Upload initial model weights

added_tokens.json

@@ -0,0 +1,27 @@
+{
+  "</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,
+  "<|vggt_end|>": 151666,
+  "<|vggt_pad|>": 151667,
+  "<|vggt_start|>": 151665,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

chat_template.json

@@ -0,0 +1,3 @@
+{
+    "chat_template": "{% set image_count = namespace(value=0) %}{% set video_count = namespace(value=0) %}{% for message in messages %}{% if loop.first and message['role'] != 'system' %}<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n{% endif %}<|im_start|>{{ message['role'] }}\n{% if message['content'] is string %}{{ message['content'] }}<|im_end|>\n{% else %}{% for content in message['content'] %}{% if content['type'] == 'image' or 'image' in content or 'image_url' in content %}{% set image_count.value = image_count.value + 1 %}{% if add_vision_id %}Picture {{ image_count.value }}: {% endif %}<|vision_start|><|image_pad|><|vision_end|>{% elif content['type'] == 'video' or 'video' in content %}{% set video_count.value = video_count.value + 1 %}{% if add_vision_id %}Video {{ video_count.value }}: {% endif %}<|vision_start|><|video_pad|><|vision_end|>{% elif 'text' in content %}{{ content['text'] }}{% endif %}{% endfor %}<|im_end|>\n{% endif %}{% endfor %}{% if add_generation_prompt %}<|im_start|>assistant\n{% endif %}"
+}

config.json

@@ -0,0 +1,75 @@
+{
+  "architectures": [
+    "Qwen2_5_VLForConditionalGenerationWithVGGT"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151645,
+  "feature_fusion_method": "add",
+  "fusion_num_layers": 1,
+  "geometry_encoder_type": "vggt",
+  "geometry_merger_type": "mlp",
+  "hidden_act": "silu",
+  "hidden_size": 2048,
+  "image_token_id": 151655,
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "max_position_embeddings": 128000,
+  "max_window_layers": 70,
+  "model_type": "qwen2_5_vl",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 36,
+  "num_key_value_heads": 2,
+  "reference_frame": "first",
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": {
+    "mrope_section": [
+      16,
+      24,
+      24
+    ],
+    "rope_type": "default",
+    "type": "default"
+  },
+  "rope_theta": 1000000.0,
+  "sliding_window": 32768,
+  "stage": "stage4",
+  "tie_word_embeddings": true,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.50.0",
+  "use_cache": false,
+  "use_geometry_encoder": true,
+  "use_sliding_window": false,
+  "vggt_end_token_id": 151666,
+  "vggt_pad_token_id": 151667,
+  "vggt_start_token_id": 151665,
+  "video_token_id": 151656,
+  "vision_config": {
+    "depth": 32,
+    "fullatt_block_indexes": [
+      7,
+      15,
+      23,
+      31
+    ],
+    "hidden_act": "silu",
+    "hidden_size": 1280,
+    "in_channels": 3,
+    "in_chans": 3,
+    "intermediate_size": 3420,
+    "model_type": "qwen2_5_vl",
+    "num_heads": 16,
+    "out_hidden_size": 2048,
+    "patch_size": 14,
+    "spatial_merge_size": 2,
+    "spatial_patch_size": 14,
+    "temporal_patch_size": 2,
+    "tokens_per_second": 2,
+    "torch_dtype": "bfloat16",
+    "window_size": 112
+  },
+  "vision_end_token_id": 151653,
+  "vision_start_token_id": 151652,
+  "vision_token_id": 151654,
+  "vocab_size": 151668
+}

generation_config.json

@@ -0,0 +1,12 @@
+{
+  "bos_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "repetition_penalty": 1.05,
+  "temperature": 1e-06,
+  "transformers_version": "4.50.0"
+}

global_step4500/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:a316044a37df890d00a763bd341ebd9cfc404a74ed7e79fec1ae2962fae58379
+size 6328110698

global_step4500/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:d385d40cc7e06a95069505b2481e34620ef0a8b4498bb3e63eab1df24ed9e8af
+size 6328113706

global_step4500/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ec50fd9ec7f9a2168b280ff3cb065174e2a57e3ed44385c46cacb97c57b2cfaa
+size 6328113514

