Upload folder using huggingface_hub

.gitattributes

@@ -57,3 +57,7 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 26_03_11_distill_2b_8b_off_policy_per_4d_attn_uniform_masking_01_1_2epoch/checkpoint-74/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 26_03_11_distill_2b_8b_off_policy_per_4d_attn_uniform_masking_01_1_2epoch/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 26_03_05_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_01_1_2epoch/checkpoint-37/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-111/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-37/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-74/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/tokenizer.json filter=lfs diff=lfs merge=lfs -text

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/added_tokens.json

@@ -0,0 +1,28 @@
+{
+  "</think>": 151668,
+  "</tool_call>": 151658,
+  "</tool_response>": 151666,
+  "<think>": 151667,
+  "<tool_call>": 151657,
+  "<tool_response>": 151665,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/chat_template.jinja

@@ -0,0 +1,120 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {%- if messages[0].content is string %}
+            {{- messages[0].content }}
+        {%- else %}
+            {%- for content in messages[0].content %}
+                {%- if 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' }}
+        {%- if messages[0].content is string %}
+            {{- messages[0].content }}
+        {%- else %}
+            {%- for content in messages[0].content %}
+                {%- if 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- for message in messages %}
+    {%- if message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- 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 %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role + '\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content_item in message.content %}
+                {%- if 'text' in content_item %}
+                    {{- content_item.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and message.content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- 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 %}
+        {%- endif %}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-111/added_tokens.json

@@ -0,0 +1,28 @@
+{
+  "</think>": 151668,
+  "</tool_call>": 151658,
+  "</tool_response>": 151666,
+  "<think>": 151667,
+  "<tool_call>": 151657,
+  "<tool_response>": 151665,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-111/chat_template.jinja

@@ -0,0 +1,120 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {%- if messages[0].content is string %}
+            {{- messages[0].content }}
+        {%- else %}
+            {%- for content in messages[0].content %}
+                {%- if 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' }}
+        {%- if messages[0].content is string %}
+            {{- messages[0].content }}
+        {%- else %}
+            {%- for content in messages[0].content %}
+                {%- if 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- for message in messages %}
+    {%- if message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- 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 %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role + '\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content_item in message.content %}
+                {%- if 'text' in content_item %}
+                    {{- content_item.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and message.content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- 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 %}
+        {%- endif %}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-111/config.json

@@ -0,0 +1,68 @@
+{
+  "architectures": [
+    "Qwen3VLForConditionalGeneration"
+  ],
+  "dtype": "bfloat16",
+  "eos_token_id": 151645,
+  "image_token_id": 151655,
+  "model_type": "qwen3_vl",
+  "pad_token_id": 151643,
+  "text_config": {
+    "attention_bias": false,
+    "attention_dropout": 0.0,
+    "bos_token_id": 151643,
+    "dtype": "bfloat16",
+    "eos_token_id": 151645,
+    "head_dim": 128,
+    "hidden_act": "silu",
+    "hidden_size": 2048,
+    "initializer_range": 0.02,
+    "intermediate_size": 6144,
+    "max_position_embeddings": 262144,
+    "model_type": "qwen3_vl_text",
+    "num_attention_heads": 16,
+    "num_hidden_layers": 28,
+    "num_key_value_heads": 8,
+    "rms_norm_eps": 1e-06,
+    "rope_scaling": {
+      "mrope_interleaved": true,
+      "mrope_section": [
+        24,
+        20,
+        20
+      ],
+      "rope_type": "default"
+    },
+    "rope_theta": 5000000,
+    "tie_word_embeddings": true,
+    "use_cache": true,
+    "vocab_size": 151936
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "4.57.1",
+  "use_cache": false,
+  "video_token_id": 151656,
+  "vision_config": {
+    "deepstack_visual_indexes": [
+      5,
+      11,
+      17
+    ],
+    "depth": 24,
+    "dtype": "bfloat16",
+    "hidden_act": "gelu_pytorch_tanh",
+    "hidden_size": 1024,
+    "in_channels": 3,
+    "initializer_range": 0.02,
+    "intermediate_size": 4096,
+    "model_type": "qwen3_vl",
+    "num_heads": 16,
+    "num_position_embeddings": 2304,
+    "out_hidden_size": 2048,
+    "patch_size": 16,
+    "spatial_merge_size": 2,
+    "temporal_patch_size": 2
+  },
+  "vision_end_token_id": 151653,
+  "vision_start_token_id": 151652
+}

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-111/generation_config.json

@@ -0,0 +1,13 @@
+{
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "temperature": 0.7,
+  "top_k": 20,
+  "top_p": 0.8,
+  "transformers_version": "4.57.1"
+}

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-111/global_step111/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:a25d5d858f6c03c2a06727bf144ed454c464c75b1151d1c78b6b70ca06cc9545
+size 10323465201

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-111/global_step111/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:844ae58c99465ce6c46545c1fe365947e8a88ce8dac8e8f89bebf7f50885cc42
+size 10323472049

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-111/global_step111/mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9689f3717fd77d9fabbdcb8df92b0a5a99c21e47e9bd2eeff2144f6e67e665bf
+size 5069326589

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-111/latest

@@ -0,0 +1 @@
+global_step111

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-111/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b3d27d92084d333d832f076404d04b2a52f476b35980d082dc1be83b118e952f
+size 4877470336

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-111/preprocessor_config.json

@@ -0,0 +1,39 @@
+{
+  "crop_size": null,
+  "data_format": "channels_first",
+  "default_to_square": true,
+  "device": null,
+  "disable_grouping": null,
+  "do_center_crop": null,
+  "do_convert_rgb": true,
+  "do_normalize": true,
+  "do_pad": null,
+  "do_rescale": true,
+  "do_resize": true,
+  "image_mean": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "image_processor_type": "Qwen2VLImageProcessorFast",
+  "image_std": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "input_data_format": null,
+  "max_pixels": null,
+  "merge_size": 2,
+  "min_pixels": null,
+  "pad_size": null,
+  "patch_size": 16,
+  "processor_class": "Qwen3VLProcessor",
+  "resample": 3,
+  "rescale_factor": 0.00392156862745098,
+  "return_tensors": null,
+  "size": {
+    "longest_edge": 16777216,
+    "shortest_edge": 65536
+  },
+  "temporal_patch_size": 2
+}

