Add files using upload-large-folder tool

checkpoints/Qwen2.5-VL-7B-Instruct-tp4pp1-back-exp2-mixed-3-datasets-iter500-frame16-sampled/chat_template.json

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

checkpoints/Qwen2.5-VL-7B-Instruct-tp4pp1-back-exp2-mixed-3-datasets-iter500-frame16-sampled/config.json

@@ -0,0 +1,61 @@
+{
+  "architectures": [
+    "Qwen2_5_VLForConditionalGeneration"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151645,
+  "vision_start_token_id": 151652,
+  "vision_end_token_id": 151653,
+  "vision_token_id": 151654,
+  "image_token_id": 151655,
+  "video_token_id": 151656,
+  "hidden_act": "silu",
+  "hidden_size": 3584,
+  "initializer_range": 0.02,
+  "intermediate_size": 18944,
+  "max_position_embeddings": 128000,
+  "max_window_layers": 28,
+  "model_type": "qwen2_5_vl",
+  "num_attention_heads": 28,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 4,
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 1000000.0,
+  "sliding_window": 32768,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.41.2",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vision_config": {
+    "depth": 32,
+    "hidden_act": "silu",
+    "hidden_size": 1280,
+    "intermediate_size": 3420,
+    "num_heads": 16,
+    "in_chans": 3,
+    "out_hidden_size": 3584,
+    "patch_size": 14,
+    "spatial_merge_size": 2,
+    "spatial_patch_size": 14,
+    "window_size": 112,
+    "fullatt_block_indexes": [
+      7,
+      15,
+      23,
+      31
+    ],
+    "tokens_per_second": 2,
+    "temporal_patch_size": 2
+  },
+  "rope_scaling": {
+    "type": "mrope",
+    "mrope_section": [
+      16,
+      24,
+      24
+    ]
+  },
+  "vocab_size": 152064
+}

checkpoints/Qwen2.5-VL-7B-Instruct-tp4pp1-back-exp2-mixed-3-datasets-iter500-frame16-sampled/generation_config.json

@@ -0,0 +1,12 @@
+{
+  "bos_token_id": 151643,
+  "pad_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "repetition_penalty": 1.05,
+  "temperature": 0.000001,
+  "transformers_version": "4.37.0"
+}

checkpoints/Qwen2.5-VL-7B-Instruct-tp4pp1-back-exp2-mixed-3-datasets-iter500-frame16-sampled/model.safetensors.index.json

@@ -0,0 +1,736 @@
+{
+  "metadata": {
+    "total_size": 16584333312
+  },
+  "weight_map": {
+    "lm_head.weight": "model-00002-of-00002.safetensors",
+    "model.embed_tokens.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.16.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.17.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.17.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.17.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.18.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.18.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.18.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.19.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.19.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.19.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.norm.weight": "model-00002-of-00002.safetensors",
+    "visual.blocks.0.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.0.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.0.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.0.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.0.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.0.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.0.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.0.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.0.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.1.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.1.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.1.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.1.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.1.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.1.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.1.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.1.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.1.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.10.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.10.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.10.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.10.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.10.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.10.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.10.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.10.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.10.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.11.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.11.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.11.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.11.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.11.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.11.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.11.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.11.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.11.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.12.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.12.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.12.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.12.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.12.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.12.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.12.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.12.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.12.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.12.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.12.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.12.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.13.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.13.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.13.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.13.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.13.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.13.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.13.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.13.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.13.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.13.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.13.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.13.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.14.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.14.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.14.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.14.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.14.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.14.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.14.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.14.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.14.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.14.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.14.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.14.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.15.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.15.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.15.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.15.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.15.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.15.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.15.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.15.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.15.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.15.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.15.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.15.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.16.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.16.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.16.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.16.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.16.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.16.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.16.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.16.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.16.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.16.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.16.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.16.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.17.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.17.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.17.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.17.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.17.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.17.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.17.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.17.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.17.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.17.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.17.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.17.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.18.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.18.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.18.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.18.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.18.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.18.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.18.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.18.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.18.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.18.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.18.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.18.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.19.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.19.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.19.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.19.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.19.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.19.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.19.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.19.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.19.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.19.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.19.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.19.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.2.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.2.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.2.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.2.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.2.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.2.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.2.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.2.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.2.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.20.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.20.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.20.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.20.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.20.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.20.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.20.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.20.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.20.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.20.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.20.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.20.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.21.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.21.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.21.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.21.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.21.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.21.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.21.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.21.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.21.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.21.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.21.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.21.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.22.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.22.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.22.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.22.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.22.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.22.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.22.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.22.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.22.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.22.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.22.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.22.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.23.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.23.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.23.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.23.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.23.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.23.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.23.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.23.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.23.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.23.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.23.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.23.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.24.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.24.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.24.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.24.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.24.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.24.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.24.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.24.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.24.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.24.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.24.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.24.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.25.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.25.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.25.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.25.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.25.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.25.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.25.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.25.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.25.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.25.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.25.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.25.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.26.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.26.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.26.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.26.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.26.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.26.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.26.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.26.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.26.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.26.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.26.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.26.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.27.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.27.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.27.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.27.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.27.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.27.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.27.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.27.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.27.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.27.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.27.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.27.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.28.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.28.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.28.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.28.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.28.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.28.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.28.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.28.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.28.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.28.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.28.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.28.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.29.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.29.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.29.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.29.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.29.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.29.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.29.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.29.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.29.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.29.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.29.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.29.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.3.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.3.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.3.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.3.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.3.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.3.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.3.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.3.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.3.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.30.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.30.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.30.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.30.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.30.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.30.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.30.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.30.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.30.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.30.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.30.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.30.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.31.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.31.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.31.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.31.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.31.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.31.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.31.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.31.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.31.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.31.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.31.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.31.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.4.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.4.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.4.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.4.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.4.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.4.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.4.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.4.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.4.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.5.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.5.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.5.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.5.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.5.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.5.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.5.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.5.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.5.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.6.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.6.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.6.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.6.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.6.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.6.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.6.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.6.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.6.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.7.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.7.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.7.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.7.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.7.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.7.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.7.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.7.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.7.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.8.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.8.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.8.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.8.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.8.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.8.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.8.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.8.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.8.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.9.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.9.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.9.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.9.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.9.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.9.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.9.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.9.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.9.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.merger.ln_q.weight": "model-00001-of-00002.safetensors",
+    "visual.merger.mlp.0.bias": "model-00001-of-00002.safetensors",
+    "visual.merger.mlp.0.weight": "model-00001-of-00002.safetensors",
+    "visual.merger.mlp.2.bias": "model-00001-of-00002.safetensors",
+    "visual.merger.mlp.2.weight": "model-00001-of-00002.safetensors",
+    "visual.patch_embed.proj.weight": "model-00001-of-00002.safetensors"
+  }
+}

checkpoints/Qwen2.5-VL-7B-Instruct-tp4pp1-back-exp2-mixed-3-datasets-iter500-frame16-sampled/preprocessor_config.json

@@ -0,0 +1,19 @@
+{
+  "min_pixels": 3136,
+  "max_pixels": 12845056,
+  "patch_size": 14,
+  "temporal_patch_size": 2,
+  "merge_size": 2,
+  "image_mean": [
+    0.48145466,
+    0.4578275,
+    0.40821073
+  ],
+  "image_std": [
+    0.26862954,
+    0.26130258,
+    0.27577711
+  ],
+  "image_processor_type": "Qwen2VLImageProcessor",
+  "processor_class": "Qwen2_5_VLProcessor"
+}

checkpoints/Qwen2.5-VL-7B-Instruct-tp4pp1-back-exp2-mixed-3-datasets-iter500-frame16-sampled/tokenizer_config.json

@@ -0,0 +1,207 @@
+{
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "chat_template": "{% set image_count = namespace(value=0) %}{% set video_count = namespace(value=0) %}{% for message in messages %}{% if loop.first and message['role'] != 'system' %}<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n{% endif %}<|im_start|>{{ message['role'] }}\n{% if message['content'] is string %}{{ message['content'] }}<|im_end|>\n{% else %}{% for content in message['content'] %}{% if content['type'] == 'image' or 'image' in content or 'image_url' in content %}{% set image_count.value = image_count.value + 1 %}{% if add_vision_id %}Picture {{ image_count.value }}: {% endif %}<|vision_start|><|image_pad|><|vision_end|>{% elif content['type'] == 'video' or 'video' in content %}{% set video_count.value = video_count.value + 1 %}{% if add_vision_id %}Video {{ video_count.value }}: {% endif %}<|vision_start|><|video_pad|><|vision_end|>{% elif 'text' in content %}{{ content['text'] }}{% endif %}{% endfor %}<|im_end|>\n{% endif %}{% endfor %}{% if add_generation_prompt %}<|im_start|>assistant\n{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null,
+  "add_bos_token": false
+}

checkpoints/Qwen2.5-VL-7B-Instruct-tp4pp1-back-exp2-mixed-7-datasets-iter2000-frame16-sampled-85k/chat_template.json

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

checkpoints/Qwen2.5-VL-7B-Instruct-tp4pp1-back-exp2-mixed-7-datasets-iter2000-frame16-sampled-85k/config.json

@@ -0,0 +1,61 @@
+{
+  "architectures": [
+    "Qwen2_5_VLForConditionalGeneration"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151645,
+  "vision_start_token_id": 151652,
+  "vision_end_token_id": 151653,
+  "vision_token_id": 151654,
+  "image_token_id": 151655,
+  "video_token_id": 151656,
+  "hidden_act": "silu",
+  "hidden_size": 3584,
+  "initializer_range": 0.02,
+  "intermediate_size": 18944,
+  "max_position_embeddings": 128000,
+  "max_window_layers": 28,
+  "model_type": "qwen2_5_vl",
+  "num_attention_heads": 28,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 4,
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 1000000.0,
+  "sliding_window": 32768,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.41.2",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vision_config": {
+    "depth": 32,
+    "hidden_act": "silu",
+    "hidden_size": 1280,
+    "intermediate_size": 3420,
+    "num_heads": 16,
+    "in_chans": 3,
+    "out_hidden_size": 3584,
+    "patch_size": 14,
+    "spatial_merge_size": 2,
+    "spatial_patch_size": 14,
+    "window_size": 112,
+    "fullatt_block_indexes": [
+      7,
+      15,
+      23,
+      31
+    ],
+    "tokens_per_second": 2,
+    "temporal_patch_size": 2
+  },
+  "rope_scaling": {
+    "type": "mrope",
+    "mrope_section": [
+      16,
+      24,
+      24
+    ]
+  },
+  "vocab_size": 152064
+}

checkpoints/Qwen2.5-VL-7B-Instruct-tp4pp1-back-exp2-mixed-7-datasets-iter2000-frame16-sampled-85k/generation_config.json

@@ -0,0 +1,12 @@
+{
+  "bos_token_id": 151643,
+  "pad_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "repetition_penalty": 1.05,
+  "temperature": 0.000001,
+  "transformers_version": "4.37.0"
+}

checkpoints/Qwen2.5-VL-7B-Instruct-tp4pp1-back-exp2-mixed-7-datasets-iter2000-frame16-sampled-85k/model.safetensors.index.json