global_step4500/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9da3d68650409476bcc8cb9e5d0291409ac75463364ad28313d93377e3c06346
+size 6328113834

global_step4500/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bfc5cf90d0530e82847d038afeb88922a17e520a7e862d2268dc7d49def92248
+size 6328113770

global_step4500/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:29e1c8b9d93539667a77c2599fc032cee882be22d27d67656406863d89faaac5
+size 6328113642

global_step4500/mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:95ed6d3f2c6727d9176ef48be4ad6b1f5479b7fa98c198592bdb4ee9a8af0f4c
+size 12493934564

latest

@@ -0,0 +1 @@
+global_step4500

model-00001-of-00003.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:14746dbe6da6869de62af7b4d0da296b8bc8f22bef9c60a3b0f5c6319ca8bf43
+size 4958904568

model-00002-of-00003.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2d5da20a1b7b690b11034e8bf7f43edc96cb2e1e1c6e0a60474f5e95c528af36
+size 4451629296

model-00003-of-00003.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9464eb037ae10fdff94268378cf6f4134f9b6206db64e82639105ebd7db605c3
+size 621232256

preprocessor_config.json

@@ -0,0 +1,29 @@
+{
+  "do_convert_rgb": true,
+  "do_normalize": true,
+  "do_rescale": true,
+  "do_resize": true,
+  "image_mean": [
+    0.48145466,
+    0.4578275,
+    0.40821073
+  ],
+  "image_processor_type": "Qwen2VLImageProcessor",
+  "image_std": [
+    0.26862954,
+    0.26130258,
+    0.27577711
+  ],
+  "max_pixels": 451584,
+  "merge_size": 2,
+  "min_pixels": 12544,
+  "patch_size": 14,
+  "processor_class": "Qwen2_5_VLProcessor",
+  "resample": 3,
+  "rescale_factor": 0.00392156862745098,
+  "size": {
+    "longest_edge": 451584,
+    "shortest_edge": 12544
+  },
+  "temporal_patch_size": 2
+}

rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d877a82d7c2eac5101648c444cf545257799a6300049989f9293e5255bb5daed
+size 15472

rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b2c5c982def324178eef7b3b30cc2ec46b38e46a01a1ebe77ed1476a0f6150e8
+size 15472

rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:78c447a29db247d0c65aa3239179d602c8d24f407b99abe9eac23b6fa7c21304
+size 15472

rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:60307992e79e4a35a01da1869c4155030405fe6e06afc3ccbf361145ff0a183a
+size 15472

rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:101027e87495f3127f05e7dd10590bf5958808724213fb0ee9abcee1955eb59d
+size 15472

rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0e14a17a4ba5e47a3f767d3edeb93b01b4cc2eed66e38c016b15ff499bf4f497
+size 15472

scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:08c82649bedf9149c8c385f8a8acbae29140974d8a8182405f9988a738898cf4
+size 1064

special_tokens_map.json

@@ -0,0 +1,21 @@
+{
+  "additional_special_tokens": [
+    "<|vggt_start|>",
+    "<|vggt_end|>",
+    "<|vggt_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_config.json

@@ -0,0 +1,223 @@
+{
+  "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": "<|vggt_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151666": {
+      "content": "<|vggt_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151667": {
+      "content": "<|vggt_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<|vggt_start|>",
+    "<|vggt_end|>",
+    "<|vggt_pad|>"
+  ],
+  "bos_token": null,
+  "chat_template": "{% set image_count = namespace(value=0) %}{% set video_count = namespace(value=0) %}{% for message in messages %}{% if loop.first and message['role'] != 'system' %}<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n{% endif %}<|im_start|>{{ message['role'] }}\n{% if message['content'] is string %}{{ message['content'] }}<|im_end|>\n{% else %}{% for content in message['content'] %}{% if content['type'] == 'image' or 'image' in content or 'image_url' in content %}{% set image_count.value = image_count.value + 1 %}{% if add_vision_id %}Picture {{ image_count.value }}: {% endif %}<|vision_start|><|image_pad|><|vision_end|>{% elif content['type'] == 'video' or 'video' in content %}{% set video_count.value = video_count.value + 1 %}{% if add_vision_id %}Video {{ video_count.value }}: {% endif %}<|vision_start|><|video_pad|><|vision_end|>{% elif 'text' in content %}{{ content['text'] }}{% endif %}{% endfor %}<|im_end|>\n{% endif %}{% endfor %}{% if add_generation_prompt %}<|im_start|>assistant\n{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 20000,
+  "pad_token": "<|endoftext|>",
+  "padding_side": "right",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