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-111/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61e957b4cd785256be4cb26eb03060ef689e1d58f1766d7f26ca36a62bec4994
+size 14917

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-111/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:550c54d430b44b77b0abe44c6e3ceba90a155305315c081b7616b35e2c18d1ce
+size 14917

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-111/special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-111/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aeb13307a71acd8fe81861d94ad54ab689df773318809eed3cbe794b4492dae4
+size 11422654

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-111/tokenizer_config.json

@@ -0,0 +1,240 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151665": {
+      "content": "<tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151666": {
+      "content": "</tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151667": {
+      "content": "<think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151668": {
+      "content": "</think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 262144,
+  "pad_token": "<|endoftext|>",
+  "processor_class": "Qwen3VLProcessor",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-111/trainer_state.json

@@ -0,0 +1,811 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 111,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.027429551055394834,
+      "grad_norm": 551.8389282226562,
+      "learning_rate": 0.0,
+      "loss": 31.7803,
+      "step": 1
+    },
+    {
+      "epoch": 0.05485910211078967,
+      "grad_norm": 546.7216796875,
+      "learning_rate": 5e-07,
+      "loss": 30.82,
+      "step": 2
+    },
+    {
+      "epoch": 0.0822886531661845,
+      "grad_norm": 548.2788696289062,
+      "learning_rate": 7.924812503605781e-07,
+      "loss": 31.3991,
+      "step": 3
+    },
+    {
+      "epoch": 0.10971820422157934,
+      "grad_norm": 565.9237060546875,
+      "learning_rate": 1e-06,
+      "loss": 31.7,
+      "step": 4
+    },
+    {
+      "epoch": 0.13714775527697418,
+      "grad_norm": 561.6377563476562,
+      "learning_rate": 1e-06,
+      "loss": 30.9113,
+      "step": 5
+    },
+    {
+      "epoch": 0.164577306332369,
+      "grad_norm": 563.2583618164062,
+      "learning_rate": 9.906542056074767e-07,
+      "loss": 31.0517,
+      "step": 6
+    },
+    {
+      "epoch": 0.19200685738776385,
+      "grad_norm": 565.8189697265625,
+      "learning_rate": 9.813084112149532e-07,
+      "loss": 30.4734,
+      "step": 7
+    },
+    {
+      "epoch": 0.21943640844315868,
+      "grad_norm": 587.2005004882812,
+      "learning_rate": 9.7196261682243e-07,
+      "loss": 30.7924,
+      "step": 8
+    },
+    {
+      "epoch": 0.24686595949855353,
+      "grad_norm": 570.0631713867188,
+      "learning_rate": 9.626168224299064e-07,
+      "loss": 30.7344,
+      "step": 9
+    },
+    {
+      "epoch": 0.27429551055394835,
+      "grad_norm": 574.8858642578125,
+      "learning_rate": 9.532710280373831e-07,
+      "loss": 30.2849,
+      "step": 10
+    },
+    {
+      "epoch": 0.3017250616093432,
+      "grad_norm": 598.7982177734375,
+      "learning_rate": 9.439252336448598e-07,
+      "loss": 28.2476,
+      "step": 11
+    },
+    {
+      "epoch": 0.329154612664738,
+      "grad_norm": 598.5196533203125,
+      "learning_rate": 9.345794392523364e-07,
+      "loss": 28.5003,
+      "step": 12
+    },
+    {
+      "epoch": 0.35658416372013285,
+      "grad_norm": 594.8805541992188,
+      "learning_rate": 9.25233644859813e-07,
+      "loss": 27.634,
+      "step": 13
+    },
+    {
+      "epoch": 0.3840137147755277,
+      "grad_norm": 605.1944580078125,
+      "learning_rate": 9.158878504672897e-07,
+      "loss": 28.4566,
+      "step": 14
+    },
+    {
+      "epoch": 0.41144326583092256,
+      "grad_norm": 620.5889892578125,
+      "learning_rate": 9.065420560747663e-07,
+      "loss": 27.5533,
+      "step": 15
+    },
+    {
+      "epoch": 0.43887281688631735,
+      "grad_norm": 612.4210815429688,
+      "learning_rate": 8.97196261682243e-07,
+      "loss": 27.2171,
+      "step": 16
+    },
+    {
+      "epoch": 0.4663023679417122,
+      "grad_norm": 607.5828857421875,
+      "learning_rate": 8.878504672897196e-07,
+      "loss": 27.0387,
+      "step": 17
+    },
+    {
+      "epoch": 0.49373191899710706,
+      "grad_norm": 616.777587890625,
+      "learning_rate": 8.785046728971962e-07,
+      "loss": 26.9732,
+      "step": 18
+    },
+    {
+      "epoch": 0.5211614700525019,
+      "grad_norm": 487.4756774902344,
+      "learning_rate": 8.691588785046728e-07,
+      "loss": 18.871,
+      "step": 19
+    },
+    {
+      "epoch": 0.5485910211078967,
+      "grad_norm": 463.1073303222656,
+      "learning_rate": 8.598130841121495e-07,
+      "loss": 18.9919,
+      "step": 20
+    },
+    {
+      "epoch": 0.5760205721632915,
+      "grad_norm": 447.2720642089844,
+      "learning_rate": 8.504672897196261e-07,
+      "loss": 18.5918,
+      "step": 21
+    },
+    {
+      "epoch": 0.6034501232186864,
+      "grad_norm": 445.87921142578125,
+      "learning_rate": 8.411214953271028e-07,
+      "loss": 18.355,
+      "step": 22
+    },
+    {
+      "epoch": 0.6308796742740812,
+      "grad_norm": 421.7401123046875,