@@ -0,0 +1,736 @@
+{
+  "metadata": {
+    "total_size": 16584333312
+  },
+  "weight_map": {
+    "lm_head.weight": "model-00002-of-00002.safetensors",
+    "model.embed_tokens.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.16.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.17.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.17.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.17.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.18.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.18.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.18.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.19.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.19.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.19.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.norm.weight": "model-00002-of-00002.safetensors",
+    "visual.blocks.0.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.0.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.0.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.0.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.0.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.0.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.0.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.0.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.0.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.1.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.1.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.1.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.1.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.1.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.1.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.1.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.1.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.1.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.10.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.10.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.10.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.10.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.10.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.10.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.10.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.10.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.10.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.11.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.11.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.11.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.11.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.11.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.11.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.11.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.11.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.11.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.12.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.12.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.12.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.12.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.12.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.12.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.12.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.12.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.12.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.12.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.12.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.12.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.13.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.13.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.13.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.13.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.13.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.13.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.13.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.13.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.13.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.13.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.13.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.13.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.14.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.14.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.14.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.14.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.14.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.14.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.14.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.14.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.14.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.14.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.14.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.14.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.15.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.15.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.15.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.15.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.15.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.15.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.15.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.15.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.15.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.15.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.15.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.15.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.16.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.16.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.16.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.16.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.16.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.16.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.16.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.16.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.16.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.16.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.16.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.16.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.17.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.17.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.17.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.17.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.17.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.17.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.17.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.17.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.17.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.17.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.17.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.17.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.18.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.18.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.18.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.18.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.18.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.18.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.18.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.18.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.18.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.18.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.18.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.18.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.19.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.19.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.19.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.19.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.19.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.19.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.19.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.19.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.19.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.19.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.19.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.19.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.2.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.2.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.2.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.2.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.2.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.2.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.2.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.2.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.2.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.20.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.20.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.20.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.20.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.20.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.20.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.20.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.20.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.20.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.20.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.20.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.20.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.21.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.21.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.21.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.21.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.21.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.21.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.21.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.21.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.21.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.21.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.21.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.21.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.22.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.22.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.22.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.22.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.22.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.22.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.22.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.22.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.22.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.22.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.22.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.22.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.23.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.23.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.23.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.23.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.23.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.23.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.23.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.23.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.23.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.23.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.23.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.23.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.24.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.24.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.24.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.24.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.24.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.24.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.24.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.24.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.24.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.24.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.24.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.24.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.25.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.25.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.25.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.25.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.25.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.25.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.25.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.25.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.25.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.25.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.25.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.25.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.26.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.26.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.26.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.26.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.26.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.26.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.26.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.26.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.26.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.26.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.26.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.26.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.27.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.27.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.27.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.27.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.27.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.27.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.27.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.27.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.27.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.27.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.27.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.27.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.28.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.28.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.28.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.28.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.28.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.28.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.28.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.28.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.28.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.28.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.28.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.28.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.29.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.29.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.29.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.29.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.29.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.29.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.29.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.29.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.29.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.29.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.29.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.29.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.3.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.3.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.3.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.3.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.3.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.3.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.3.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.3.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.3.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.30.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.30.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.30.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.30.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.30.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.30.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.30.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.30.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.30.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.30.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.30.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.30.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.31.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.31.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.31.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.31.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.31.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.31.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.31.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.31.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.31.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.31.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.31.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.31.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.4.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.4.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.4.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.4.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.4.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.4.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.4.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.4.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.4.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.5.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.5.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.5.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.5.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.5.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.5.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.5.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.5.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.5.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.6.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.6.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.6.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.6.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.6.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.6.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.6.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.6.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.6.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.7.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.7.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.7.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.7.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.7.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.7.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.7.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.7.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.7.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.8.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.8.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.8.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.8.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.8.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.8.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.8.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.8.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.8.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.9.attn.proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.9.attn.proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.9.attn.qkv.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.9.attn.qkv.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.9.mlp.down_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.9.mlp.gate_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.9.mlp.up_proj.bias": "model-00001-of-00002.safetensors",
+    "visual.blocks.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.9.norm1.weight": "model-00001-of-00002.safetensors",
+    "visual.blocks.9.norm2.weight": "model-00001-of-00002.safetensors",
+    "visual.merger.ln_q.weight": "model-00001-of-00002.safetensors",
+    "visual.merger.mlp.0.bias": "model-00001-of-00002.safetensors",
+    "visual.merger.mlp.0.weight": "model-00001-of-00002.safetensors",
+    "visual.merger.mlp.2.bias": "model-00001-of-00002.safetensors",
+    "visual.merger.mlp.2.weight": "model-00001-of-00002.safetensors",
+    "visual.patch_embed.proj.weight": "model-00001-of-00002.safetensors"
+  }
+}

checkpoints/Qwen2.5-VL-7B-Instruct-tp4pp1-back-exp2-mixed-7-datasets-iter2000-frame16-sampled-85k/preprocessor_config.json

@@ -0,0 +1,19 @@
+{
+  "min_pixels": 3136,
+  "max_pixels": 12845056,
+  "patch_size": 14,
+  "temporal_patch_size": 2,
+  "merge_size": 2,
+  "image_mean": [
+    0.48145466,
+    0.4578275,
+    0.40821073
+  ],
+  "image_std": [
+    0.26862954,
+    0.26130258,
+    0.27577711
+  ],
+  "image_processor_type": "Qwen2VLImageProcessor",
+  "processor_class": "Qwen2_5_VLProcessor"
+}

checkpoints/Qwen2.5-VL-7B-Instruct-tp4pp1-back-exp2-mixed-7-datasets-iter2000-frame16-sampled-85k/tokenizer_config.json

@@ -0,0 +1,207 @@
+{
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "chat_template": "{% set image_count = namespace(value=0) %}{% set video_count = namespace(value=0) %}{% for message in messages %}{% if loop.first and message['role'] != 'system' %}<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n{% endif %}<|im_start|>{{ message['role'] }}\n{% if message['content'] is string %}{{ message['content'] }}<|im_end|>\n{% else %}{% for content in message['content'] %}{% if content['type'] == 'image' or 'image' in content or 'image_url' in content %}{% set image_count.value = image_count.value + 1 %}{% if add_vision_id %}Picture {{ image_count.value }}: {% endif %}<|vision_start|><|image_pad|><|vision_end|>{% elif content['type'] == 'video' or 'video' in content %}{% set video_count.value = video_count.value + 1 %}{% if add_vision_id %}Video {{ video_count.value }}: {% endif %}<|vision_start|><|video_pad|><|vision_end|>{% elif 'text' in content %}{{ content['text'] }}{% endif %}{% endfor %}<|im_end|>\n{% endif %}{% endfor %}{% if add_generation_prompt %}<|im_start|>assistant\n{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null,
+  "add_bos_token": false
+}

prediction_results/iter2000_sampled_85k/-qIj6mLoFTM-Scene-036-Start-17052-End-17275+cropped_split_0.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/-qIj6mLoFTM-Scene-036-Start-17052-End-17275+cropped_split_0.mp4",
+  "gt_caption": "The video opens with a dynamic scene from a TEDx talk in Le\u00f3n, featuring a charismatic middle-aged speaker with salt-and-pepper hair delivering an insightful presentation. He is dressed in a dark blazer over a vertically striped shirt and dark trousers, projecting a smart-casual yet professional image. The stage is set against a black backdrop with a prominent red \"TEDx\" logo partially obscured on the left, and the word \"Le\u00f3n\" in a lighter, off-white serif font to its right. Above the speaker, a large, luminous teal-blue screen serves as a backdrop, featuring a small, square blue graphic in its upper right quadrant, potentially a sun-like emblem.\n\nThe speaker stands in a focused and inviting atmosphere, actively engaging his audience with animated speech and expressive hand gestures. He shifts his weight slightly, leaning forward to emphasize key points before pulling back. His facial expressions are lively, with a subtly furrowed brow suggesting concentration and a mouth frequently open mid-sentence. His eye gaze sweeps across the audience, establishing a connection, with occasional glances to his left or right, possibly engaging with different sections of the crowd.\n\nThroughout the presentation, the speaker's hands are in constant motion, punctuating his words. While his right hand remains close to his body, holding a small, black remote control with minimal movement, his left hand is far more expressive. He frequently gestures with an open palm and splayed fingers, moving his hand up and down to underscore his points and add a visual rhythm to his communication. His head movements are natural, accompanying his verbal emphasis with subtle nods and turns.\n\nThe camera maintains a steady, static medium shot, framing the speaker from the waist up. The eye-level angle creates a direct and personal connection with the presenter, immersing the viewer in the experience as if they were part of the audience. The even lighting softly illuminates the speaker against the expansive teal-blue screen, creating a sense of anticipation and intellectual curiosity without any harsh shadows.",
+  "prediction": "A medium shot, from roughly the waist up, captures a middle-aged man delivering a presentation on a stage.  The camera remains static throughout, offering a direct and engaging perspective of the speaker.  The backdrop is a large, dark banner featuring the bold white text \"TEDx Le\u00f3n\" against a black background.  A smaller, red rectangle with a star-like pattern is visible above the text.  The lighting is focused on the speaker, creating a professional and formal atmosphere.\n\nThe speaker, a man with short, neatly styled grey hair, is dressed in a dark blazer over a light-colored shirt with thin vertical stripes (likely blue or purple) and dark trousers. He stands upright, maintaining a confident posture.  His facial expression is animated and engaged, his mouth open as he speaks, and his eyes directed forward, likely at the audience.  His head moves subtly, emphasizing points in his speech.\n\nHe holds a small, dark rectangular object, possibly a presentation clicker, in his right hand.  This object is used to punctuate his speech, moving it slightly up and down and gesturing towards the audience.  His left hand is more expressive, frequently opening into an open palm gesture, then closing, and sometimes pointing, to emphasize key points.  These gestures are fluid and natural, complementing his verbal delivery.  The speaker's movements are contained yet purposeful, reflecting a practiced and comfortable public speaking style.  The overall impression is one of a dynamic and engaging presentation, typical of a TEDx event.",
+  "avg_length": 242
+}

prediction_results/iter2000_sampled_85k/-uyx2ls34yw-Scene-060-Start-21466-End-21825+cropped_split_0.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/-uyx2ls34yw-Scene-060-Start-21466-End-21825+cropped_split_0.mp4",
+  "gt_caption": "The video captures a dynamic moment from a TED presentation, focusing on the speaker's engaging stage presence. The stage is defined by large, bold, three-dimensional red letters spelling \"TED\" in the lower left of the frame, contrasting with the dark stage floor. The backdrop features a large, light-colored screen or wall. Initially, the auburn-haired speaker, dressed in a stylish black leather jacket worn open over a blush pink high-necked top and light-colored skirt, is captured mid-stride, her face beaming with a wide, joyful smile, radiating enthusiasm. She holds a slender, black remote-like device in her right hand, held low and relaxed. As the seconds pass, her smile softens into a pleasant and engaging expression, and she takes a few measured steps to her left, gliding across the stage with a purposeful yet fluid walk. Her left arm, initially out of frame, smoothly swings into view, accompanying her natural gait. She then comes to a confident and composed stop, her gaze consistently directed forward, presumably at her audience, conveying attentiveness. The video maintains a static medium shot, framing the speaker from the waist up, allowing for an intimate focus on her expressions, gestures, and upper body movements. The overall atmosphere is one of focused attention and intellectual exchange, fostered by the bright stage lights evenly illuminating the speaker.",
+  "prediction": "A woman, likely a presenter, stands center stage, delivering a presentation to an unseen audience.  The scene is set on a stage with a large, light grey screen or wall behind her, and a dark blue structure, possibly part of the stage design, positioned to her left.  Bright, even lighting illuminates the speaker, creating a professional atmosphere.\n\nThe camera maintains a static medium shot, framing the speaker from roughly the waist up, providing a clear view of her upper body and facial expressions.  The angle is slightly below eye level, giving a sense of presence.\n\nThe speaker, a woman with shoulder-length reddish-brown hair, is dressed in a black leather jacket over a light-colored top (possibly white or pale pink) and a light-colored skirt (likely beige or cream).  She holds a small, dark rectangular object, resembling a presentation clicker, in her right hand.  Her posture is upright and confident, and she stands relatively still, though subtle shifts in weight and slight turns of her head suggest engagement with the audience.\n\nHer expression is consistently positive and engaging; a warm smile plays on her lips, and her eyes are directed towards the audience, indicating direct communication.  Her mouth moves actively as she speaks, conveying enthusiasm and confidence.  Her left hand hangs relaxed at her side, while her right hand, holding the clicker, remains steady, occasionally making small adjustments.  These subtle movements and gestures, combined with her animated facial expressions, suggest a dynamic and engaging presentation style.  The overall impression is one of a confident and enthusiastic speaker, actively communicating with her audience.",
+  "avg_length": 260
+}