trainer_state.json

@@ -0,0 +1,664 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.43873012004143563,
+  "eval_steps": 500,
+  "global_step": 4500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004874779111571507,
+      "grad_norm": 581.0810546875,
+      "learning_rate": 1.6233766233766232e-07,
+      "loss": 42.9948,
+      "step": 50
+    },
+    {
+      "epoch": 0.009749558223143015,
+      "grad_norm": 331.4380798339844,
+      "learning_rate": 3.2467532467532465e-07,
+      "loss": 36.4433,
+      "step": 100
+    },
+    {
+      "epoch": 0.014624337334714521,
+      "grad_norm": 399.9220886230469,
+      "learning_rate": 4.87012987012987e-07,
+      "loss": 33.9757,
+      "step": 150
+    },
+    {
+      "epoch": 0.01949911644628603,
+      "grad_norm": 227.4508514404297,
+      "learning_rate": 6.493506493506493e-07,
+      "loss": 24.5654,
+      "step": 200
+    },
+    {
+      "epoch": 0.024373895557857534,
+      "grad_norm": 169.74786376953125,
+      "learning_rate": 8.116883116883116e-07,
+      "loss": 18.221,
+      "step": 250
+    },
+    {
+      "epoch": 0.029248674669429042,
+      "grad_norm": 151.637451171875,
+      "learning_rate": 9.74025974025974e-07,
+      "loss": 15.1792,
+      "step": 300
+    },
+    {
+      "epoch": 0.03412345378100055,
+      "grad_norm": 140.70602416992188,
+      "learning_rate": 9.99956019474448e-07,
+      "loss": 13.3062,
+      "step": 350
+    },
+    {
+      "epoch": 0.03899823289257206,
+      "grad_norm": 163.72286987304688,
+      "learning_rate": 9.997889850109673e-07,
+      "loss": 12.1289,
+      "step": 400
+    },
+    {
+      "epoch": 0.04387301200414356,
+      "grad_norm": 205.9079132080078,
+      "learning_rate": 9.994973425669175e-07,
+      "loss": 11.2206,
+      "step": 450
+    },
+    {
+      "epoch": 0.04874779111571507,
+      "grad_norm": 190.47525024414062,
+      "learning_rate": 9.990811648549374e-07,
+      "loss": 10.3846,
+      "step": 500
+    },
+    {
+      "epoch": 0.05362257022728657,
+      "grad_norm": 63.44021224975586,
+      "learning_rate": 9.98540555636946e-07,
+      "loss": 10.0807,
+      "step": 550
+    },
+    {
+      "epoch": 0.058497349338858085,
+      "grad_norm": 134.91310119628906,
+      "learning_rate": 9.978756496982724e-07,
+      "loss": 9.2068,
+      "step": 600
+    },
+    {
+      "epoch": 0.0633721284504296,
+      "grad_norm": 125.61437225341797,
+      "learning_rate": 9.97086612814052e-07,
+      "loss": 8.8829,
+      "step": 650
+    },
+    {
+      "epoch": 0.0682469075620011,
+      "grad_norm": 148.3848876953125,
+      "learning_rate": 9.961736417078928e-07,
+      "loss": 8.8043,
+      "step": 700
+    },
+    {
+      "epoch": 0.0731216866735726,
+      "grad_norm": 360.50970458984375,
+      "learning_rate": 9.951369640028304e-07,
+      "loss": 9.8165,
+      "step": 750
+    },
+    {
+      "epoch": 0.07799646578514412,
+      "grad_norm": 89.09446716308594,
+      "learning_rate": 9.939768381645761e-07,
+      "loss": 8.9056,
+      "step": 800
+    },
+    {
+      "epoch": 0.08287124489671562,
+      "grad_norm": 163.72415161132812,
+      "learning_rate": 9.92693553437075e-07,
+      "loss": 9.5952,
+      "step": 850
+    },
+    {
+      "epoch": 0.08774602400828713,
+      "grad_norm": 147.0449981689453,