+      "learning_rate": 8.317757009345794e-07,
+      "loss": 18.0186,
+      "step": 23
+    },
+    {
+      "epoch": 0.658309225329476,
+      "grad_norm": 418.1053161621094,
+      "learning_rate": 8.22429906542056e-07,
+      "loss": 17.953,
+      "step": 24
+    },
+    {
+      "epoch": 0.6857387763848709,
+      "grad_norm": 404.4615173339844,
+      "learning_rate": 8.130841121495326e-07,
+      "loss": 17.6013,
+      "step": 25
+    },
+    {
+      "epoch": 0.7131683274402657,
+      "grad_norm": 380.7596435546875,
+      "learning_rate": 8.037383177570093e-07,
+      "loss": 17.3135,
+      "step": 26
+    },
+    {
+      "epoch": 0.7405978784956606,
+      "grad_norm": 337.3852844238281,
+      "learning_rate": 7.94392523364486e-07,
+      "loss": 16.6483,
+      "step": 27
+    },
+    {
+      "epoch": 0.7680274295510554,
+      "grad_norm": 307.840087890625,
+      "learning_rate": 7.850467289719626e-07,
+      "loss": 16.05,
+      "step": 28
+    },
+    {
+      "epoch": 0.7954569806064502,
+      "grad_norm": 272.3890075683594,
+      "learning_rate": 7.757009345794392e-07,
+      "loss": 15.468,
+      "step": 29
+    },
+    {
+      "epoch": 0.8228865316618451,
+      "grad_norm": 254.71994018554688,
+      "learning_rate": 7.663551401869158e-07,
+      "loss": 15.4758,
+      "step": 30
+    },
+    {
+      "epoch": 0.8503160827172399,
+      "grad_norm": 231.67564392089844,
+      "learning_rate": 7.570093457943924e-07,
+      "loss": 15.4315,
+      "step": 31
+    },
+    {
+      "epoch": 0.8777456337726347,
+      "grad_norm": 215.486328125,
+      "learning_rate": 7.476635514018691e-07,
+      "loss": 15.1236,
+      "step": 32
+    },
+    {
+      "epoch": 0.9051751848280296,
+      "grad_norm": 206.07510375976562,
+      "learning_rate": 7.383177570093458e-07,
+      "loss": 15.2038,
+      "step": 33
+    },
+    {
+      "epoch": 0.9326047358834244,
+      "grad_norm": 119.7702865600586,
+      "learning_rate": 7.289719626168224e-07,
+      "loss": 14.0311,
+      "step": 34
+    },
+    {
+      "epoch": 0.9600342869388192,
+      "grad_norm": 196.92446899414062,
+      "learning_rate": 7.19626168224299e-07,
+      "loss": 14.7591,
+      "step": 35
+    },
+    {
+      "epoch": 0.9874638379942141,
+      "grad_norm": 222.88058471679688,
+      "learning_rate": 7.102803738317756e-07,
+      "loss": 14.4945,
+      "step": 36
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 85.30135345458984,
+      "learning_rate": 7.009345794392522e-07,
+      "loss": 6.2501,
+      "step": 37
+    },
+    {
+      "epoch": 1.0274295510553948,
+      "grad_norm": 163.3282928466797,
+      "learning_rate": 6.91588785046729e-07,
+      "loss": 13.6152,
+      "step": 38
+    },
+    {
+      "epoch": 1.0548591021107896,
+      "grad_norm": 127.98493957519531,
+      "learning_rate": 6.822429906542056e-07,
+      "loss": 13.091,
+      "step": 39
+    },
+    {
+      "epoch": 1.0822886531661844,
+      "grad_norm": 104.3193359375,
+      "learning_rate": 6.728971962616822e-07,
+      "loss": 12.6728,
+      "step": 40
+    },
+    {
+      "epoch": 1.1097182042215794,
+      "grad_norm": 91.09891510009766,
+      "learning_rate": 6.635514018691588e-07,
+      "loss": 12.4405,
+      "step": 41
+    },
+    {
+      "epoch": 1.1371477552769742,
+      "grad_norm": 81.36380004882812,
+      "learning_rate": 6.542056074766354e-07,
+      "loss": 12.4788,
+      "step": 42
+    },
+    {
+      "epoch": 1.164577306332369,
+      "grad_norm": 77.23246765136719,
+      "learning_rate": 6.44859813084112e-07,
+      "loss": 12.0745,
+      "step": 43
+    },
+    {
+      "epoch": 1.1920068573877638,
+      "grad_norm": 79.51660919189453,
+      "learning_rate": 6.355140186915888e-07,
+      "loss": 12.3292,
+      "step": 44
+    },
+    {
+      "epoch": 1.2194364084431586,
+      "grad_norm": 79.39995574951172,
+      "learning_rate": 6.261682242990654e-07,
+      "loss": 11.9485,
+      "step": 45
+    },
+    {
+      "epoch": 1.2468659594985536,
+      "grad_norm": 77.93059539794922,
+      "learning_rate": 6.16822429906542e-07,
+      "loss": 11.8875,
+      "step": 46
+    },
+    {
+      "epoch": 1.2742955105539484,
+      "grad_norm": 76.22134399414062,
+      "learning_rate": 6.074766355140186e-07,
+      "loss": 11.736,
+      "step": 47
+    },
+    {
+      "epoch": 1.3017250616093432,
+      "grad_norm": 70.79058837890625,
+      "learning_rate": 5.981308411214952e-07,
+      "loss": 11.6677,
+      "step": 48
+    },
+    {
+      "epoch": 1.329154612664738,
+      "grad_norm": 68.9483642578125,
+      "learning_rate": 5.887850467289718e-07,
+      "loss": 11.6169,
+      "step": 49
+    },
+    {
+      "epoch": 1.3565841637201328,
+      "grad_norm": 67.60340881347656,
+      "learning_rate": 5.794392523364486e-07,
+      "loss": 11.5749,
+      "step": 50
+    },
+    {
+      "epoch": 1.3840137147755276,
+      "grad_norm": 67.21224212646484,
+      "learning_rate": 5.700934579439252e-07,
+      "loss": 11.5583,
+      "step": 51
+    },
+    {
+      "epoch": 1.4114432658309226,
+      "grad_norm": 65.26056671142578,
+      "learning_rate": 5.607476635514018e-07,
+      "loss": 11.5118,
+      "step": 52
+    },
+    {
+      "epoch": 1.4388728168863174,
+      "grad_norm": 62.7562255859375,
