n    {%- else %}\n        {{- 'You are a helpful assistant.' }}\n    {%- endif %}\n    {{- \"\\n\\n# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>\" }}\n    {%- for tool in tools %}\n        {{- \"\\n\" }}\n        {{- tool | tojson }}\n    {%- endfor %}\n    {{- \"\\n</tools>\\n\\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\\n<tool_call>\\n{\\\"name\\\": <function-name>, \\\"arguments\\\": <args-json-object>}\\n</tool_call><|im_end|>\\n\" }}\n{%- else %}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- '<|im_start|>system\\n' + messages[0]['content'] + '<|im_end|>\\n' }}\n    {%- else %}\n        {{- '<|im_start|>system\\nYou are a helpful assistant.<|im_end|>\\n' }}\n    {%- endif %}\n{%- endif %}\n{%- for message in messages %}\n    {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) or (message.role == \"assistant\" and not message.tool_calls) %}\n        {{- '<|im_start|>' + message.role + '\\n' + message.content + '<|im_end|>' + '\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {{- '<|im_start|>' + message.role }}\n        {%- if message.content %}\n            {{- '\\n' + message.content }}\n        {%- endif %}\n        {%- for tool_call in message.tool_calls %}\n            {%- if tool_call.function is defined %}\n                {%- set tool_call = tool_call.function %}\n            {%- endif %}\n            {{- '\\n<tool_call>\\n{\"name\": \"' }}\n            {{- tool_call.name }}\n            {{- '\", \"arguments\": ' }}\n            {{- tool_call.arguments | tojson }}\n            {{- '}\\n</tool_call>' }}\n        {%- endfor %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"tool\" %}\n        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != \"tool\") %}\n            {{- '<|im_start|>user' }}\n        {%- endif %}\n        {{- '\\n<tool_response>\\n' }}\n        {{- message.content }}\n        {{- '\\n</tool_response>' }}\n        {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n            {{- '<|im_end|>\\n' }}\n        {%- endif %}\n    {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n    {{- '<|im_start|>assistant\\n' }}\n{%- endif %}\n",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "model_max_length": 2048,
+  "pad_token": "<|endoftext|>",
+  "padding_side": "right",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": "<|endoftext|>"
+}

trainer_state.json

@@ -0,0 +1,581 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.009173259947253756,
+  "eval_steps": 500,
+  "global_step": 80,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.0,
+      "learning_rate": 0.0,
+      "loss": 5.8857,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.0,
+      "learning_rate": 0.0,
+      "loss": 5.8613,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.0,
+      "learning_rate": 0.0,
+      "loss": 5.8757,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 22.244311559060773,
+      "learning_rate": 3.816793893129771e-06,
+      "loss": 6.0653,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 15.057639900452392,
+      "learning_rate": 7.633587786259541e-06,
+      "loss": 5.9572,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 38.092887646390466,
+      "learning_rate": 1.1450381679389314e-05,
+      "loss": 5.7617,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 15.31117053293346,
+      "learning_rate": 1.5267175572519083e-05,
+      "loss": 5.8945,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 21.702345688707492,
+      "learning_rate": 1.9083969465648855e-05,
+      "loss": 5.9337,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 21.702345688707492,
+      "learning_rate": 1.9083969465648855e-05,
+      "loss": 5.8771,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 26.25713504824,
+      "learning_rate": 2.2900763358778628e-05,
+      "loss": 5.879,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 88.79206625984197,
+      "learning_rate": 2.6717557251908397e-05,
+      "loss": 5.6244,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 13.979462978780887,
+      "learning_rate": 3.0534351145038166e-05,
+      "loss": 5.7982,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 13.979462978780887,
+      "learning_rate": 3.0534351145038166e-05,
+      "loss": 5.8216,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 58.311502538831355,
+      "learning_rate": 3.435114503816794e-05,
+      "loss": 5.7539,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 20.71528208128746,
+      "learning_rate": 3.816793893129771e-05,
+      "loss": 5.7114,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 33.69759468630642,
+      "learning_rate": 4.198473282442748e-05,
+      "loss": 5.6904,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 17.28319140833229,
+      "learning_rate": 4.5801526717557256e-05,
+      "loss": 5.6718,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 44.39956377017333,
+      "learning_rate": 4.9618320610687025e-05,
+      "loss": 5.4446,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 91.68382396043916,