+      "learning_rate": 9.912874297703925e-07,
+      "loss": 9.0044,
+      "step": 900
+    },
+    {
+      "epoch": 0.09262080311985862,
+      "grad_norm": 171.43394470214844,
+      "learning_rate": 9.897588177409434e-07,
+      "loss": 9.1141,
+      "step": 950
+    },
+    {
+      "epoch": 0.09749558223143014,
+      "grad_norm": 206.17628479003906,
+      "learning_rate": 9.88108098464086e-07,
+      "loss": 8.1566,
+      "step": 1000
+    },
+    {
+      "epoch": 0.10237036134300165,
+      "grad_norm": 131.0735321044922,
+      "learning_rate": 9.863356834991016e-07,
+      "loss": 8.4912,
+      "step": 1050
+    },
+    {
+      "epoch": 0.10724514045457315,
+      "grad_norm": 140.00730895996094,
+      "learning_rate": 9.844420147465848e-07,
+      "loss": 8.1491,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11211991956614466,
+      "grad_norm": 163.5810089111328,
+      "learning_rate": 9.824275643382676e-07,
+      "loss": 8.6904,
+      "step": 1150
+    },
+    {
+      "epoch": 0.11699469867771617,
+      "grad_norm": 129.7275848388672,
+      "learning_rate": 9.802928345193068e-07,
+      "loss": 8.1686,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12186947778928767,
+      "grad_norm": 115.78919219970703,
+      "learning_rate": 9.780383575230648e-07,
+      "loss": 7.6378,
+      "step": 1250
+    },
+    {
+      "epoch": 0.1267442569008592,
+      "grad_norm": 74.11811828613281,
+      "learning_rate": 9.756646954384115e-07,
+      "loss": 7.8103,
+      "step": 1300
+    },
+    {
+      "epoch": 0.1316190360124307,
+      "grad_norm": 93.70452880859375,
+      "learning_rate": 9.731724400695836e-07,
+      "loss": 8.122,
+      "step": 1350
+    },
+    {
+      "epoch": 0.1364938151240022,
+      "grad_norm": 110.20355987548828,
+      "learning_rate": 9.70562212788636e-07,
+      "loss": 7.8767,
+      "step": 1400
+    },
+    {
+      "epoch": 0.1413685942355737,
+      "grad_norm": 96.42841339111328,
+      "learning_rate": 9.6783466438052e-07,
+      "loss": 8.0516,
+      "step": 1450
+    },
+    {
+      "epoch": 0.1462433733471452,
+      "grad_norm": 98.32221221923828,
+      "learning_rate": 9.649904748808292e-07,
+      "loss": 7.6941,
+      "step": 1500
+    },
+    {
+      "epoch": 0.1511181524587167,
+      "grad_norm": 107.42027282714844,
+      "learning_rate": 9.620303534062518e-07,
+      "loss": 8.0057,
+      "step": 1550
+    },
+    {
+      "epoch": 0.15599293157028823,
+      "grad_norm": 55.05694580078125,
+      "learning_rate": 9.589550379777732e-07,
+      "loss": 7.4756,
+      "step": 1600
+    },
+    {
+      "epoch": 0.16086771068185973,
+      "grad_norm": 117.27649688720703,
+      "learning_rate": 9.557652953366717e-07,
+      "loss": 6.8833,
+      "step": 1650
+    },
+    {
+      "epoch": 0.16574248979343123,
+      "grad_norm": 174.34263610839844,
+      "learning_rate": 9.52461920753353e-07,
+      "loss": 7.4795,
+      "step": 1700
+    },
+    {
+      "epoch": 0.17061726890500276,
+      "grad_norm": 119.58318328857422,
+      "learning_rate": 9.490457378290737e-07,
+      "loss": 7.7871,
+      "step": 1750
+    },
+    {
+      "epoch": 0.17549204801657425,
+      "grad_norm": 142.45582580566406,
+      "learning_rate": 9.455175982905988e-07,
+      "loss": 8.1505,
+      "step": 1800
+    },
+    {