+      "learning_rate": 5.514018691588784e-07,
+      "loss": 11.3273,
+      "step": 53
+    },
+    {
+      "epoch": 1.4663023679417122,
+      "grad_norm": 63.19300842285156,
+      "learning_rate": 5.42056074766355e-07,
+      "loss": 11.1034,
+      "step": 54
+    },
+    {
+      "epoch": 1.493731918997107,
+      "grad_norm": 60.84135437011719,
+      "learning_rate": 5.327102803738317e-07,
+      "loss": 11.4488,
+      "step": 55
+    },
+    {
+      "epoch": 1.521161470052502,
+      "grad_norm": 63.09079360961914,
+      "learning_rate": 5.233644859813084e-07,
+      "loss": 11.1026,
+      "step": 56
+    },
+    {
+      "epoch": 1.5485910211078968,
+      "grad_norm": 60.59257888793945,
+      "learning_rate": 5.14018691588785e-07,
+      "loss": 11.166,
+      "step": 57
+    },
+    {
+      "epoch": 1.5760205721632916,
+      "grad_norm": 59.48110580444336,
+      "learning_rate": 5.046728971962616e-07,
+      "loss": 10.9686,
+      "step": 58
+    },
+    {
+      "epoch": 1.6034501232186864,
+      "grad_norm": 58.81517791748047,
+      "learning_rate": 4.953271028037383e-07,
+      "loss": 11.0533,
+      "step": 59
+    },
+    {
+      "epoch": 1.6308796742740812,
+      "grad_norm": 57.43783950805664,
+      "learning_rate": 4.85981308411215e-07,
+      "loss": 10.9696,
+      "step": 60
+    },
+    {
+      "epoch": 1.658309225329476,
+      "grad_norm": 55.98307418823242,
+      "learning_rate": 4.7663551401869157e-07,
+      "loss": 10.8266,
+      "step": 61
+    },
+    {
+      "epoch": 1.6857387763848708,
+      "grad_norm": 55.058387756347656,
+      "learning_rate": 4.672897196261682e-07,
+      "loss": 10.7082,
+      "step": 62
+    },
+    {
+      "epoch": 1.7131683274402656,
+      "grad_norm": 54.193199157714844,
+      "learning_rate": 4.5794392523364486e-07,
+      "loss": 10.4601,
+      "step": 63
+    },
+    {
+      "epoch": 1.7405978784956606,
+      "grad_norm": 54.41346740722656,
+      "learning_rate": 4.485981308411215e-07,
+      "loss": 10.6191,
+      "step": 64
+    },
+    {
+      "epoch": 1.7680274295510554,
+      "grad_norm": 54.253849029541016,
+      "learning_rate": 4.392523364485981e-07,
+      "loss": 10.5553,
+      "step": 65
+    },
+    {
+      "epoch": 1.7954569806064502,
+      "grad_norm": 53.43233871459961,
+      "learning_rate": 4.2990654205607476e-07,
+      "loss": 10.6199,
+      "step": 66
+    },
+    {
+      "epoch": 1.8228865316618452,
+      "grad_norm": 53.78797149658203,
+      "learning_rate": 4.205607476635514e-07,
+      "loss": 10.2618,
+      "step": 67
+    },
+    {
+      "epoch": 1.85031608271724,
+      "grad_norm": 51.88896560668945,
+      "learning_rate": 4.11214953271028e-07,
+      "loss": 10.3923,
+      "step": 68
+    },
+    {
+      "epoch": 1.8777456337726348,
+      "grad_norm": 52.391883850097656,
+      "learning_rate": 4.0186915887850467e-07,
+      "loss": 10.3294,
+      "step": 69
+    },
+    {
+      "epoch": 1.9051751848280296,
+      "grad_norm": 49.57695770263672,
+      "learning_rate": 3.925233644859813e-07,
+      "loss": 10.298,
+      "step": 70
+    },
+    {
+      "epoch": 1.9326047358834244,
+      "grad_norm": 52.91679763793945,
+      "learning_rate": 3.831775700934579e-07,
+      "loss": 10.3645,
+      "step": 71
+    },
+    {
+      "epoch": 1.9600342869388192,
+      "grad_norm": 49.12158203125,
+      "learning_rate": 3.7383177570093457e-07,
+      "loss": 10.1515,
+      "step": 72
+    },
+    {
+      "epoch": 1.987463837994214,
+      "grad_norm": 51.0526237487793,
+      "learning_rate": 3.644859813084112e-07,
+      "loss": 10.2743,
+      "step": 73
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 22.302366256713867,
+      "learning_rate": 3.551401869158878e-07,
+      "loss": 4.7568,
+      "step": 74
+    },
+    {
+      "epoch": 2.027429551055395,
+      "grad_norm": 49.09823989868164,
+      "learning_rate": 3.457943925233645e-07,
+      "loss": 10.0536,
+      "step": 75
+    },
+    {
+      "epoch": 2.0548591021107896,
+      "grad_norm": 46.24396896362305,
+      "learning_rate": 3.364485981308411e-07,
+      "loss": 10.1908,
+      "step": 76
+    },
+    {
+      "epoch": 2.0822886531661844,
+      "grad_norm": 46.49886703491211,
+      "learning_rate": 3.271028037383177e-07,
+      "loss": 9.9399,
+      "step": 77
+    },
+    {
+      "epoch": 2.109718204221579,
+      "grad_norm": 46.75298309326172,
+      "learning_rate": 3.177570093457944e-07,
+      "loss": 9.9158,
+      "step": 78
+    },
+    {
+      "epoch": 2.137147755276974,
+      "grad_norm": 45.0136604309082,
+      "learning_rate": 3.08411214953271e-07,
+      "loss": 9.8675,
+      "step": 79
+    },
+    {
+      "epoch": 2.1645773063323688,
+      "grad_norm": 43.98538589477539,
+      "learning_rate": 2.990654205607476e-07,
+      "loss": 9.9633,
+      "step": 80
+    },
+    {
+      "epoch": 2.192006857387764,
+      "grad_norm": 43.370426177978516,
+      "learning_rate": 2.897196261682243e-07,
+      "loss": 9.6898,
+      "step": 81
+    },
+    {
+      "epoch": 2.219436408443159,
+      "grad_norm": 44.13566589355469,
+      "learning_rate": 2.803738317757009e-07,
+      "loss": 9.6963,
+      "step": 82
+    },
+    {
+      "epoch": 2.2468659594985536,
+      "grad_norm": 43.51216506958008,
+      "learning_rate": 2.710280373831775e-07,