prediction_results/iter2000_sampled_85k/0_AzXpHnx6M-Scene-046-Start-15208-End-15583+cropped_split_1.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/0_AzXpHnx6M-Scene-046-Start-15208-End-15583+cropped_split_1.mp4",
+  "gt_caption": "The video opens with a static, eye-level medium shot of a man on stage, likely during a presentation or lecture. He stands in front of dark, possibly navy blue, floor-length curtains which serve as a backdrop. The lighting is direct, focused on him and the immediate foreground. He is dressed in a smart casual style, including a light blue button-down shirt under a dark V-neck sweater vest, blue denim jeans held by a brown leather belt, and a watch on his left wrist. A white name tag or ID badge is clipped to his sweater vest, suggesting a professional setting. The man has a well-groomed dark beard and dark, curly hair. His posture is poised and grounded.\n\nHe holds a small, dark, cylindrical object, possibly made of black or dark grey plastic, in his left hand. Its size suggests it might be a clicker, a specialized tool, or a small component. Initially, his right hand hovers near the object. The fingers of his right hand then move purposefully to manipulate the object, perhaps pressing a button or making an adjustment. The manipulation is subtle and deliberate. His facial expression is one of quiet concentration; his brow is slightly furrowed, and his gaze shifts between the object and a point forward, presumably towards an unseen audience. Although silent, his focused demeanor suggests he is communicating a complex detail. As the clip progresses, the index finger of his right hand extends upwards in a clear, emphatic gesture, signaling a transition to a new point or a conclusion to a thought. This gesture serves as a non-verbal cue, effectively guiding the audience's attention.\n\nIn the foreground, several pristine white, rectangular blocks are arranged, functioning as minimalist stage props. To the far right, a vibrant, variegated green and yellow plant, possibly a snake plant (Sansevieria), adds a subtle organic touch to the otherwise stark stage. The overall atmosphere is one of focused attention and educational discourse, typical of a conference or specialized presentation. The camera remains static, maintaining a medium shot of the man from just above his waist to the top of his head, ensuring the viewer's attention is fixed on his controlled gestures and interaction with the object.",
+  "prediction": "A medium shot, from roughly the waist up, captures a man delivering a presentation or lecture.  The camera remains static throughout, offering a direct and focused view of the speaker.  He stands upright, his posture confident and engaged, against a dark, textured backdrop, likely a curtain or wall.  The lighting is even and bright, creating a professional atmosphere.\n\nThe speaker, a man with a full, dark beard and short, dark hair, wears a light brown or tan collared shirt, layered under a dark brown or black V-neck sweater vest.  A small white name tag is visible on his vest.  He's wearing blue jeans and a watch on his left wrist.  His expression is serious and focused, his brow slightly furrowed, suggesting concentration.  His eyes are directed forward, likely at an audience or a screen just out of frame.\n\nIn his right hand, he holds a small, dark, rectangular object, possibly a presentation clicker or remote control.  Its smooth surface reflects the light.  In his left hand, he holds a small, white, rectangular object, resembling a business card or a small piece of paper.  This object has faint, unreadable text or graphics on it.  Both objects are held close to his body, and his hands are positioned near his chest.\n\nThe speaker begins by holding both objects, then subtly shifts his grip.  His right hand, holding the dark object, makes a slight adjustment, while his left hand, holding the white object, also makes a minor movement.  These actions suggest he is preparing to use the objects for a presentation or demonstration.  The overall impression is one of a speaker poised and ready to engage with his material.  The scene is calm and focused, with no significant movement beyond the speaker's subtle adjustments.",
+  "avg_length": 291
+}

prediction_results/iter2000_sampled_85k/1010070-Split-000-Start-10-End-369+cropped_split_0.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/1010070-Split-000-Start-10-End-369+cropped_split_0.mp4",
+  "gt_caption": "The video presents a serene scene of a solitary figure walking slowly through a vast, golden field of ripened grain. Bathed in the warm glow of either sunrise or sunset, the countless stalks of grain, colored in rich amber hues, reach approximately mid-thigh on the character, creating a textured expanse that stretches towards the horizon. A thin, dark line of distant trees marks the edge of the plain beneath a pale blue sky that transitions into a bright, golden-white light near the horizon.\n\nThe main character, seen from behind, moves deliberately forward into the golden expanse. They wear a wide-brimmed, light straw hat that partially obscures their face, revealing glimpses of dark, flowing hair. They are dressed in light, layered clothing: a long-sleeved inner garment, possibly white or cream-colored, beneath a flowing, sleeveless outer vest or duster in soft grey or beige. This outer layer extends past their knees, creating an elongated silhouette. Their arms are slightly outstretched with open hands hanging loosely, and their fingers gently brush the tops of the grain as they walk, emphasizing a tactile engagement with their surroundings.\n\nThe character\u2019s upright, relaxed posture and unhurried pace communicate a profound sense of calm and contentment despite the lack of visible facial expressions. Their open-handed gesture suggests a readiness to receive the environment, and their gentle interaction with the grain indicates a reverent connection.\n\nThe background, dominated by the boundless golden field, is bathed in soft, warm light, creating long shadows and highlights that accentuate the grain's texture. The largely clear sky, with a few faint clouds, adds to the expansive feel. The steady, static medium-long shot maintains an eye-level perspective from behind the character, emphasizing the scale of the landscape and their journey into it. The sharp focus on the character and foreground, with a gentle softening of the distant background, enhances the scene's depth. The lack of camera movements, cuts, or transitions allows uninterrupted observation of the continuous, unfolding scene, evoking a feeling of tranquility, natural beauty, and timeless simplicity.",
+  "prediction": "The video opens with a serene scene of a woman walking through a vast, golden wheat field under a clear, pale blue sky dotted with wispy clouds. The overall theme evokes tranquility and connection with nature, suggesting a peaceful journey or escape into the countryside. The camera, positioned at eye-level, employs a medium-long shot, framing the woman from her mid-thighs to just above her head, maintaining a steady, smooth tracking motion as she walks directly away from the viewer.\n\nThe woman is seen from behind, wearing a long, flowing white dress that reaches her ankles, and a wide-brimmed straw hat that casts a soft shadow over her face, obscuring her features. Her dark hair is visible beneath the hat. She walks with a calm and unhurried gait, her posture upright and relaxed. Her arms are extended slightly outwards, and both hands gently brush against the tops of the tall, ripe wheat stalks, creating a rhythmic, caressing motion. This deliberate interaction with the wheat suggests a tactile appreciation for the environment.\n\nThe wheat field stretches endlessly towards the horizon, its dense, golden-brown stalks forming a textured carpet. The lighting is soft and warm, characteristic of late afternoon or early morning sun, casting a gentle glow on the wheat and creating subtle shadows. In the far distance, a faint line of trees or low hills marks the horizon, adding depth to the expansive landscape. The atmosphere is profoundly peaceful and idyllic, conveying a sense of calm and natural beauty. The continuous, uninterrupted shot emphasizes the uninterrupted flow of the moment, reinforcing the tranquil mood.",
+  "avg_length": 262
+}

prediction_results/iter2000_sampled_85k/1014792-Split-000-Start-10-End-348+cropped_split_1.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/1014792-Split-000-Start-10-End-348+cropped_split_1.mp4",
+  "gt_caption": "Bathed in the warm glow of a tropical sunset, a solitary man enjoys a tranquil moment on a unique swing set located directly in the ocean's shallow waters. Two rustic swing frames, each composed of tall, dark wooden posts firmly planted in the seabed, support a horizontal beam from which a simple wooden seat hangs by ropes. These weathered structures stand as a focal point amidst the shimmering water where numerous dark, irregular rocks and coral protrude. The sun, a brilliant orb of orange and yellow, dominates the right side of the frame, casting a long, glistening path of golden reflection across the gently rippling ocean surface.\n\nThe man, bare-chested and wearing dark shorts, sits upright on the swing seat, his toned back and shoulders visible as he faces away from the camera. His hair color is not discernible. He holds the ropes of the swing with both hands, his bare feet partially submerged in the shallow water. With the static, eye-level medium-wide shot, he gently sways back and forth with the water's movement, embodying a state of calm absorption in his surroundings.\n\nThe scene unfolds at a quiet, unhurried pace, emphasizing the peaceful passage of time as the sunset light remains constant. The setting is a tropical coastal area at low tide, evident from the shallow water and exposed rocks. The sky transitions from bright orange and gold near the sun to pale blue above. The ocean stretches calmly to the horizon, reflecting the vivid sunlight. This serene and picturesque atmosphere evokes a feeling of peace, solitude, and natural wonder, capturing a moment of tranquil contemplation and a deep connection with the marine environment at dusk. The absence of camera movements further contributes to the undisturbed feel of the scene.",
+  "prediction": "The video opens with a serene, wide shot of a tranquil sunset scene, establishing an atmosphere of peace and solitude. Two dark, rectangular swing frames stand in the calm, shallow waters, their sturdy posts firmly rooted in the seabed. The frame on the left is empty, while the one on the right holds a lone figure seated on its swing. The individual, silhouetted against the bright horizon, appears to be a young adult with short hair, wearing dark swimwear or shorts. Their posture is relaxed yet attentive, with hands resting casually on their lap, conveying a sense of calm and introspection.\n\nThe primary action involves the gentle, rhythmic swinging of the main character's swing, synchronized with the subtle undulations of the water. This continuous, unhurried motion emphasizes the peaceful passage of time. The vast expanse of the ocean stretches out, reflecting the golden light of the setting sun, which casts a shimmering path across the water. Scattered rocks and coral formations dot the seabed, adding texture to the otherwise smooth surface.\n\nThe background features a clear sky transitioning from soft blue at the top to warm orange and yellow near the horizon where the sun is positioned. The overall mood is one of profound tranquility and natural beauty, evoking a sense of escape and quiet contemplation. The camera remains static throughout, maintaining a consistent eye-level perspective that allows the viewer to fully absorb the peaceful scene without distraction. There are no cuts or transitions, contributing to the uninterrupted flow of the moment.",
+  "avg_length": 252
+}

prediction_results/iter2000_sampled_85k/1026004-Split-000-Start-10-End-385+cropped_split_0.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/1026004-Split-000-Start-10-End-385+cropped_split_0.mp4",
+  "gt_caption": "The video opens with a static, eye-level, medium-long shot of a serene mountain clearing, presenting a tranquil moment in the wilderness. Two hikers, equipped for a multi-day trek, enter the frame from the left. The first hiker, a man with dark hair wearing a striking red hooded jacket and dark olive or grey trousers, walks with a steady, measured pace, indicative of someone accustomed to carrying heavy loads. He carries a substantial, dark-colored hiking backpack, appearing well-filled. He takes a few steps and then pauses, subtly tilting his head and surveying the clearing with a calm, observant expression. His hands briefly adjust something on the sides of his backpack, a gesture suggesting relief or repositioning of his load.\n\nMeanwhile, his companion, clad in a dark blue or black puffy jacket and dark trousers, continues walking a short distance toward a primitive wooden bench. The bench is crafted from a split log, weathered and sturdy, and is accompanied by several cut tree stumps serving as potential seating or a fire pit surround. With a practiced movement, the second hiker bends significantly at the waist, reaching downwards towards the bench, likely intending to rest his heavy, dark-colored backpack or inspect the ground. The first hiker remains standing, patiently observing his companion.\n\nThe foreground features a lush expanse of slightly undulating green grass leading to a darker, rugged terrain in the distance. The backdrop is dominated by a dense, imposing wall of dark green coniferous trees, their pointed crowns reaching towards a bright, partly cloudy sky. The sky is a serene pale blue, punctuated by gentle, diffused clouds, suggesting fair weather. The overall atmosphere is one of peaceful tranquility, emphasizing the enduring spirit of outdoor adventure and the simple beauty of nature, portraying a quiet shared experience between companions in a remote setting. The absence of camera movement enhances this sense of calm observation.",
+  "prediction": "In a serene mountain landscape, two hikers pause for respite on a grassy hillside, their journey momentarily interrupted by the need to prepare for a fire. The scene is set against a backdrop of towering evergreen trees, their dark green needles contrasting with the lighter sky and soft white clouds. The atmosphere is tranquil and natural, bathed in bright daylight.\n\nThe hiker on the left, dressed in a vibrant blue jacket, dark grey pants, and sturdy hiking boots, stands upright, observing his companion. He carries a large, dark grey backpack. Meanwhile, the hiker on the right, sporting a dark olive-green jacket, dark trousers, and similar footwear, bends down to interact with a rustic wooden bench crafted from rough-hewn logs. He also has a dark grey backpack. To the right of the bench, several neatly stacked pieces of cut wood are visible, suggesting a designated spot for campfires.\n\nThe hiker on the right carefully places or adjusts something on the wooden bench, his movements deliberate and focused. His posture shifts from standing to bending at the waist and knees, indicating a task requiring precision. The hiker on the left remains largely still, his hands hanging naturally at his sides, his gaze fixed on his companion's actions. His expression is neutral, conveying quiet observation rather than active participation. There is no verbal communication between them; their interaction is purely physical and task-oriented.\n\nThe camera maintains a static, medium-long shot throughout, capturing both hikers and their immediate surroundings without any panning, tilting, or zooming. This steady perspective allows the viewer to observe the subtle actions and details of the scene without distraction, emphasizing the peaceful solitude and the practicalities of outdoor living.",
+  "avg_length": 279
+}