+      "epoch": 0.18036682712814575,
+      "grad_norm": 122.0265884399414,
+      "learning_rate": 9.418783817778484e-07,
+      "loss": 7.6914,
+      "step": 1850
+    },
+    {
+      "epoch": 0.18524160623971725,
+      "grad_norm": 96.58927917480469,
+      "learning_rate": 9.381289956245861e-07,
+      "loss": 7.5846,
+      "step": 1900
+    },
+    {
+      "epoch": 0.19011638535128877,
+      "grad_norm": 238.81964111328125,
+      "learning_rate": 9.342703746321997e-07,
+      "loss": 7.7886,
+      "step": 1950
+    },
+    {
+      "epoch": 0.19499116446286027,
+      "grad_norm": 61.6027946472168,
+      "learning_rate": 9.303034808366366e-07,
+      "loss": 7.2491,
+      "step": 2000
+    },
+    {
+      "epoch": 0.19986594357443177,
+      "grad_norm": 96.19196319580078,
+      "learning_rate": 9.262293032685475e-07,
+      "loss": 6.8776,
+      "step": 2050
+    },
+    {
+      "epoch": 0.2047407226860033,
+      "grad_norm": 72.44068908691406,
+      "learning_rate": 9.220488577066996e-07,
+      "loss": 7.2714,
+      "step": 2100
+    },
+    {
+      "epoch": 0.2096155017975748,
+      "grad_norm": 160.9955291748047,
+      "learning_rate": 9.177631864247226e-07,
+      "loss": 7.4344,
+      "step": 2150
+    },
+    {
+      "epoch": 0.2144902809091463,
+      "grad_norm": 78.55003356933594,
+      "learning_rate": 9.133733579312468e-07,
+      "loss": 7.2211,
+      "step": 2200
+    },
+    {
+      "epoch": 0.21936506002071782,
+      "grad_norm": 141.0493621826172,
+      "learning_rate": 9.088804667035016e-07,
+      "loss": 7.3533,
+      "step": 2250
+    },
+    {
+      "epoch": 0.22423983913228931,
+      "grad_norm": 106.55406951904297,
+      "learning_rate": 9.042856329144392e-07,
+      "loss": 7.526,
+      "step": 2300
+    },
+    {
+      "epoch": 0.2291146182438608,
+      "grad_norm": 116.47844696044922,
+      "learning_rate": 8.995900021534517e-07,
+      "loss": 6.5839,
+      "step": 2350
+    },
+    {
+      "epoch": 0.23398939735543234,
+      "grad_norm": 87.43718719482422,
+      "learning_rate": 8.947947451407512e-07,
+      "loss": 7.2284,
+      "step": 2400
+    },
+    {
+      "epoch": 0.23886417646700384,
+      "grad_norm": 135.8431854248047,
+      "learning_rate": 8.89901057435485e-07,
+      "loss": 7.6484,
+      "step": 2450
+    },
+    {
+      "epoch": 0.24373895557857533,
+      "grad_norm": 87.54926300048828,
+      "learning_rate": 8.849101591376568e-07,
+      "loss": 7.2991,
+      "step": 2500
+    },
+    {
+      "epoch": 0.24861373469014686,
+      "grad_norm": 109.77268981933594,
+      "learning_rate": 8.798232945839304e-07,
+      "loss": 6.9895,
+      "step": 2550
+    },
+    {
+      "epoch": 0.2534885138017184,
+      "grad_norm": 118.82627868652344,
+      "learning_rate": 8.746417320373896e-07,
+      "loss": 7.4786,
+      "step": 2600
+    },
+    {
+      "epoch": 0.25836329291328985,
+      "grad_norm": 175.50918579101562,
+      "learning_rate": 8.693667633713338e-07,
+      "loss": 6.6877,
+      "step": 2650
+    },
+    {
+      "epoch": 0.2632380720248614,
+      "grad_norm": 190.9921875,
+      "learning_rate": 8.639997037471867e-07,
+      "loss": 6.8118,
+      "step": 2700
+    },
+    {
+      "epoch": 0.2681128511364329,
+      "grad_norm": 163.4025115966797,
+      "learning_rate": 8.585418912865986e-07,