+      "loss": 9.7773,
+      "step": 83
+    },
+    {
+      "epoch": 2.2742955105539484,
+      "grad_norm": 43.27804183959961,
+      "learning_rate": 2.616822429906542e-07,
+      "loss": 9.6374,
+      "step": 84
+    },
+    {
+      "epoch": 2.301725061609343,
+      "grad_norm": 43.48250198364258,
+      "learning_rate": 2.523364485981308e-07,
+      "loss": 9.7281,
+      "step": 85
+    },
+    {
+      "epoch": 2.329154612664738,
+      "grad_norm": 43.23092269897461,
+      "learning_rate": 2.429906542056075e-07,
+      "loss": 9.5983,
+      "step": 86
+    },
+    {
+      "epoch": 2.356584163720133,
+      "grad_norm": 40.64580535888672,
+      "learning_rate": 2.336448598130841e-07,
+      "loss": 9.5143,
+      "step": 87
+    },
+    {
+      "epoch": 2.3840137147755276,
+      "grad_norm": 41.178253173828125,
+      "learning_rate": 2.2429906542056074e-07,
+      "loss": 9.441,
+      "step": 88
+    },
+    {
+      "epoch": 2.4114432658309224,
+      "grad_norm": 41.0638542175293,
+      "learning_rate": 2.1495327102803738e-07,
+      "loss": 9.2656,
+      "step": 89
+    },
+    {
+      "epoch": 2.438872816886317,
+      "grad_norm": 39.70026779174805,
+      "learning_rate": 2.05607476635514e-07,
+      "loss": 9.2483,
+      "step": 90
+    },
+    {
+      "epoch": 2.4663023679417124,
+      "grad_norm": 39.37113571166992,
+      "learning_rate": 1.9626168224299064e-07,
+      "loss": 9.2907,
+      "step": 91
+    },
+    {
+      "epoch": 2.493731918997107,
+      "grad_norm": 40.414371490478516,
+      "learning_rate": 1.8691588785046729e-07,
+      "loss": 9.1898,
+      "step": 92
+    },
+    {
+      "epoch": 2.521161470052502,
+      "grad_norm": 40.054298400878906,
+      "learning_rate": 1.775700934579439e-07,
+      "loss": 9.2199,
+      "step": 93
+    },
+    {
+      "epoch": 2.548591021107897,
+      "grad_norm": 39.530765533447266,
+      "learning_rate": 1.6822429906542055e-07,
+      "loss": 9.223,
+      "step": 94
+    },
+    {
+      "epoch": 2.5760205721632916,
+      "grad_norm": 38.541099548339844,
+      "learning_rate": 1.588785046728972e-07,
+      "loss": 9.1145,
+      "step": 95
+    },
+    {
+      "epoch": 2.6034501232186864,
+      "grad_norm": 38.837013244628906,
+      "learning_rate": 1.495327102803738e-07,
+      "loss": 9.2341,
+      "step": 96
+    },
+    {
+      "epoch": 2.630879674274081,
+      "grad_norm": 38.75645065307617,
+      "learning_rate": 1.4018691588785045e-07,
+      "loss": 9.1859,
+      "step": 97
+    },
+    {
+      "epoch": 2.658309225329476,
+      "grad_norm": 38.22486114501953,
+      "learning_rate": 1.308411214953271e-07,
+      "loss": 9.1635,
+      "step": 98
+    },
+    {
+      "epoch": 2.685738776384871,
+      "grad_norm": 37.80083465576172,
+      "learning_rate": 1.2149532710280374e-07,
+      "loss": 9.1447,
+      "step": 99
+    },
+    {
+      "epoch": 2.7131683274402656,
+      "grad_norm": 37.75882339477539,
+      "learning_rate": 1.1214953271028037e-07,
+      "loss": 9.1514,
+      "step": 100
+    },
+    {
+      "epoch": 2.7405978784956604,
+      "grad_norm": 38.09444046020508,
+      "learning_rate": 1.02803738317757e-07,
+      "loss": 9.044,
+      "step": 101
+    },
+    {
+      "epoch": 2.768027429551055,
+      "grad_norm": 35.683841705322266,
+      "learning_rate": 9.345794392523364e-08,
+      "loss": 8.9984,
+      "step": 102
+    },
+    {
+      "epoch": 2.79545698060645,
+      "grad_norm": 37.42098617553711,
+      "learning_rate": 8.411214953271027e-08,
+      "loss": 9.0771,
+      "step": 103
+    },
+    {
+      "epoch": 2.822886531661845,
+      "grad_norm": 36.868045806884766,
+      "learning_rate": 7.47663551401869e-08,
+      "loss": 8.9911,
+      "step": 104
+    },
+    {
+      "epoch": 2.85031608271724,
+      "grad_norm": 38.63874816894531,
+      "learning_rate": 6.542056074766355e-08,
+      "loss": 9.0046,
+      "step": 105
+    },
+    {
+      "epoch": 2.877745633772635,
+      "grad_norm": 36.24422836303711,
+      "learning_rate": 5.6074766355140185e-08,
+      "loss": 8.8876,
+      "step": 106
+    },
+    {
+      "epoch": 2.9051751848280296,
+      "grad_norm": 36.39665222167969,
+      "learning_rate": 4.672897196261682e-08,
+      "loss": 9.0893,
+      "step": 107
+    },
+    {
+      "epoch": 2.9326047358834244,
+      "grad_norm": 36.73887634277344,
+      "learning_rate": 3.738317757009345e-08,
+      "loss": 9.2107,
+      "step": 108
+    },
+    {
+      "epoch": 2.960034286938819,
+      "grad_norm": 35.88780975341797,
+      "learning_rate": 2.8037383177570092e-08,
+      "loss": 8.8774,
+      "step": 109
+    },
+    {
+      "epoch": 2.987463837994214,
+      "grad_norm": 37.03651428222656,
+      "learning_rate": 1.8691588785046726e-08,
+      "loss": 9.0682,
+      "step": 110
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 17.341136932373047,
+      "learning_rate": 9.345794392523363e-09,
+      "loss": 4.1435,
+      "step": 111
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 111,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6.021753466907525e+17,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-111/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d3ab7bae7cf0600a169f62e008dd726b852e1dd85b53866c7363a371c45df950
+size 7953