prediction_results/iter2000_sampled_85k/1036703-Split-000-Start-10-End-401+cropped_split_0.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/1036703-Split-000-Start-10-End-401+cropped_split_0.mp4",
+  "gt_caption": "Against the backdrop of a breathtaking sunset, two figures stand silhouetted atop a mountain peak, radiating triumph and camaraderie. Their forms, rendered dark against the brilliant sun, suggest they are climbers or hikers equipped with outdoor gear and harnesses. The scene captures a moment of shared achievement, the culmination of effort in reaching a formidable goal.\n\nThe figures, one slightly taller than the other, stand confidently on the dark, indistinct peak. In a decisive, synchronized movement, they raise their hands and meet in a quick, emphatic high-five, a universal gesture of success and partnership. This contact lasts only a fraction of a second, after which they hold the celebratory pose, arms raised, emphasizing the completion of their shared feat. Although their facial expressions are obscured by the backlighting, their body language speaks volumes. Their upright posture reflects the successful climb, and their coordinated movements underscore their unified effort.\n\nThe setting sun casts a golden-orange glow across the horizon, creating the dramatic silhouettes. Above the intense light, the sky transitions to a serene, pale blue, adorned with delicate cirrus clouds that add depth. Below the peak, successive layers of dark mountain ranges stretch into the distance, forming a majestic backdrop that instills a sense of vastness and accomplishment. The atmosphere is one of profound peace intertwined with the exhilarating feeling of conquest.\n\nThe camera remains static in a medium-long shot, perfectly framing the figures against the expansive sky and receding mountains. A slightly low angle emphasizes their elevated position and the grandeur of their achievement. The absence of camera movement allows the viewer to fully absorb the powerful imagery and the impactful moment without distraction.",
+  "prediction": "Against a breathtaking backdrop of a mountain range silhouetted against a vibrant sunset, two figures stand atop a rocky peak, their forms rendered as stark black silhouettes. The sky transitions from soft blues and purples at the horizon to warm oranges and yellows near the sun, which casts a brilliant lens flare directly behind them. The scene evokes a sense of accomplishment, shared joy, and the triumph of human endeavor over nature's grandeur.\n\nThe two individuals, likely hikers or mountaineers, are equipped with backpacks and climbing gear, suggesting a challenging ascent. They are engaged in a celebratory high-five, their hands meeting in the center of the frame. The figure on the left, slightly taller, initiates the gesture, extending their right hand upwards towards the other. The figure on the right responds by raising their left hand to meet it. Their synchronized movement is fluid and deliberate, culminating in a firm, satisfying clasp. This action, performed with a slight lean forward, conveys a strong sense of mutual respect, shared success, and unspoken communication of achievement. The high-five is held for a moment, symbolizing their shared victory before they slowly lower their hands back down, completing the gesture.\n\nThe camera remains static throughout, employing a medium-long shot that frames the two figures centrally against the expansive sky. The eye-level angle provides a direct and immersive perspective, allowing the viewer to fully appreciate the serene beauty of the natural environment and the intimate moment between the two individuals. The overall atmosphere is one of profound tranquility and awe-inspiring majesty, imbued with a sense of peace and accomplishment.",
+  "avg_length": 264
+}

prediction_results/iter2000_sampled_85k/1044042-Split-000-Start-10-End-360+cropped_split_0.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/1044042-Split-000-Start-10-End-360+cropped_split_0.mp4",
+  "gt_caption": "In a scene capturing modern digital absorption, a middle-aged man with short, grey-tinged hair, wearing clear-framed glasses, a blue plaid shirt, and ripped blue jeans, is deeply engrossed in his sleek, silver-backed smartphone. He is comfortably settled on a large, plush sofa upholstered in a rich, dark teal or petrol blue chenille fabric, set against a plain, light grey wall in a contemporary living space. The lighting is even and bright, contributing to a calm and tranquil domestic atmosphere.\n\nFrom a slightly low, static camera angle, the man's gaze remains fixed on the phone screen, his brow slightly furrowed in concentration as his left thumb subtly interacts with the device. His body, broad and sturdy, is relaxed and still, nestled into the cushions. At approximately 0:02, a fluffy black and white cat, with significant white markings on its chest and belly, enters the frame from the lower left. It gracefully leaps onto the sofa and walks across the cushions behind the man's left leg, its tail swishing gently, before exiting to the right by 0:05. Throughout this brief feline traverse, the man remains utterly oblivious, never once acknowledging the cat's presence, highlighting his complete immersion in the digital world and showcasing the theme of quiet, solitary leisure in a modern home. The shot remains a medium-long shot from the knees up, showing the entire man on the couch, maintaining an uninterrupted, observational view of the scene.",
+  "prediction": "The video opens with a static, eye-level medium shot of a middle-aged man relaxing on a plush, dark green velvet sofa. He is dressed casually in a brown and black plaid flannel shirt and light blue denim jeans, his short, neatly combed grey hair and clear-framed glasses contributing to his relaxed demeanor. Initially, he is deeply engrossed in his dark-colored smartphone, held horizontally in both hands, his brow slightly furrowed in concentration as he taps or swipes at the screen. The scene is set indoors, likely a living room, with soft, even lighting creating a calm and intimate atmosphere.\n\nAfter a few seconds, a small, fluffy black and white cat with distinct markings on its head and tail enters the frame from the left, walking slowly across the sofa towards the right. The man's focus shifts from the phone to the cat, his expression softening into a gentle smile. His right hand, previously supporting the phone, now rests on the sofa cushion beside him, while his left hand continues to hold the device. The cat continues its unhurried journey, eventually exiting the frame, leaving the man once again absorbed in his phone. Throughout this interaction, the camera remains fixed, maintaining a consistent perspective and emphasizing the quiet, domestic tranquility of the moment.",
+  "avg_length": 212
+}

prediction_results/iter2000_sampled_85k/1064708-Split-000-Start-10-End-365+cropped_split_1.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/1064708-Split-000-Start-10-End-365+cropped_split_1.mp4",
+  "gt_caption": "In a serene and cozy living space, a slender young woman with long, dark hair partially hidden under a red and white Santa hat puts the finishing touches on a beautifully decorated Christmas tree. The room is dimly lit, emphasizing the warm, golden glow emanating from the multitude of fairy lights adorning the lush, green tree. Interspersed among the lights are numerous vibrant red spherical ornaments of varying sizes, complemented by golden accents and possibly red ribbons or bows. A distinctive, ornate red and gold tree topper with a pointed spire crowns the tree.\n\nDressed in a long-sleeved red top that echoes the red ornaments, a dark, likely black, short skirt, and dark tights or stockings, the woman embodies a chic and minimalist holiday style. Initially, she leans slightly forward, her face expressing focused concentration as she meticulously adjusts the red and gold tree topper with both hands, ensuring its perfect placement. Her gaze is fixed upwards at the tree's pinnacle, guiding her precise and controlled hand movements.\n\nOnce the topper is secured, her hands gracefully descend. Her left hand comes to rest by her hip, while her right hand briefly and gently touches a lower ornament, a final, almost affectionate check of her decorating. The brief touch reveals a tactile and affectionate engagement with her handiwork. Her expression softens slightly, suggesting quiet satisfaction and contentment, though she remains composed.\n\nThe setting is an interior room, likely a cozy living space, enveloped in a soft, warm glow primarily emanating from the illuminated Christmas tree. To the left, light-colored curtains hang, suggesting a window beyond, while the wall to the right is plain and light-toned. The camera maintains a static, eye-level, medium-full shot, capturing the woman and the Christmas tree from approximately the waist up. There are no camera movements, cuts, or transitions, creating an uninterrupted and steady view. The overall atmosphere is intimate, tranquil, and distinctly festive, imbued with the comforting warmth of the holiday season.",
+  "prediction": "The video captures a serene and intimate moment of holiday preparation, focusing on a young woman meticulously decorating her Christmas tree. The scene is set indoors, likely in a living room or bedroom, with light-colored walls and a window covered by sheer white curtains that diffuse natural light, creating a soft, ambient glow. A white radiator is visible on the right wall.\n\nThe central object is a lush green Christmas tree, adorned with numerous small, warm-toned lights that cast a gentle, inviting glow. Red spherical ornaments, some with gold accents, are scattered throughout the branches, complementing the vibrant blue and silver tinsel that drapes around its lower half. The woman, dressed in a dark, long-sleeved top and a knee-length black skirt, wears a classic red Santa hat with a white brim and pom-pom, adding to the festive atmosphere.\n\nShe stands beside the tree, her posture relaxed yet attentive. With focused concentration, she reaches up with her left hand to carefully place a slender, dark blue or purple pointed ornament onto the very top of the tree, just below the star. Her movements are deliberate and precise, reflecting care and attention to detail. After securing the ornament, she briefly withdraws her hand before extending it again to make a final adjustment, ensuring the ornament is perfectly positioned. Her gaze remains fixed on the tree, conveying a sense of quiet satisfaction and enjoyment in the task at hand. The camera maintains a static medium shot from an eye-level perspective, framing the woman and the Christmas tree centrally, allowing for uninterrupted observation of this peaceful and joyful holiday ritual.",
+  "avg_length": 266
+}

prediction_results/iter2000_sampled_85k/1088836-Split-001-Start-370-End-587+cropped_split_0.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/1088836-Split-001-Start-370-End-587+cropped_split_0.mp4",
+  "gt_caption": "The video opens in a bright, contemporary architectural design office, focusing on a moment of professional dedication and collaborative creativity. A young African woman with neatly braided hair, dressed in a crisp light blue button-up shirt, stands attentively studying a large blueprint, her brows slightly furrowed in concentration. Her right hand grips the bottom edge of the blueprint, while her left hand holds the top-middle part, ensuring it's held securely. The office features muted blue walls adorned with abstract white linear patterns suggestive of architectural schematics, complemented by lush green potted plants and modern office equipment. Sunlight streams in from large windows, illuminating the scene and contributing to a clean, spacious feel.\n\nIn the foreground, a collection of stark white architectural models forms a miniature cityscape on a large table. A tall, slender, elegantly fluted skyscraper model stands out, flanked by smaller, cuboid and rectilinear building blocks of varying heights, emphasizing their structural forms. In the background, other professionals are subtly active; one handles a large blueprint on a drafting table, while another is seated at a desk, engaged in computer work, adding to the productive and collaborative atmosphere.\n\nAfter a few moments, the woman's gaze slowly lifts from the blueprint, making direct and confident eye contact with the camera. Simultaneously, a broad, warm smile blossoms on her face, transforming her focused expression into one of approachable professionalism. Her head lifts, and a subtle relaxation in her shoulders accompanies her smile, radiating an aura of ease and confidence. This transition communicates a positive acknowledgment of the observer and suggests pride in her work, encapsulating the video's theme of progress and innovation in architectural design. The camera remains static in a medium, eye-level shot, keeping the architectural models prominently framed at the bottom, grounding the scene and emphasizing the nature of the work.",
+  "prediction": "The video opens with a medium shot of a modern architectural office, immediately establishing a theme of innovation and collaboration within the design industry. The camera remains static at eye-level throughout, providing a direct and engaging perspective. In the foreground, a meticulously crafted white architectural model dominates the scene. It features a prominent, slender skyscraper with a fluted or ribbed facade, flanked by smaller, block-like structures, all resting on a flat base. Behind the model, a large, light-colored blueprint is held by an African American woman.\n\nThe woman, dressed in a smart, light yellow collared shirt, stands poised and upright, holding the blueprint with both hands. Her dark hair is neatly styled, possibly in braids or twists, pulled back from her face. Initially, she looks down at the blueprint with focused concentration, her brow slightly furrowed. Then, she slowly lifts her head, her gaze shifting to meet the camera directly. As she does so, her expression transforms into a warm, genuine smile that crinkles the corners of her eyes, conveying approachability and confidence. This shift in expression is the primary temporal event, marking a transition from internal focus to outward engagement.\n\nIn the background, other colleagues are visible, some standing and reviewing documents, others seated at desks working on computers. The office environment is bright and airy, illuminated by natural light streaming through large windows. The walls are a clean, light color, adorned with abstract geometric art, contributing to a contemporary aesthetic. The overall mood is one of quiet productivity and collaborative energy, suggesting a positive and conducive workspace for creative work. The presence of multiple individuals engaged in their tasks reinforces this sense of a busy yet organized environment. The woman's posture remains steady and professional as she holds the blueprint, her movements deliberate and controlled, reflecting a calm and composed demeanor.",
+  "avg_length": 303
+}