+      "loss": 6.9759,
+      "step": 2750
+    },
+    {
+      "epoch": 0.2729876302480044,
+      "grad_norm": 149.16815185546875,
+      "learning_rate": 8.529946867378241e-07,
+      "loss": 7.2147,
+      "step": 2800
+    },
+    {
+      "epoch": 0.2778624093595759,
+      "grad_norm": 97.56202697753906,
+      "learning_rate": 8.473594731364587e-07,
+      "loss": 7.1163,
+      "step": 2850
+    },
+    {
+      "epoch": 0.2827371884711474,
+      "grad_norm": 104.49602508544922,
+      "learning_rate": 8.416376554606195e-07,
+      "loss": 7.5656,
+      "step": 2900
+    },
+    {
+      "epoch": 0.2876119675827189,
+      "grad_norm": 130.78414916992188,
+      "learning_rate": 8.358306602806534e-07,
+      "loss": 6.901,
+      "step": 2950
+    },
+    {
+      "epoch": 0.2924867466942904,
+      "grad_norm": 152.71543884277344,
+      "learning_rate": 8.299399354034633e-07,
+      "loss": 7.0534,
+      "step": 3000
+    },
+    {
+      "epoch": 0.29736152580586195,
+      "grad_norm": 128.23233032226562,
+      "learning_rate": 8.239669495115393e-07,
+      "loss": 7.2949,
+      "step": 3050
+    },
+    {
+      "epoch": 0.3022363049174334,
+      "grad_norm": 212.286865234375,
+      "learning_rate": 8.179131917967852e-07,
+      "loss": 7.1819,
+      "step": 3100
+    },
+    {
+      "epoch": 0.30711108402900494,
+      "grad_norm": 146.38832092285156,
+      "learning_rate": 8.117801715892306e-07,
+      "loss": 7.3945,
+      "step": 3150
+    },
+    {
+      "epoch": 0.31198586314057647,
+      "grad_norm": 98.95829010009766,
+      "learning_rate": 8.05569417980724e-07,
+      "loss": 7.0111,
+      "step": 3200
+    },
+    {
+      "epoch": 0.31686064225214794,
+      "grad_norm": 147.78128051757812,
+      "learning_rate": 7.992824794436971e-07,
+      "loss": 7.1754,
+      "step": 3250
+    },
+    {
+      "epoch": 0.32173542136371946,
+      "grad_norm": 137.8080596923828,
+      "learning_rate": 7.92920923445098e-07,
+      "loss": 7.2801,
+      "step": 3300
+    },
+    {
+      "epoch": 0.326610200475291,
+      "grad_norm": 156.37132263183594,
+      "learning_rate": 7.864863360555886e-07,
+      "loss": 7.1625,
+      "step": 3350
+    },
+    {
+      "epoch": 0.33148497958686246,
+      "grad_norm": 176.6288299560547,
+      "learning_rate": 7.799803215541036e-07,
+      "loss": 7.5386,
+      "step": 3400
+    },
+    {
+      "epoch": 0.336359758698434,
+      "grad_norm": 150.56468200683594,
+      "learning_rate": 7.734045020278694e-07,
+      "loss": 6.9751,
+      "step": 3450
+    },
+    {
+      "epoch": 0.3412345378100055,
+      "grad_norm": 138.31723022460938,
+      "learning_rate": 7.667605169679842e-07,
+      "loss": 6.8245,
+      "step": 3500
+    },
+    {
+      "epoch": 0.346109316921577,
+      "grad_norm": 100.9895248413086,
+      "learning_rate": 7.600500228606573e-07,
+      "loss": 6.947,
+      "step": 3550
+    },
+    {
+      "epoch": 0.3509840960331485,
+      "grad_norm": 142.1031036376953,
+      "learning_rate": 7.532746927742119e-07,
+      "loss": 6.9751,
+      "step": 3600
+    },
+    {
+      "epoch": 0.35585887514472,
+      "grad_norm": 182.1207275390625,
+      "learning_rate": 7.464362159419551e-07,
+      "loss": 7.1473,
+      "step": 3650
+    },
+    {
+      "epoch": 0.3607336542562915,