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-111/video_preprocessor_config.json

@@ -0,0 +1,41 @@
+{
+  "crop_size": null,
+  "data_format": "channels_first",
+  "default_to_square": true,
+  "device": null,
+  "do_center_crop": null,
+  "do_convert_rgb": true,
+  "do_normalize": true,
+  "do_rescale": true,
+  "do_resize": true,
+  "do_sample_frames": true,
+  "fps": 2,
+  "image_mean": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "image_std": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "input_data_format": null,
+  "max_frames": 768,
+  "merge_size": 2,
+  "min_frames": 4,
+  "num_frames": null,
+  "pad_size": null,
+  "patch_size": 16,
+  "processor_class": "Qwen3VLProcessor",
+  "resample": 3,
+  "rescale_factor": 0.00392156862745098,
+  "return_metadata": false,
+  "size": {
+    "longest_edge": 25165824,
+    "shortest_edge": 4096
+  },
+  "temporal_patch_size": 2,
+  "video_metadata": null,
+  "video_processor_type": "Qwen3VLVideoProcessor"
+}

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-111/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)

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-37/added_tokens.json

@@ -0,0 +1,28 @@
+{
+  "</think>": 151668,
+  "</tool_call>": 151658,
+  "</tool_response>": 151666,
+  "<think>": 151667,
+  "<tool_call>": 151657,
+  "<tool_response>": 151665,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-37/chat_template.jinja

@@ -0,0 +1,120 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {%- if messages[0].content is string %}
+            {{- messages[0].content }}
+        {%- else %}
+            {%- for content in messages[0].content %}
+                {%- if 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' }}
+        {%- if messages[0].content is string %}
+            {{- messages[0].content }}
+        {%- else %}
+            {%- for content in messages[0].content %}
+                {%- if 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- for message in messages %}
+    {%- if message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- 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 %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role + '\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content_item in message.content %}
+                {%- if 'text' in content_item %}
+                    {{- content_item.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and message.content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- 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 %}
+        {%- endif %}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-37/config.json

@@ -0,0 +1,68 @@
+{
+  "architectures": [
+    "Qwen3VLForConditionalGeneration"
+  ],
+  "dtype": "bfloat16",
+  "eos_token_id": 151645,
+  "image_token_id": 151655,
+  "model_type": "qwen3_vl",
+  "pad_token_id": 151643,
+  "text_config": {
+    "attention_bias": false,
+    "attention_dropout": 0.0,
+    "bos_token_id": 151643,
+    "dtype": "bfloat16",
+    "eos_token_id": 151645,
+    "head_dim": 128,
+    "hidden_act": "silu",
+    "hidden_size": 2048,
+    "initializer_range": 0.02,
+    "intermediate_size": 6144,
+    "max_position_embeddings": 262144,
+    "model_type": "qwen3_vl_text",
+    "num_attention_heads": 16,
+    "num_hidden_layers": 28,
+    "num_key_value_heads": 8,
+    "rms_norm_eps": 1e-06,
+    "rope_scaling": {
+      "mrope_interleaved": true,
+      "mrope_section": [
+        24,
+        20,
+        20
+      ],
+      "rope_type": "default"
+    },
+    "rope_theta": 5000000,
+    "tie_word_embeddings": true,
+    "use_cache": true,
+    "vocab_size": 151936
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "4.57.1",
+  "use_cache": false,
+  "video_token_id": 151656,
+  "vision_config": {
+    "deepstack_visual_indexes": [
+      5,
+      11,
+      17
+    ],
+    "depth": 24,
+    "dtype": "bfloat16",
+    "hidden_act": "gelu_pytorch_tanh",
+    "hidden_size": 1024,
+    "in_channels": 3,
+    "initializer_range": 0.02,
+    "intermediate_size": 4096,
+    "model_type": "qwen3_vl",
+    "num_heads": 16,
+    "num_position_embeddings": 2304,
+    "out_hidden_size": 2048,
+    "patch_size": 16,
+    "spatial_merge_size": 2,
+    "temporal_patch_size": 2
+  },
+  "vision_end_token_id": 151653,
+  "vision_start_token_id": 151652
+}

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-37/generation_config.json

@@ -0,0 +1,13 @@
+{
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "temperature": 0.7,
+  "top_k": 20,
+  "top_p": 0.8,
+  "transformers_version": "4.57.1"
+}

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-37/global_step37/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1f22b3d4aa67ec79bc7b201a810da55beeb9f581510c1990adcfd1c46492433b
+size 10323465201

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-37/global_step37/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:420b5b6338bd85d34292c4c0ec77247795cf2c3b9582d9fbb65577d50005d79f
+size 10323472049

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-37/global_step37/mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:28b07c2f0eb9be386570d3925c2429839fa815e75b614395c8cd8b1f1c845fa5
+size 5069326589

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-37/latest

@@ -0,0 +1 @@
+global_step37

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-37/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ff04ac727d0d84bd2cbbefc8fa0cb7e75c7aab4372aa06769ad7d2fd6c8131a1
+size 4877470336

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-37/preprocessor_config.json

@@ -0,0 +1,39 @@
+{
+  "crop_size": null,
+  "data_format": "channels_first",
+  "default_to_square": true,
+  "device": null,
+  "disable_grouping": null,
+  "do_center_crop": null,
+  "do_convert_rgb": true,
+  "do_normalize": true,
+  "do_pad": null,
+  "do_rescale": true,
+  "do_resize": true,
+  "image_mean": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "image_processor_type": "Qwen2VLImageProcessorFast",
+  "image_std": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "input_data_format": null,
+  "max_pixels": null,
+  "merge_size": 2,
+  "min_pixels": null,
+  "pad_size": null,
+  "patch_size": 16,
+  "processor_class": "Qwen3VLProcessor",
+  "resample": 3,
+  "rescale_factor": 0.00392156862745098,
+  "return_tensors": null,
+  "size": {
+    "longest_edge": 16777216,
+    "shortest_edge": 65536
+  },
+  "temporal_patch_size": 2
+}

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-37/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4dd7671ce88d469c49c0530724ac76b2306574002d1ecd1ca9294e41621fd96a
+size 14917

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-37/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3246ef1170ccca541a03b89ad6f20e01c51eb6834a2c2211c78c71c70f896879
+size 14917

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-37/special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-37/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aeb13307a71acd8fe81861d94ad54ab689df773318809eed3cbe794b4492dae4
+size 11422654

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-37/tokenizer_config.json

@@ -0,0 +1,240 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151665": {
+      "content": "<tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151666": {
+      "content": "</tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151667": {
+      "content": "<think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151668": {
+      "content": "</think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 262144,
+  "pad_token": "<|endoftext|>",
+  "processor_class": "Qwen3VLProcessor",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-37/trainer_state.json