prediction_results/iter2000_sampled_85k/1098658-Split-000-Start-10-End-459+cropped_split_1.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/1098658-Split-000-Start-10-End-459+cropped_split_1.mp4",
+  "gt_caption": "The video opens with a serene, static medium shot of a young woman meditating outdoors, capturing a moment of tranquility and mindfulness against a bright, natural backdrop. The woman, slender and athletic, is the focal point. She is dressed in a light green, short-sleeved crew-neck t-shirt and form-fitting dark blue leggings, paired with light-colored athletic shoes. Her dark hair is neatly pulled back, revealing clear skin and a natural look. She is seated in Baddha Konasana (Butterfly Pose) on a dark, flat surface, possibly a yoga mat or low platform resting on a light wooden boardwalk. Her spine is straight, and her shoulders are relaxed, showcasing excellent posture.\n\nThroughout the video, she maintains a deeply meditative pose, her stillness underscoring her inner calm and focused attention. Her hands are elegantly positioned in Gyan Mudra, with the tips of her thumbs and index fingers lightly touching, resting on her knees. Her face exhibits a profound sense of peace; her eyes are closed, indicating deep concentration. She maintains a constant, placid demeanor, with no discernible facial movements or verbal communication. Her entire body language emanates calmness and control, demonstrating a disciplined and mindful approach to her practice.\n\nThe background features lush green palm trees with brown trunks, creating a fresh and open atmosphere. Light-colored buildings are visible beyond the trees under a clear, bright blue sky. Blue railings or fences are also discernible in the mid-ground, suggesting a public park or promenade near a coastal area. The abundance of natural sunlight suggests a warm and pleasant day, further enhancing the peaceful environment. The presence of the wooden boardwalk and implied sand suggests a beachside park setting, though distant urban architecture indicates that the location is not entirely isolated. The camera remains static, at eye-level or slightly lower, maintaining a steady focus on the woman\u2019s meditation and reinforcing the stillness of the scene, creating a stable and contemplative viewing experience. The overall atmosphere is one of tranquility, natural beauty, and calm, offering a serene escape.",
+  "prediction": "The video opens with a serene scene of a young woman meditating outdoors, embodying tranquility and mindfulness. She sits cross-legged on a light-colored wooden deck, likely a boardwalk or patio, surrounded by tall palm trees with rough, textured trunks and vibrant green fronds. In the background, hints of urban architecture suggest a coastal town or resort area, with distant buildings and a yellow railing visible. The bright, natural lighting indicates a clear day, possibly morning or late afternoon, casting soft shadows.\n\nThe woman, appearing to be in her early to mid-teens, has dark hair neatly pulled back into a low bun or ponytail. She is dressed in comfortable athletic wear: a plain, light beige or off-white short-sleeved t-shirt, form-fitting brown leggings, and clean white sneakers with light laces. A subtle red bracelet adorns her left wrist. Her posture is impeccable, with an erect spine and relaxed shoulders, demonstrating good physical control. Her hands rest gently on her knees, palms up, in the Gyan Mudra, symbolizing wisdom and knowledge.\n\nThroughout the video, the camera remains static, employing a medium shot that frames the woman from just above her head to her feet. The eye-level angle creates a direct and intimate perspective. The woman maintains unwavering stillness, her eyes gently closed, conveying deep concentration and inner peace. There are no discernible movements, gestures, or shifts in her posture, emphasizing the sustained nature of her meditation. The overall atmosphere is one of profound serenity and calm, inviting contemplation and relaxation.",
+  "avg_length": 247
+}

prediction_results/iter2000_sampled_85k/1114415-Split-000-Start-10-End-266+cropped_split_0.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/1114415-Split-000-Start-10-End-266+cropped_split_0.mp4",
+  "gt_caption": "The video presents a serene scene of a medium-sized duck, likely a female mallard, gracefully gliding across a calm body of water, creating a tranquil and captivating moment in what appears to be an urban park environment. The duck, covered in a mosaic of earthy brown tones with darker grey and subtle cream speckles, floats effortlessly with its body resting low in the water. Its bright, almost luminous yellowish-olive beak tapers gently from its slightly darker brown head, and its dark, intelligent eyes peer calmly forward. The duck propels itself from right to left across the frame at a steady, unhurried pace, paddling subtly beneath the surface, evidenced by the concentric ripples that emanate around its body. Its head is held erect, maintaining a calm composure and natural alertness as it navigates the water, displaying no signs of distress or excitement. No vocalizations are heard, adding to the tranquil atmosphere. The background is rendered in a soft, dreamy bokeh, blurring lush green foliage, possibly trees or bushes, and faint shapes that could be distant park visitors or structures. The soft, diffused lighting, suggesting an overcast day or early morning/late afternoon, contributes to the peaceful and natural atmosphere. The camera employs a static, medium close-up shot, maintaining a low angle almost at water level, which immerses the viewer into the duck's perspective. The shallow depth of field keeps the duck in sharp focus while blurring the immediate water and background, isolating the subject and creating a soft, artistic aesthetic.",
+  "prediction": "The video presents a serene scene of a single, medium-sized mallard duck gracefully floating on calm water, likely within a park or natural pond setting. The duck, the central figure, is characterized by its mottled brown and grey plumage, with a darker, almost black head featuring a distinct white ring around its neck and a small, bright yellow beak. Its body is covered in intricate patterns of brown and black feathers, creating a speckled effect. The duck maintains a stable, horizontal posture, its body perfectly balanced on the water's surface, demonstrating its natural buoyancy.\n\nThe water itself is a muted green-grey, reflecting the soft, diffused light of an overcast day, creating gentle ripples that subtly shift the duck's position. The background is intentionally blurred, suggesting a park-like environment with distant trees and indistinct human figures, contributing to the overall atmosphere of tranquility and natural beauty.\n\nThroughout the video, the duck remains largely stationary, occasionally making slight adjustments to its head or body, perhaps scanning its surroundings. Its expression appears calm and alert, its dark eyes fixed forward, conveying a sense of quiet awareness. There are no vocalizations or overt communicative gestures; instead, its presence is one of quiet contemplation. The camera employs a static, eye-level shot, maintaining a consistent focus on the duck while utilizing a shallow depth of field to blur the background, effectively isolating the subject and enhancing the peaceful mood. The absence of camera movement further contributes to the stillness and contemplative nature of the scene.",
+  "avg_length": 249
+}

prediction_results/iter2000_sampled_85k/1150355-Split-000-Start-10-End-459+cropped_split_1.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/1150355-Split-000-Start-10-End-459+cropped_split_1.mp4",
+  "gt_caption": "In a cozy and inviting living room, a group of friends share a moment of collective exhilaration, likely triggered by a thrilling sports match or an intense show. The atmosphere transitions from calm anticipation to boisterous celebration. The room is furnished with a plush, grey fabric sofa adorned with striped throw pillows, and a dark, rectangular coffee table. On the coffee table sits a clear glass bowl overflowing with bright orange potato chips. The lighting is soft and consistent, illuminating the scene without harsh shadows.\n\nThe central figure, seated on the sofa, wears a striking red and dark-hued plaid shirt and has long, light brown hair cascading down her back. Initially, she holds a dark-colored remote control in her left hand and an amber-colored glass bottle, likely containing a beverage, in her right, exhibiting a relaxed yet attentive posture. As the scene unfolds, a palpable shift in energy occurs. The central figure and the individual beside her erupt in enthusiasm. Simultaneously, she thrusts both arms upwards in a gesture of celebration, brandishing the remote control like a trophy in her left hand while her right hand, still clutching the bottle, mimics the celebratory gesture. This synchronized action is mirrored by another friend's excited fist pump. The overhead, static camera angle provides a bird's-eye view, capturing the entire tableau and accentuating the synchronized reactions of the friends, framing their collective excitement. A vibrant green plant in a simple white pot adds a touch of life to the background.",
+  "prediction": "The video captures a moment of shared excitement among friends watching a sports event on television, likely a football match or similar game. The scene unfolds in a cozy living room with light-colored wooden flooring and a dark grey sofa where the friends are seated. A sleek, black flat-screen television dominates the background, displaying a vibrant bowl of golden-yellow potato chips. The lighting is soft and ambient, creating a relaxed atmosphere.\n\nThe main character, a young woman with long, light brown hair styled in two braids, sits centrally on the sofa. She wears a red, blue, and white plaid shirt, suggesting casual comfort. Initially, she holds a black remote control in her right hand and a blue glass bottle in her left, her expression one of focused anticipation as she watches the game. Her posture is relaxed yet attentive, leaning slightly forward.\n\nAs the game progresses, her body language shifts dramatically. Her left hand, previously holding the bottle, clenches into a fist, signaling rising tension. Then, both hands rise in unison, fists clenched, in a powerful gesture of victory. This culminates in a triumphant yell, her mouth open wide, conveying pure elation. Her eyes are fixed intently on the television screen, reflecting her deep engagement with the unfolding action. The other friends mirror this reaction, their own celebratory gestures reinforcing the collective joy. The camera maintains a static, high-angle shot throughout, providing an overhead view that emphasizes the group's dynamic and the central role of the television in their shared experience.",
+  "avg_length": 251
+}

prediction_results/iter2000_sampled_85k/1220402-Split-000-Start-10-End-322+cropped_split_1.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/1220402-Split-000-Start-10-End-322+cropped_split_1.mp4",
+  "gt_caption": "In a serene and well-lit living room, a woman with long, straight black hair is deeply engrossed in a multi-person video call on her sleek, black smartphone. The camera, positioned in a stable medium shot, captures her from the waist up, focusing on her interaction with the device. The phone, held horizontally, displays a brightly illuminated grid of approximately nine to twelve individuals, a digital mosaic connecting her to others remotely. The scene highlights the contemporary shift toward digital engagement within the comfort of home.\n\nThe woman, seen mostly from behind and slightly to the side, wears a light-colored top and a delicate gold-toned ring on her right hand, suggesting a casual and comfortable setting. Her left hand steadily holds the smartphone, while her right hand rests gently near her collarbone, conveying a sense of ease and introspection. The overall atmosphere is quiet and contemplative, with soft, diffused indoor light bathing the room.\n\nThe background, intentionally blurred, features a large flat-screen television atop a clean, white media console and a warm, reddish-brown brick wall, subtly adding texture to the domestic environment. The camera remains static, with a shallow depth of field that isolates the subject, drawing attention to her and the digital screen. While her direct expression is unseen, her unwavering focus on the smartphone implies an attentive and engaged demeanor, her body language entirely oriented towards the phone, reinforcing her deep immersion in the digital interaction. There are no dramatic movements; the action is confined to the subtle, ongoing presence of the video call participants on the phone's screen, suggesting a continuous conversation or presentation unfolding digitally.",
+  "prediction": "The video depicts a woman participating in a virtual meeting or video call on her smartphone, highlighting the integration of technology into daily life for remote communication and connection. The scene is set in a modern living room with a light blue sofa where the woman sits comfortably, her back to the camera. A white entertainment unit with open shelves stands behind her, holding decorative items and a small, dark-colored speaker. A large flat-screen television displays a colorful image, possibly a cartoon or animated show, in the background.\n\nThe woman has long, straight, dark hair cascading down her back and wears a light-colored, possibly white or cream, collared shirt. She holds a sleek, black smartphone horizontally, its screen displaying a grid of multiple video feeds from other participants. Her left hand supports the phone from below while her right hand rests casually on her lap. Her posture is relaxed yet attentive, indicating her engagement with the virtual meeting.\n\nThroughout the video, the woman remains still, her gaze fixed on the smartphone screen, suggesting active listening and participation. Her facial expression is not visible, but her posture conveys focus and attentiveness. There are no discernible gestures or movements, emphasizing her quiet observation of the ongoing virtual interaction. The lighting is soft and ambient, creating a calm and inviting atmosphere. The camera maintains a static medium shot, positioned slightly over her right shoulder, providing an over-the-shoulder perspective that frames her and the smartphone as the central focus. The overall mood is one of quiet domesticity and modern connectivity.",
+  "avg_length": 256
+}