+      "grad_norm": 162.8542938232422,
+      "learning_rate": 7.395362973410145e-07,
+      "loss": 7.7815,
+      "step": 3700
+    },
+    {
+      "epoch": 0.36560843336786303,
+      "grad_norm": 168.93276977539062,
+      "learning_rate": 7.325766572672528e-07,
+      "loss": 7.7646,
+      "step": 3750
+    },
+    {
+      "epoch": 0.3704832124794345,
+      "grad_norm": 159.7053985595703,
+      "learning_rate": 7.255590309063604e-07,
+      "loss": 6.4885,
+      "step": 3800
+    },
+    {
+      "epoch": 0.375357991591006,
+      "grad_norm": 96.07608795166016,
+      "learning_rate": 7.184851679012374e-07,
+      "loss": 6.9556,
+      "step": 3850
+    },
+    {
+      "epoch": 0.38023277070257755,
+      "grad_norm": 158.5734100341797,
+      "learning_rate": 7.113568319157707e-07,
+      "loss": 6.9754,
+      "step": 3900
+    },
+    {
+      "epoch": 0.385107549814149,
+      "grad_norm": 151.5749969482422,
+      "learning_rate": 7.041758001951149e-07,
+      "loss": 6.6478,
+      "step": 3950
+    },
+    {
+      "epoch": 0.38998232892572054,
+      "grad_norm": 118.84528350830078,
+      "learning_rate": 6.969438631225877e-07,
+      "loss": 6.3464,
+      "step": 4000
+    },
+    {
+      "epoch": 0.39485710803729207,
+      "grad_norm": 130.48410034179688,
+      "learning_rate": 6.896628237732894e-07,
+      "loss": 7.0122,
+      "step": 4050
+    },
+    {
+      "epoch": 0.39973188714886354,
+      "grad_norm": 92.12020874023438,
+      "learning_rate": 6.823344974645576e-07,
+      "loss": 7.2089,
+      "step": 4100
+    },
+    {
+      "epoch": 0.40460666626043507,
+      "grad_norm": 113.18313598632812,
+      "learning_rate": 6.749607113033709e-07,
+      "loss": 7.2546,
+      "step": 4150
+    },
+    {
+      "epoch": 0.4094814453720066,
+      "grad_norm": 106.96129608154297,
+      "learning_rate": 6.675433037308119e-07,
+      "loss": 7.3078,
+      "step": 4200
+    },
+    {
+      "epoch": 0.41435622448357806,
+      "grad_norm": 150.419677734375,
+      "learning_rate": 6.600841240637052e-07,
+      "loss": 7.1537,
+      "step": 4250
+    },
+    {
+      "epoch": 0.4192310035951496,
+      "grad_norm": 88.62400817871094,
+      "learning_rate": 6.525850320335433e-07,
+      "loss": 7.0714,
+      "step": 4300
+    },
+    {
+      "epoch": 0.4241057827067211,
+      "grad_norm": 83.91474151611328,
+      "learning_rate": 6.450478973228162e-07,
+      "loss": 6.9181,
+      "step": 4350
+    },
+    {
+      "epoch": 0.4289805618182926,
+      "grad_norm": 135.38629150390625,
+      "learning_rate": 6.374745990988598e-07,
+      "loss": 7.1421,
+      "step": 4400
+    },
+    {
+      "epoch": 0.4338553409298641,
+      "grad_norm": 101.2317123413086,
+      "learning_rate": 6.298670255453404e-07,
+      "loss": 6.6926,
+      "step": 4450
+    },
+    {
+      "epoch": 0.43873012004143563,
+      "grad_norm": 112.95464324951172,
+      "learning_rate": 6.222270733914895e-07,
+      "loss": 6.7252,
+      "step": 4500
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 10256,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.8045681436358345e+19,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:554281b6e557f24360d9bf0a8018b0566c180166de1d97947e7c8d2e1fb17277
+size 7288

zero_to_fp32.py

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