@@ -0,0 +1,293 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 37,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.027429551055394834,
+      "grad_norm": 551.8389282226562,
+      "learning_rate": 0.0,
+      "loss": 31.7803,
+      "step": 1
+    },
+    {
+      "epoch": 0.05485910211078967,
+      "grad_norm": 546.7216796875,
+      "learning_rate": 5e-07,
+      "loss": 30.82,
+      "step": 2
+    },
+    {
+      "epoch": 0.0822886531661845,
+      "grad_norm": 548.2788696289062,
+      "learning_rate": 7.924812503605781e-07,
+      "loss": 31.3991,
+      "step": 3
+    },
+    {
+      "epoch": 0.10971820422157934,
+      "grad_norm": 565.9237060546875,
+      "learning_rate": 1e-06,
+      "loss": 31.7,
+      "step": 4
+    },
+    {
+      "epoch": 0.13714775527697418,
+      "grad_norm": 561.6377563476562,
+      "learning_rate": 1e-06,
+      "loss": 30.9113,
+      "step": 5
+    },
+    {
+      "epoch": 0.164577306332369,
+      "grad_norm": 563.2583618164062,
+      "learning_rate": 9.906542056074767e-07,
+      "loss": 31.0517,
+      "step": 6
+    },
+    {
+      "epoch": 0.19200685738776385,
+      "grad_norm": 565.8189697265625,
+      "learning_rate": 9.813084112149532e-07,
+      "loss": 30.4734,
+      "step": 7
+    },
+    {
+      "epoch": 0.21943640844315868,
+      "grad_norm": 587.2005004882812,
+      "learning_rate": 9.7196261682243e-07,
+      "loss": 30.7924,
+      "step": 8
+    },
+    {
+      "epoch": 0.24686595949855353,
+      "grad_norm": 570.0631713867188,
+      "learning_rate": 9.626168224299064e-07,
+      "loss": 30.7344,
+      "step": 9
+    },
+    {
+      "epoch": 0.27429551055394835,
+      "grad_norm": 574.8858642578125,
+      "learning_rate": 9.532710280373831e-07,
+      "loss": 30.2849,
+      "step": 10
+    },
+    {
+      "epoch": 0.3017250616093432,
+      "grad_norm": 598.7982177734375,
+      "learning_rate": 9.439252336448598e-07,
+      "loss": 28.2476,
+      "step": 11
+    },
+    {
+      "epoch": 0.329154612664738,
+      "grad_norm": 598.5196533203125,
+      "learning_rate": 9.345794392523364e-07,
+      "loss": 28.5003,
+      "step": 12
+    },
+    {
+      "epoch": 0.35658416372013285,
+      "grad_norm": 594.8805541992188,
+      "learning_rate": 9.25233644859813e-07,
+      "loss": 27.634,
+      "step": 13
+    },
+    {
+      "epoch": 0.3840137147755277,
+      "grad_norm": 605.1944580078125,
+      "learning_rate": 9.158878504672897e-07,
+      "loss": 28.4566,
+      "step": 14
+    },
+    {
+      "epoch": 0.41144326583092256,
+      "grad_norm": 620.5889892578125,
+      "learning_rate": 9.065420560747663e-07,
+      "loss": 27.5533,
+      "step": 15
+    },
+    {
+      "epoch": 0.43887281688631735,
+      "grad_norm": 612.4210815429688,
+      "learning_rate": 8.97196261682243e-07,
+      "loss": 27.2171,
+      "step": 16
+    },
+    {
+      "epoch": 0.4663023679417122,
+      "grad_norm": 607.5828857421875,
+      "learning_rate": 8.878504672897196e-07,
+      "loss": 27.0387,
+      "step": 17
+    },
+    {
+      "epoch": 0.49373191899710706,
+      "grad_norm": 616.777587890625,
+      "learning_rate": 8.785046728971962e-07,
+      "loss": 26.9732,
+      "step": 18
+    },
+    {
+      "epoch": 0.5211614700525019,
+      "grad_norm": 487.4756774902344,
+      "learning_rate": 8.691588785046728e-07,
+      "loss": 18.871,
+      "step": 19
+    },
+    {
+      "epoch": 0.5485910211078967,
+      "grad_norm": 463.1073303222656,
+      "learning_rate": 8.598130841121495e-07,
+      "loss": 18.9919,
+      "step": 20
+    },
+    {
+      "epoch": 0.5760205721632915,
+      "grad_norm": 447.2720642089844,
+      "learning_rate": 8.504672897196261e-07,
+      "loss": 18.5918,
+      "step": 21
+    },
+    {
+      "epoch": 0.6034501232186864,
+      "grad_norm": 445.87921142578125,
+      "learning_rate": 8.411214953271028e-07,
+      "loss": 18.355,
+      "step": 22
+    },
+    {
+      "epoch": 0.6308796742740812,
+      "grad_norm": 421.7401123046875,
+      "learning_rate": 8.317757009345794e-07,
+      "loss": 18.0186,
+      "step": 23
+    },
+    {
+      "epoch": 0.658309225329476,
+      "grad_norm": 418.1053161621094,
+      "learning_rate": 8.22429906542056e-07,
+      "loss": 17.953,
+      "step": 24
+    },
+    {
+      "epoch": 0.6857387763848709,
+      "grad_norm": 404.4615173339844,
+      "learning_rate": 8.130841121495326e-07,
+      "loss": 17.6013,
+      "step": 25
+    },
+    {
+      "epoch": 0.7131683274402657,
+      "grad_norm": 380.7596435546875,
+      "learning_rate": 8.037383177570093e-07,
+      "loss": 17.3135,
+      "step": 26
+    },
+    {
+      "epoch": 0.7405978784956606,
+      "grad_norm": 337.3852844238281,
+      "learning_rate": 7.94392523364486e-07,
+      "loss": 16.6483,
+      "step": 27
+    },
+    {
+      "epoch": 0.7680274295510554,
+      "grad_norm": 307.840087890625,
+      "learning_rate": 7.850467289719626e-07,
+      "loss": 16.05,
+      "step": 28
+    },
+    {
+      "epoch": 0.7954569806064502,
+      "grad_norm": 272.3890075683594,
+      "learning_rate": 7.757009345794392e-07,
+      "loss": 15.468,
+      "step": 29
+    },
+    {
+      "epoch": 0.8228865316618451,
+      "grad_norm": 254.71994018554688,
+      "learning_rate": 7.663551401869158e-07,
+      "loss": 15.4758,
+      "step": 30
+    },
+    {
+      "epoch": 0.8503160827172399,
+      "grad_norm": 231.67564392089844,
+      "learning_rate": 7.570093457943924e-07,
+      "loss": 15.4315,
+      "step": 31
+    },
+    {
+      "epoch": 0.8777456337726347,
+      "grad_norm": 215.486328125,
+      "learning_rate": 7.476635514018691e-07,
+      "loss": 15.1236,
+      "step": 32
+    },
+    {
+      "epoch": 0.9051751848280296,
+      "grad_norm": 206.07510375976562,
+      "learning_rate": 7.383177570093458e-07,
+      "loss": 15.2038,
+      "step": 33
+    },
+    {
+      "epoch": 0.9326047358834244,
+      "grad_norm": 119.7702865600586,
+      "learning_rate": 7.289719626168224e-07,
+      "loss": 14.0311,
+      "step": 34
+    },
+    {
+      "epoch": 0.9600342869388192,
+      "grad_norm": 196.92446899414062,
+      "learning_rate": 7.19626168224299e-07,
+      "loss": 14.7591,
+      "step": 35
+    },
+    {
+      "epoch": 0.9874638379942141,
+      "grad_norm": 222.88058471679688,
+      "learning_rate": 7.102803738317756e-07,
+      "loss": 14.4945,
+      "step": 36
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 85.30135345458984,
+      "learning_rate": 7.009345794392522e-07,
+      "loss": 6.2501,
+      "step": 37
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 111,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "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": 2.007249513705636e+17,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-37/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d3ab7bae7cf0600a169f62e008dd726b852e1dd85b53866c7363a371c45df950
+size 7953