prediction_results/iter2000_sampled_85k/1232898-Split-000-Start-10-End-298+cropped_split_1.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/1232898-Split-000-Start-10-End-298+cropped_split_1.mp4",
+  "gt_caption": "The video opens with a static, medium-close shot capturing a bustling yet organized medical or administrative reception area. Three professional women are engaged in collaborative work around a dark reception desk, on which sits a white tissue box, a clear glass of water, a white landline phone, and a black mesh cup filled with pens and pencils. The camera maintains an eye-level perspective, fostering a sense of direct observation.\n\nThe primary focus is a sleek black computer monitor displaying a uniform green screen with white plus-sign tracking markers, indicating a chroma key setup ready for digital content. A standard black keyboard sits beneath it, with an African American woman positioned with her back mostly to the camera on the right, acting as the main character. She has her dark, braided hair neatly tied back and is attired in a dark brown or maroon blazer over a lighter top, with a subtle gold earring visible. Her head is slightly bowed, suggesting intense focus on the keyboard and screen. Her right hand is dynamically active, rapidly typing on the keyboard, while her left hand rests nearby, ready for commands.\n\nParallel to her, the middle character, a woman in a lab coat, is actively conversing and gesturing with a pen. To the left, the woman in blue scrubs diligently writes or reviews documents on a clipboard, occasionally looking up to participate in the dialogue. This synchronized interplay of actions paints a picture of a well-coordinated team managing multiple tasks. Her body language and subtle nods imply active listening, even without verbal contribution. The overall demeanor of the main character projects calm professionalism and dedicated concentration.\n\nIn the background to the right, a large, multi-functional white and grey printer/copier stands, emphasizing the office's operational capacity. The setting features stark, clean white walls under bright overhead lighting, creating a sterile atmosphere characteristic of healthcare facilities. The overall mood is one of subdued activity and focused productivity, devoid of clutter, fostering an efficient work environment. There is no camera movement, maintaining an uninterrupted focus on the subtle dynamics within the reception area. The underlying theme suggests efficiency, shared responsibilities, and the seamless interaction required to manage administrative and patient care tasks, capturing a moment where technology and human collaboration intersect.",
+  "prediction": "The video depicts a scene inside a modern medical clinic reception area, focusing on the interaction between a receptionist and two medical professionals. The reception desk, a dark grey or black surface, is neatly arranged with office supplies including a white tissue box, a clear glass of water, a pen holder filled with writing instruments, and a landline telephone. A prominent computer monitor displaying a vibrant green screen with white tracking markers sits centrally, suggesting its use for digital content overlay.\n\nThe receptionist, an African American woman with her hair neatly pulled back into a bun, wears a professional dark blue blazer over a lighter top, complemented by small, elegant earrings. She maintains a focused and composed expression, her eyes primarily directed towards the medical professionals. Her posture is upright and attentive as she types on the keyboard with both hands, her fingers moving with precision. Her head occasionally tilts slightly, indicating active listening to the ongoing conversation between the two medical staff members seated at the counter.\n\nThe two medical professionals are engaged in a discussion, occasionally glancing at each other and the green-screen monitor. The receptionist's demeanor suggests a calm and efficient approach to her duties, maintaining a professional distance while actively participating in the workflow. The background features light-colored walls and glimpses of medical equipment, reinforcing the clinical setting. The lighting is bright and even, contributing to a clean and functional atmosphere. The camera remains static throughout, employing a medium shot that frames the receptionist from the waist up, along with the medical professionals and the computer monitor, providing a clear view of their interactions and workspace. There are no camera movements, zooms, or cuts, allowing the viewer to observe the scene unfold naturally.",
+  "avg_length": 287
+}

prediction_results/iter2000_sampled_85k/1280999-Split-000-Start-10-End-381+cropped_split_0.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/1280999-Split-000-Start-10-End-381+cropped_split_0.mp4",
+  "gt_caption": "Against a breathtaking golden sky, indicative of either sunrise or sunset, a video captures a timeless romantic moment between a couple silhouetted in a medium close-up shot. The scene, filmed from an eye-level angle, focuses on their loving embrace, emphasizing intimacy and a shared connection. Subtle vertical lines, potentially rigging from a boat, suggest a maritime setting, adding a sense of adventure.\n\nThe silhouetted figures, devoid of distinct color or texture due to intense backlighting, become iconic representations of love. The male figure, taller and broader, stands to the right, wearing what appears to be a collared, long-sleeved shirt, and his hair is short and neat. He embraces the female figure on the left, who wears a light-colored, sleeveless dress, possibly a wedding gown, and her long, wavy hair falls gently over her shoulders. His left hand gently cradles her waist or lower back, drawing her close, while his right hand rests softly on her left shoulder or upper arm. The embrace is reciprocated, signifying mutual comfort and connection.\n\nThough their faces are indiscernible, their heads are tilted towards each other, suggesting an impending kiss or a profound closeness. Their body language communicates tender affection, mutual adoration, and a deep emotional bond as they focus entirely on each other, lost in their shared moment. Minimal actions occur, with the couple maintaining their loving posture, exhibiting only subtle shifts in weight within their tight hold.\n\nThe background features wispy clouds drifting lazily across the sky, catching the golden light and enhancing the warm, ethereal atmosphere. The overall mood is incredibly romantic, peaceful, and deeply intimate, enhanced by the natural beauty of the sky and the stillness of the moment. The camera remains static throughout the brief clip, with no zooms or cuts, allowing the viewer to experience a timeless, undisturbed glimpse into the couple's private moment of love.",
+  "prediction": "Against a breathtaking twilight sky, a couple shares an intimate moment of profound connection and romantic bliss. The scene is bathed in the soft, ethereal glow of dusk, with a gradient of deep blue transitioning to lighter hues near the horizon, punctuated by scattered, wispy clouds. The camera, employing a static medium close-up shot from a slightly low angle, focuses on the couple's silhouettes, emphasizing their closeness.\n\nThe woman, with long, flowing hair cascading down her back, wears what appears to be a light-colored dress or top, possibly white or cream, with a delicate texture. Her left hand rests gently on the man's shoulder, while her right hand is placed tenderly on his lower back, drawing him closer. The man, dressed in a light-colored polo shirt, likely yellow or beige, has his right arm wrapped around her waist, pulling her into a secure embrace. His left hand is positioned on her upper back, completing their intertwined posture.\n\nTheir faces are pressed together, their foreheads touching, conveying a deep sense of intimacy and mutual adoration. The woman's eyes are closed, suggesting a state of blissful surrender, while the man's eyes are also closed, indicating a shared moment of peace. Their bodies are pressed intimately against each other, their postures relaxed yet engaged, reflecting a comfortable and loving bond. The overall atmosphere is one of tranquility, romance, and quiet beauty, evoking feelings of peace, love, and a dreamlike escape. The absence of camera movement further emphasizes the stillness and intimacy of the moment, allowing the viewer to fully absorb the emotional depth of the scene.",
+  "avg_length": 263
+}

prediction_results/iter2000_sampled_85k/1293461-Split-000-Start-10-End-369+cropped_split_0.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/1293461-Split-000-Start-10-End-369+cropped_split_0.mp4",
+  "gt_caption": "The video opens in a contemporary, comfortable home office, showcasing an African American woman with medium-length, dark brown dreadlocks with reddish undertones, wearing clear-rimmed eyeglasses and a casual, olive-green collared shirt. Initially, she is deeply engrossed in her work at a light-colored desk, focused on a sleek, dark grey laptop. Her fingers move rhythmically across the keyboard as she types, her brow slightly furrowed in concentration, lips gently pressed together and her gaze fixed downwards. A white spiral-bound notebook and a small, vibrant green succulent plant sit to the left of the laptop. The background, featuring an exposed brick wall, includes a tall, silver telescope on a tripod, a multi-tiered bookshelf filled with books of various sizes and colorful spines, and a large, dark television screen. A light-colored sofa with a bright red spherical cushion hints at a living area. After three seconds, the woman subtly lifts her head, her typing ceases, and her eyes shift to meet the camera\u2019s lens. Her hands, previously engaged in quick and precise typing, gently come to rest over the keyboard. A warm, genuine smile gradually spreads across her face, softening her expression and transforming her demeanor from one of focused concentration to one of friendliness and approachability. Her eyes, magnified by her glasses, meet the viewer's gaze, creating a sense of welcome and personal connection. The static, eye-level, medium-close-up shot, framed from her chest upwards, utilizes a shallow depth of field to keep the woman and her laptop in sharp focus, while the background remains gently blurred. The lighting is bright and airy, enhanced by natural light entering from a window to the left, contributing to an overall mood of quiet productivity, warmth, and intellectual curiosity.",
+  "prediction": "The video opens with a serene scene of a young Black woman working on her laptop at home, highlighting the integration of technology into daily life and the potential for remote work or study. She is seated at a clean, white desk, upon which sits a sleek, silver-grey laptop, its screen facing away from the camera. To the left of the laptop is a small, vibrant green succulent plant in a dark pot, while to the right, a minimalist black desk lamp stands ready. The background features a rustic red brick wall adorned with framed pictures and shelves holding various items, including a large telescope on a tripod, suggesting an interest in astronomy.\n\nInitially, the woman is deeply engrossed in her work, her fingers moving rhythmically across the keyboard. She wears clear-framed glasses and a short-sleeved, olive-green collared shirt with subtle vertical stripes. Her long, dark dreadlocks cascade over her shoulders. Her expression is one of calm concentration, her brow slightly furrowed as she focuses on the screen. After a few moments, she pauses typing, lifts her head, and makes direct eye contact with the camera. This transition is marked by a subtle shift in her facial muscles, culminating in a warm, genuine smile that radiates friendliness and approachability. Her eyes, magnified by her glasses, convey sincerity and openness. Throughout this interaction, she maintains an upright posture, her hands resting gently on the laptop's palm rest, conveying a sense of ease and control. The lighting is bright and even, creating a comfortable and inviting atmosphere. The camera remains static in a medium shot, framing the woman from the waist up at eye-level, allowing viewers to observe her actions and expressions without distraction. The overall mood is one of quiet productivity and domestic comfort.",
+  "avg_length": 293
+}