+      "learning_rate": 5.3435114503816794e-05,
+      "loss": 5.6799,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 91.68382396043916,
+      "learning_rate": 5.3435114503816794e-05,
+      "loss": 5.606,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 192.72226388224067,
+      "learning_rate": 5.725190839694656e-05,
+      "loss": 5.5796,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 45.566936397354795,
+      "learning_rate": 6.106870229007633e-05,
+      "loss": 5.6014,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 54.111992406676734,
+      "learning_rate": 6.488549618320611e-05,
+      "loss": 5.4642,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 47.77738772861109,
+      "learning_rate": 6.870229007633588e-05,
+      "loss": 5.3002,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 52.88816210898902,
+      "learning_rate": 7.251908396946565e-05,
+      "loss": 5.8573,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 43.38379795566033,
+      "learning_rate": 7.633587786259542e-05,
+      "loss": 5.5942,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 112.9679807995391,
+      "learning_rate": 8.015267175572518e-05,
+      "loss": 5.7442,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 112.9679807995391,
+      "learning_rate": 8.015267175572518e-05,
+      "loss": 5.447,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 164.11451980650267,
+      "learning_rate": 8.396946564885496e-05,
+      "loss": 5.7154,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 115.9334173332188,
+      "learning_rate": 8.778625954198472e-05,
+      "loss": 5.9088,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 145.9836977981191,
+      "learning_rate": 9.160305343511451e-05,
+      "loss": 5.4605,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 114.64052697776405,
+      "learning_rate": 9.541984732824429e-05,
+      "loss": 5.697,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 202.12636675775389,
+      "learning_rate": 9.923664122137405e-05,
+      "loss": 6.0274,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 160.88000887426793,
+      "learning_rate": 0.00010305343511450383,
+      "loss": 6.2896,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 145.16182847186317,
+      "learning_rate": 0.00010687022900763359,
+      "loss": 5.9883,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 104.5781091944148,
+      "learning_rate": 0.00011068702290076336,
+      "loss": 6.1505,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 55.72279835011099,
+      "learning_rate": 0.00011450381679389313,
+      "loss": 6.458,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 72.60539121615658,
+      "learning_rate": 0.0001183206106870229,
+      "loss": 6.4766,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 152.31919342671264,
+      "learning_rate": 0.00012213740458015266,
+      "loss": 6.6228,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 195.38778604806365,
+      "learning_rate": 0.00012595419847328244,
+      "loss": 6.5874,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 98.21218214875543,
+      "learning_rate": 0.00012977099236641222,
+      "loss": 6.2979,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 117.40378533203793,
+      "learning_rate": 0.000133587786259542,
+      "loss": 6.1422,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 76.43242080692808,
+      "learning_rate": 0.00013740458015267177,
+      "loss": 6.0982,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 161.5295826913437,
+      "learning_rate": 0.00014122137404580154,
+      "loss": 5.9792,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 54.30211860707633,
+      "learning_rate": 0.0001450381679389313,
+      "loss": 6.0895,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 96.35953226922737,
+      "learning_rate": 0.00014885496183206107,
+      "loss": 6.1023,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 49.71381292121367,
+      "learning_rate": 0.00015267175572519084,
+      "loss": 5.9927,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 92.40570872689418,
+      "learning_rate": 0.00015648854961832062,
+      "loss": 5.8947,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 70.58634543270558,
+      "learning_rate": 0.00016030534351145037,
+      "loss": 5.5419,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 99.21861402306824,
+      "learning_rate": 0.00016412213740458014,
+      "loss": 5.533,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 60.43737769128788,
+      "learning_rate": 0.00016793893129770992,
+      "loss": 5.7271,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 40.38259047816709,
+      "learning_rate": 0.0001717557251908397,
+      "loss": 5.7707,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 50.37624352755525,
+      "learning_rate": 0.00017557251908396944,
+      "loss": 5.4807,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 105.31786701509579,
+      "learning_rate": 0.00017938931297709925,