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-37/video_preprocessor_config.json

@@ -0,0 +1,41 @@
+{
+  "crop_size": null,
+  "data_format": "channels_first",
+  "default_to_square": true,
+  "device": null,
+  "do_center_crop": null,
+  "do_convert_rgb": true,
+  "do_normalize": true,
+  "do_rescale": true,
+  "do_resize": true,
+  "do_sample_frames": true,
+  "fps": 2,
+  "image_mean": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "image_std": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "input_data_format": null,
+  "max_frames": 768,
+  "merge_size": 2,
+  "min_frames": 4,
+  "num_frames": null,
+  "pad_size": null,
+  "patch_size": 16,
+  "processor_class": "Qwen3VLProcessor",
+  "resample": 3,
+  "rescale_factor": 0.00392156862745098,
+  "return_metadata": false,
+  "size": {
+    "longest_edge": 25165824,
+    "shortest_edge": 4096
+  },
+  "temporal_patch_size": 2,
+  "video_metadata": null,
+  "video_processor_type": "Qwen3VLVideoProcessor"
+}

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-37/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)

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-74/added_tokens.json

@@ -0,0 +1,28 @@
+{
+  "</think>": 151668,
+  "</tool_call>": 151658,
+  "</tool_response>": 151666,
+  "<think>": 151667,
+  "<tool_call>": 151657,
+  "<tool_response>": 151665,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-74/chat_template.jinja

@@ -0,0 +1,120 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {%- if messages[0].content is string %}
+            {{- messages[0].content }}
+        {%- else %}
+            {%- for content in messages[0].content %}
+                {%- if 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' }}
+        {%- if messages[0].content is string %}
+            {{- messages[0].content }}
+        {%- else %}
+            {%- for content in messages[0].content %}
+                {%- if 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- for message in messages %}
+    {%- if message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- 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 %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role + '\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content_item in message.content %}
+                {%- if 'text' in content_item %}
+                    {{- content_item.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and message.content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- 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 %}
+        {%- endif %}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-74/config.json

@@ -0,0 +1,68 @@
+{
+  "architectures": [
+    "Qwen3VLForConditionalGeneration"
+  ],
+  "dtype": "bfloat16",
+  "eos_token_id": 151645,
+  "image_token_id": 151655,
+  "model_type": "qwen3_vl",
+  "pad_token_id": 151643,
+  "text_config": {
+    "attention_bias": false,
+    "attention_dropout": 0.0,
+    "bos_token_id": 151643,
+    "dtype": "bfloat16",
+    "eos_token_id": 151645,
+    "head_dim": 128,
+    "hidden_act": "silu",
+    "hidden_size": 2048,
+    "initializer_range": 0.02,
+    "intermediate_size": 6144,
+    "max_position_embeddings": 262144,
+    "model_type": "qwen3_vl_text",
+    "num_attention_heads": 16,
+    "num_hidden_layers": 28,
+    "num_key_value_heads": 8,
+    "rms_norm_eps": 1e-06,
+    "rope_scaling": {
+      "mrope_interleaved": true,
+      "mrope_section": [
+        24,
+        20,
+        20
+      ],
+      "rope_type": "default"
+    },
+    "rope_theta": 5000000,
+    "tie_word_embeddings": true,
+    "use_cache": true,
+    "vocab_size": 151936
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "4.57.1",
+  "use_cache": false,
+  "video_token_id": 151656,
+  "vision_config": {
+    "deepstack_visual_indexes": [
+      5,
+      11,
+      17
+    ],
+    "depth": 24,
+    "dtype": "bfloat16",
+    "hidden_act": "gelu_pytorch_tanh",
+    "hidden_size": 1024,
+    "in_channels": 3,
+    "initializer_range": 0.02,
+    "intermediate_size": 4096,
+    "model_type": "qwen3_vl",
+    "num_heads": 16,
+    "num_position_embeddings": 2304,
+    "out_hidden_size": 2048,
+    "patch_size": 16,
+    "spatial_merge_size": 2,
+    "temporal_patch_size": 2
+  },
+  "vision_end_token_id": 151653,
+  "vision_start_token_id": 151652
+}

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-74/generation_config.json

@@ -0,0 +1,13 @@
+{
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "temperature": 0.7,
+  "top_k": 20,
+  "top_p": 0.8,
+  "transformers_version": "4.57.1"
+}

26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch/checkpoint-74/global_step74/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:e4c56ddb10949c930b2cc9828533bbfe750d45a7b662dbd3ba79d26a456e0914
+size 10323465201