prediction_results/iter2000_sampled_85k/1302467-Split-000-Start-10-End-273+cropped_split_0.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/1302467-Split-000-Start-10-End-273+cropped_split_0.mp4",
+  "gt_caption": "The video presents a young woman embodying a bold and contemporary fashion aesthetic. She stands confidently against a seamless, solid hot pink wall that perfectly matches her long, straight, brightly hued fuchsia-pink hair, complete with blunt bangs. Her makeup is equally striking, featuring dramatic winged eyeliner, defined brows, and deep berry lipstick that complements her hair and the background. She is clad in a sleek black faux leather biker jacket, layered over a classic black-and-white horizontally striped long-sleeved top and dark wash jeans, which are partially visible. A small, subtle ring adorns her right hand.\n\nThe video opens with the woman elegantly holding a single strand of her vibrant pink hair with her right hand, gently pulling it outwards. Her gaze is primarily directed upwards and to the left, suggesting a pensive state. She maintains a poised and somewhat enigmatic expression, with her lips subtly parted, projecting an air of detached coolness. This initial pose is held for several seconds, conveying a moment of introspection. Then, she slowly and gracefully releases the strand of hair, allowing her right arm to drop fluidly to her side, completing the subtle sequence of movement. Her left hand remains relaxed, resting naturally near her hip, partially obscured by the jacket.\n\nCaptured with a static, medium shot from the waist up at eye-level, the camera remains focused on the subject. The bright and even lighting, typical of a professional studio, ensures she is clearly illuminated without harsh shadows. The overall mood is one of contemporary style, boldness, and artistic simplicity, emphasizing self-expression through vibrant style.",
+  "prediction": "The video presents a stylish young woman with vibrant pink hair, embodying a confident and cool aesthetic against a solid, bright fuchsia or magenta background. She is dressed in a sleek black leather jacket with silver zippers, layered over a black t-shirt featuring bold white horizontal stripes across the chest and shoulders, paired with light brown or khaki trousers. Her makeup includes dark eyeliner, defined eyebrows, and a noticeable red lipstick.\n\nThroughout the video, the woman maintains a poised and confident posture, her arms crossed at the waist. Her right hand gently holds a strand of her long, straight pink hair near her temple, allowing it to sway slightly as she subtly shifts her head from side to side. Her gaze is directed upwards and slightly to the left, conveying a sense of thoughtful contemplation or quiet observation. The camera remains static in a medium shot, framing her from the waist up at eye-level, emphasizing her subtle movements and expressions within the minimalist setting. The lighting is even and soft, contributing to a clean and focused atmosphere that highlights the subject's unique style and self-assured demeanor.",
+  "avg_length": 186
+}

prediction_results/iter2000_sampled_85k/1336558-Split-000-Start-10-End-385+cropped_split_1.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/1336558-Split-000-Start-10-End-385+cropped_split_1.mp4",
+  "gt_caption": "The video opens with a medium close-up shot of a woman in her late twenties to early thirties, engrossed in a phone conversation within a bustling retail store. Her face is framed by stylish black-rimmed eyeglasses and neatly cut, shoulder-length dark brown hair. She wears a grey, possibly tweed or knitted, cardigan or light jacket over a dark top. Held to her left ear is a sleek, white smartphone, its smooth surface contrasting with the texture of her finger.\n\nThe setting, though intentionally blurred to emphasize the main character, is clearly a commercially active space with rows of shelves densely packed with products. Bright, overhead fluorescent lighting illuminates the scene. Price signs displaying numbers like \"399.00\" and \"89.00,\" alongside green decorative elements, suggest a supermarket or large department store setting.\n\nThroughout the video segment, the woman remains actively engaged in the phone call. Her lips move consistently as she speaks, her eyes, magnified by her glasses, are wide and attentive, often gazing slightly upwards or directly forward. Her expression is a nuanced blend of concentration and slight thoughtfulness, her mouth forming various shapes as she articulates words, sometimes pursing her lips in contemplation. Her posture is upright and composed, her left hand raised, firmly holding the white phone to her ear. There are no overt gestures or fidgeting, indicating a practiced ease in handling the call within the public setting.\n\nThere is minimal camera movement, perhaps a subtle push-in or a very gentle track, keeping the woman perfectly centered while the background seemingly shifts slightly. This subtly emphasizes her focus on the phone call as she navigates the busy environment. The overall mood is functional and everyday, highlighting the pervasive nature of modern communication and the seamless blending of personal interactions into public spaces.",
+  "prediction": "In a brightly lit, modern supermarket, a young woman is deeply engrossed in a phone conversation. The camera, employing a medium close-up shot from her chest upwards, focuses on her face and upper body as she walks slowly through the aisle. She holds a sleek, white smartphone to her right ear, its smooth surface contrasting with her dark-rimmed eyeglasses and textured grey and black speckled coat. Her shoulder-length, straight brown hair frames her fair complexion, which is subtly enhanced with natural-looking makeup.\n\nHer expression is one of focused concentration, her brow slightly furrowed and lips parted as if actively speaking or listening intently. Her eyes, magnified by her glasses, dart around, suggesting she's processing information or observing her surroundings. Her posture is upright and composed, her right hand holding the phone firmly against her ear, while her left arm swings naturally at her side. Her movements are minimal and deliberate, emphasizing her absorption in the call.\n\nThe background reveals rows of shelves stocked with various products, interspersed with green and blue price tags displaying numbers like \"399\" and \"89.\" The overhead fluorescent lighting creates a sterile yet functional atmosphere, typical of a large retail environment. The camera smoothly tracks her forward movement, maintaining a consistent distance and focus, creating an intimate and observational feel. The overall mood is quiet and ordinary, highlighting the integration of personal communication into everyday life within this mundane setting.",
+  "avg_length": 234
+}

prediction_results/iter2000_sampled_85k/1344300-Split-000-Start-10-End-385+cropped_split_1.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/1344300-Split-000-Start-10-End-385+cropped_split_1.mp4",
+  "gt_caption": "The video opens with a wide, high-angle aerial shot showcasing a vast, desolate landscape under a clear blue sky with faint wisps of high clouds, suggesting a bright, sunny day. The atmosphere is one of serene solitude and expansive openness. The terrain consists of dry, light brown, sandy earth, cracked and textured, indicative of arid conditions. Glistening, dark streaks, possibly wet or icy, snake across the ground, forming shallow depressions and contrasting sharply with the dry soil. In the distance, undulating hills and mountains fade into a hazy horizon, while closer, larger mounds of similar brown earth create a rugged backdrop.\n\nA lone individual, dressed in a dark, possibly maroon or deep red, long-sleeved top or jacket and dark trousers, stands with a medium-sized, dark-colored, short-haired dog atop a gentle rise. They stand still at the apex of the slope. The camera slowly tracks forward, maintaining a steady, high-angle perspective and gradually closing in on the scene.\n\nThe individual then initiates a deliberate movement, bending forward from the waist and slowly kneeling, bringing their hands close to the ground, possibly to touch or examine something. The dog remains calmly beside the person throughout, initially standing and then shifting its position slightly as the person lowers themselves. The person's interaction with the terrain suggests a quiet, focused engagement with their surroundings, implying curiosity or contemplation. The low sun angle, characteristic of either early morning or late afternoon, casts long shadows from the figures and small rises, enhancing the dramatic sense of the scene and emphasizing the isolation and grandeur of the natural setting.",
+  "prediction": "The video opens with a serene, high-angle aerial shot of a vast, desolate landscape under a clear, bright sky. The scene establishes a theme of quiet exploration and companionship as a lone figure and their small dog traverse this rugged terrain. The camera, likely mounted on a drone, begins a smooth, continuous tracking shot, following the pair from behind and slightly above.\n\nThe foreground is dominated by a textured expanse of light grey and brown rock formations, resembling cracked earth or ancient lava fields. These formations vary in size and shape, creating a sense of depth and scale. In the distance, rolling hills and mountains fade into a hazy horizon, suggesting an expansive wilderness. The sky transitions from a warm yellow-orange near the horizon to a pale blue overhead, indicating either early morning or late afternoon light.\n\nThe main character, dressed in dark, practical outdoor attire including a jacket and trousers, walks steadily away from the camera along a narrow ridge. Their posture is upright and relaxed, with both hands swinging naturally at their sides. A small, dark-colored dog trots closely beside them, occasionally pausing to sniff the ground before resuming its walk. The dog's movements are more erratic than the character's, reflecting its curiosity about the environment.\n\nThe camera maintains a consistent distance and height relative to the subjects, keeping them centered in the frame. The overall atmosphere is one of profound tranquility and isolation, emphasizing the raw beauty of nature and the peaceful journey of the two figures through it. The absence of abrupt cuts or transitions further contributes to the video's calm and contemplative mood.",
+  "avg_length": 268
+}

prediction_results/iter2000_sampled_85k/1345899-Split-000-Start-10-End-337+cropped_split_1.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/1345899-Split-000-Start-10-End-337+cropped_split_1.mp4",
+  "gt_caption": "In a serene scene of modern engagement, a young individual is deeply engrossed in their laptop, seated comfortably on a bed. The video captures a sense of focused activity within a relaxed, personal setting, highlighting the integration of technology into daily life.\n\nThe central focus is a sleek, silver laptop displaying content on its screen, resting on a soft, light grey or off-white textured blanket that covers the bed. A padded, grey headboard frames the bed. In the background, a minimalist room features a dark-stained wooden desk with a dark green office chair and a tall, dark wooden bookshelf packed with books and decorative items. The walls are painted in muted green and darker tones, creating a cozy ambiance.\n\nThroughout the video, the individual interacts continuously with the laptop. Their right hand actively manipulates the trackpad, fingers tapping and gliding across the surface. Simultaneously, their left hand rests steadily on the palm rest, providing balance. The subtle, precise movements of the right hand demonstrate sustained digital engagement.\n\nThe individual has long, flowing light brown hair with distinctive streaks of vibrant red or auburn, especially visible towards the ends. They are dressed in a large, comfortable knit sweater with cream or off-white and black horizontal stripes and dark-colored, likely black, casual pants, possibly jeans. Small rings adorn the fingers on their right hand.\n\nThe individual's face is largely obscured by their downward gaze, but their demeanor conveys deep concentration. Their eyes are fixed intently on the laptop screen. There are no discernible verbal cues or facial expressions of emotion, reinforcing the image of quiet, solitary engagement.\n\nAdopting a relaxed yet attentive posture, the individual sits cross-legged on the bed, their head slightly bowed towards the screen. The right hand exhibits deliberate and agile finger movements on the trackpad, while the left hand acts as a subtle anchor, stabilizing their posture. This suggests an experienced and comfortable user.\n\nThe setting is a modern bedroom or studio apartment, blending sleeping and functional areas. The lighting appears soft and dim, suggesting late afternoon or evening, contributing to a calm and introspective atmosphere. The textures of the blanket and the knitwear evoke warmth and coziness.\n\nThe video features a static, medium-wide shot, framed from a slightly low angle. The camera remains perfectly still, providing an unobstructed view of the main character's interaction with the laptop and their immediate surroundings.",
+  "prediction": "The video opens with a static, medium shot of a young woman seated cross-legged on a large bed, deeply engrossed in her dark-colored laptop. The bed is covered with a light grey, textured duvet or blanket, and a grey upholstered headboard is visible to the left. She wears a cozy, oversized sweater with bold horizontal stripes of white and black, paired with dark trousers or jeans. Her long, straight brown hair, with subtle reddish or pink highlights near the ends, falls over her shoulders as she focuses intently on the screen.\n\nHer expression is one of calm concentration, her eyes fixed downwards, occasionally shifting slightly as if processing information. Her right hand rests casually on the laptop's palm rest, while her left index finger repeatedly taps and points at specific areas on the screen, indicating active navigation or selection. This deliberate pointing motion suggests she is actively interacting with the content displayed, perhaps highlighting elements or making selections. Her posture is relaxed yet attentive, reflecting a comfortable yet focused engagement with her digital task.\n\nThe setting appears to be a modern, well-appointed bedroom or living space, with a dark grey wall serving as a backdrop. In the background, a wooden bookshelf filled with books and decorative items is visible, along with a desk and chairs, suggesting a home office or study area. The lighting is soft and ambient, contributing to a calm and inviting atmosphere conducive to work or relaxation. The camera remains static throughout, maintaining an eye-level perspective that allows for uninterrupted observation of the woman's actions and the details of her environment. The overall theme revolves around personal productivity and the integration of technology into daily life, highlighting a moment of focused engagement within a comfortable home environment.",
+  "avg_length": 290
+}