+      "loss": 5.5782,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 58.697213953188964,
+      "learning_rate": 0.00018320610687022902,
+      "loss": 5.5337,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 110.55644774315732,
+      "learning_rate": 0.0001870229007633588,
+      "loss": 5.5836,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 14.426822607818815,
+      "learning_rate": 0.00019083969465648857,
+      "loss": 5.6155,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 26.228166827626183,
+      "learning_rate": 0.00019465648854961832,
+      "loss": 5.6817,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 27.269174056089717,
+      "learning_rate": 0.0001984732824427481,
+      "loss": 5.2749,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 21.695242218914665,
+      "learning_rate": 0.00020229007633587788,
+      "loss": 5.5323,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 68.62936938874972,
+      "learning_rate": 0.00020610687022900765,
+      "loss": 5.804,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 25.97127488754509,
+      "learning_rate": 0.0002099236641221374,
+      "loss": 5.4587,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 15.325961009638357,
+      "learning_rate": 0.00021374045801526718,
+      "loss": 5.4342,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 19.875772589083574,
+      "learning_rate": 0.00021755725190839695,
+      "loss": 5.5249,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 69.24118271312939,
+      "learning_rate": 0.00022137404580152673,
+      "loss": 5.1242,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 26.355890547603202,
+      "learning_rate": 0.00022519083969465648,
+      "loss": 5.5468,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 10.563329361046026,
+      "learning_rate": 0.00022900763358778625,
+      "loss": 5.2759,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 91.00091143398366,
+      "learning_rate": 0.00023282442748091603,
+      "loss": 5.4683,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 31.924406772853743,
+      "learning_rate": 0.0002366412213740458,
+      "loss": 5.4741,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 59.162721435471,
+      "learning_rate": 0.00024045801526717558,
+      "loss": 5.2497,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 31.455685925896024,
+      "learning_rate": 0.00024427480916030533,
+      "loss": 5.3254,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 67.5878959609375,
+      "learning_rate": 0.00024809160305343513,
+      "loss": 5.4352,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 42.716427641408124,
+      "learning_rate": 0.0002519083969465649,
+      "loss": 5.1745,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 51.763664942049346,
+      "learning_rate": 0.00025572519083969463,
+      "loss": 5.3703,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 45.70575367441054,
+      "learning_rate": 0.00025954198473282443,
+      "loss": 5.3987,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 22.180135968343524,
+      "learning_rate": 0.0002633587786259542,
+      "loss": 5.4224,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 60.31973560441679,
+      "learning_rate": 0.000267175572519084,
+      "loss": 5.6587,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 33.89974772165083,
+      "learning_rate": 0.00027099236641221373,
+      "loss": 5.6768,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 34.7175502203455,
+      "learning_rate": 0.00027480916030534353,
+      "loss": 5.4154,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 34.357468729191154,
+      "learning_rate": 0.0002786259541984733,
+      "loss": 5.4172,
+      "step": 80
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 8721,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 10,
+  "total_flos": 419593936896.0,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}

zero_to_fp32.py

@@ -0,0 +1,604 @@
+#!/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 . pytorch_model.bin
+
+import argparse
+import torch
+import glob
+import math
+import os
+import re
+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 param_shapes.items():
+
+        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_file, 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_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
+        - ``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
+    """
+
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters)
+    print(f"Saving fp32 state dict to {output_file}")
+    torch.save(state_dict, output_file)
+
+
+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_file",
+        type=str,
+        help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
+    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_file,
+                                               tag=args.tag,
+                                               exclude_frozen_parameters=args.exclude_frozen_parameters)