prediction_results/iter2000_sampled_85k/1349117-Split-000-Start-10-End-448+cropped_split_1.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/1349117-Split-000-Start-10-End-448+cropped_split_1.mp4",
+  "gt_caption": "Within a tranquil animal enclosure, the video intimately portrays a guanaco with a distinctive dark face and rich, reddish-brown coat as it interacts with a suspended hay feeder, showcasing an innovative animal enrichment technique. The large, teardrop-shaped bundle of fresh, vibrant green hay hangs suspended by a dark, net-like cord structure above light-colored gravel or sand. A rustic wooden fence with vertical brown planks forms a backdrop, partially obscuring faint tree trunks and a dark, shallow water trough to the far right.\n\nThe video begins with the guanaco observing the suspended feeder from a short distance. Its face, almost velvety black from the top of its head to its muzzle, provides a striking contrast to its reddish-brown fur. Maintaining a stable, upright posture, and leaning slightly against a wooden post on the left, the guanaco gradually and deliberately extends its remarkably long neck toward the green bundle, its eyes seemingly fixed on the food source.\n\nOnce within reach, the guanaco meticulously lowers its head, inserting its muzzle into the dense hay. With calm focus and intent, it begins to pluck and consume individual strands of grass. This precise and efficient feeding behavior causes the hanging bundle to sway gently with each tug and pull, demonstrating the guanaco's adaptation to reach otherwise inaccessible foliage. The camera, employing a static, medium shot at roughly eye-level, captures these subtle movements and interactions without any distracting pans, zooms, or cuts, maintaining focus on the guanaco and the feeder. The soft and ambient lighting contributes to a peaceful and undisturbed atmosphere within the clean, contained environment established by the light-colored gravel ground and the sturdy wooden fence.",
+  "prediction": "In a serene zoo enclosure, a large, dark brown camel with shaggy fur and a prominent hump is presented with a generous portion of fresh green grass. The scene unfolds on a light-colored sandy ground, bordered by a tall, dark blue wooden fence that separates the enclosure from the lush green foliage beyond. The lighting is bright and even, suggesting a clear day.\n\nThe focal point is a substantial, irregularly shaped bundle of vibrant green grass, suspended by thin, dark ropes from an unseen support above. The grass appears fresh and somewhat coarse, with individual blades clearly visible. Initially, the camel stands patiently, its head slightly turned towards the hanging grass. Then, with deliberate grace, it slowly extends its long neck and head, its muzzle approaching the grass. Its large, dark eyes convey a sense of calm focus as it prepares to feed. The camel's expression is one of quiet concentration, its gaze fixed intently on the grass. There is no audible communication; instead, its slow, controlled movements communicate a sense of calm anticipation and focused intent.\n\nThe camel's physicality is characterized by its impressive flexibility and control. It gracefully extends its long neck and head, demonstrating remarkable dexterity for such a large animal. Its posture remains stable throughout, showcasing its inherent strength and balance. The camera maintains a static medium close-up shot at eye-level with the camel, providing an intimate perspective of this peaceful moment of feeding. The overall atmosphere is tranquil and observational, highlighting the simple act of sustenance within a managed environment.",
+  "avg_length": 255
+}

prediction_results/iter2000_sampled_85k/1352268-Split-000-Start-10-End-345+cropped_split_1.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/1352268-Split-000-Start-10-End-345+cropped_split_1.mp4",
+  "gt_caption": "The video presents a brief observational study of a businessman navigating an urban environment, embodying a theme of professional routine and quiet confidence. The man, with a lean build and casually styled dark, slightly wavy hair, walks purposefully along a paved urban walkway adjacent to a modern building with large glass panels, some frosted at eye level. Reflections on the clear upper sections of the glass reveal distant buildings and a bright, sunny sky. He is dressed in business-casual attire: a crisp white collared shirt, unbuttoned at the neck, under a dark blazer or jacket, dark-wash jeans or smart casual trousers, and classic brown dress shoes. Dark sunglasses conceal his eyes, adding an air of detachment, while his mouth maintains a neutral expression.\n\nHe moves with an upright and confident posture, shoulders slightly back, his gait smooth and even. In his right hand, he holds a dark-colored, disposable takeaway coffee cup with a lid, gripped steadily close to his body. His left hand carries a sleek, rectangular briefcase made of dark leather or a similar durable material. The briefcase, with a minimalist and professional design, swings gently with each step.\n\nThe camera remains static throughout the video, providing a medium-wide, eye-level shot of the man as he walks from the background towards the camera and out of frame. He maintains a consistent, measured pace, his head held steadily forward. He neither stops nor interacts with anything in his environment, maintaining a sense of private purpose and embodying the quiet efficiency of someone on a familiar route. The background atmosphere is quiet and orderly, devoid of distractions, highlighting the man's solitary journey.",
+  "prediction": "A young, well-dressed businessman strides purposefully along a modern urban street, embodying a sense of quiet determination and professional efficiency. He is clad in a dark suit jacket over a crisp white dress shirt, paired with light brown or khaki trousers and polished dark shoes, presenting a sharp and contemporary appearance. His dark, slightly wavy hair is neatly styled, and he wears dark-rimmed glasses that lend him an intellectual air.\n\nHe carries a sleek, dark-colored briefcase in his left hand, its smooth surface suggesting professionalism, while his right arm swings naturally as he walks at a steady, unhurried pace. His posture is upright and confident, reflecting a composed demeanor. His facial expression remains neutral and focused, his gaze directed straight ahead, conveying concentration and self-possession. There is no verbal communication or overt emotional display; his demeanor is one of quiet resolve.\n\nThe setting is a clean, contemporary urban environment, likely a business district. To the left, a long, curved building features a striking facade of large, vertical glass panels, some frosted to obscure reflections, creating a minimalist aesthetic. The ground beneath him is paved with dark, interlocking cobblestones, adding texture to the scene. In the distance, other buildings and faint greenery suggest a bustling cityscape. The lighting is soft and diffused, indicating an overcast day, contributing to a calm and somewhat subdued atmosphere.\n\nThe camera employs a smooth, continuous tracking shot, moving backward at the same pace as the man, maintaining a consistent medium full shot that captures him from head to toe. The angle is at eye-level, providing a direct and engaging perspective. The motion is fluid and stable, creating a sense of effortless observation as the man moves steadily towards the viewer.",
+  "avg_length": 284
+}

prediction_results/iter2000_sampled_85k/1368946-Split-000-Start-10-End-385+cropped_split_0.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/1368946-Split-000-Start-10-End-385+cropped_split_0.mp4",
+  "gt_caption": "The video presents a focused scene of window cleaning, emphasizing the transition from a dirty to a potentially pristine surface. The primary objects are a standard-sized green spray bottle with an orange nozzle and trigger, and a soft, porous orange scrubber or sponge. A person, wearing a simple blue t-shirt, is visible behind the window, although their face remains indistinct and out of focus. The individual's arm and hand, displaying a medium skin tone, confidently manipulate the cleaning tools.\n\nInitially, the window is vigorously sprayed with the cleaning solution, creating foamy spatters and vertical streaks. This is immediately followed by the orange scrubber being moved across the glass, spreading the solution and initiating the cleaning process. The spraying action repeats, targeting different sections, creating new rivulets of fluid cascading downwards. The arm movements are fluid and strong, demonstrating familiarity with the task. The actions suggest concentration and dedication to the task, as the character exhibits an efficient and coordinated physical approach to window cleaning.\n\nThroughout the scene, the camera maintains a static, eye-level medium shot directly in front of the window. This perspective highlights the dynamic interplay between the tools, the cleaning solution, and the glass. Visible through the window is a verdant outdoor environment, suggesting lush trees or foliage, providing a natural backdrop. The bright, natural lighting further contributes to a fresh and clean atmosphere, supporting the mood of quiet industry and routine.",
+  "prediction": "The video opens with a medium close-up, static shot focused on a man diligently cleaning a large window. The camera is positioned at eye-level, offering an intimate perspective of his task. The man, seen from the chest up and slightly out of focus, has short, dark hair and wears a casual yellow t-shirt. He stands upright, leaning slightly towards the window, his posture conveying concentration.\n\nHe holds a vibrant green spray bottle in his right hand, its bright color contrasting against his shirt. His left hand wields a small, rectangular blue sponge, which he uses to wipe the glass. The window itself is initially covered in a dense layer of white, frothy suds, creating a textured, opaque surface. The background beyond the window is softly blurred, revealing lush green foliage suggesting a natural outdoor environment.\n\nThe sequence begins with the man spraying the window, releasing a fine mist that quickly transforms into a thick, white foam. He then meticulously wipes the glass with the blue sponge, moving it horizontally across the surface. As he moves, the suds are agitated, forming long, glistening streaks that run downwards, gradually clearing the view. This action is repeated rhythmically, demonstrating a systematic approach to cleaning. The lighting is bright and even, contributing to a calm and productive atmosphere. The overall theme emphasizes domesticity and the satisfaction of tidying one's living space.",
+  "avg_length": 227
+}

prediction_results/iter2000_sampled_85k/1405481-Split-000-Start-10-End-369+cropped_split_1.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/1405481-Split-000-Start-10-End-369+cropped_split_1.mp4",
+  "gt_caption": "The video depicts an IT specialist or data center manager immersed in the demands of a modern, interconnected work life. The scene is set in a high-tech server room or data center, where rows of large server racks, illuminated by cool blue and green LEDs, stretch into the distance, creating a sense of immense digital infrastructure. The environment is predominantly dark and bathed in a futuristic blue light, contrasting with the warmer, focused illumination on the man\u2019s immediate workspace.\n\nIn the foreground, a sturdy dark wooden desk is organized with essential work tools. A sleek silver laptop sits open on the desk, subtly reflecting the ambient light. To the left, a dark computer mouse is ready for use. On the right, a small stack of papers, possibly reports or schematics, is neatly arranged, topped with a metal clip, alongside a simple mesh pen holder. From the upper right, an industrial-style desk lamp, grey with a bright white interior, casts a warm light onto the workspace.\n\nThe protagonist, a man with a clean-shaven head and a neatly trimmed, dark beard, is formally dressed in a dark suit jacket over a light, striped dress shirt. A deep red or burgundy tie adds a touch of color to his professional attire. He is engrossed in a phone conversation, holding a dark, contemporary smartphone to his right ear while intently focusing on the laptop screen. His expression is serious and concentrated, his brow slightly furrowed, indicating deep thought. His eyes are primarily directed downwards, occasionally shifting, but never making direct contact with the camera. He maintains an upright, composed posture, reflecting self-discipline.\n\nInitially, he remains still and attentive, his gaze fixed on the laptop screen, suggesting he is processing visual information while listening. Then, he subtly shifts his attention and brings his left hand up to adjust his necktie, a small, almost unconscious gesture that breaks the intense concentration. Afterwards, he resumes his primary focus on the screen. Throughout the video, there is no visible verbal communication from him; his focus is entirely on receiving information and observation. The scene is captured with a static medium shot, framing the man from roughly his chest up, and the camera is positioned at eye-level, creating a direct and engaging perspective. There are no camera movements, reinforcing the quiet professionalism of the scene.",
+  "prediction": "The video depicts a professional man deeply engaged in a phone conversation within a modern office setting, subtly juxtaposed with a large digital projection of a server room. The scene emphasizes the intersection of human interaction and technological infrastructure, highlighting the demands of managing complex digital systems.\n\nThe man, dressed in a dark suit jacket over a light-colored, possibly striped, dress shirt and a patterned tie featuring shades of blue and red, sits at a dark wooden desk. A sleek silver laptop is open before him, accompanied by a black computer mouse to its left and a stack of papers or documents on the right. A classic black desk lamp with a conical shade illuminates his workspace from the right, casting a focused glow. He holds a dark-colored smartphone to his ear, his expression serious and concentrated as he listens intently. His brow is slightly furrowed, and his eyes are directed downwards towards the laptop screen, suggesting he's cross-referencing information. His posture is upright and attentive, leaning slightly forward, indicating deep engagement with the call.\n\nThroughout the clip, the man remains largely still, his primary action being listening. Around the 0:03 mark, he briefly lowers the phone from his ear, perhaps to adjust it or take a quick glance at the screen, before returning it to his ear by 0:04. This subtle movement underscores his active participation in the conversation. The background is dominated by the large digital projection of a server room, creating a stark contrast between the warm, focused lighting of the office and the cool, ambient glow of the data center. The server room image features rows of tall, rectangular server racks stretching into the distance, their surfaces illuminated by numerous small, glowing lights that suggest active data processing. The overall atmosphere is one of quiet efficiency and technological sophistication, conveying a sense of serious work being conducted in a high-tech environment.\n\nThe camera maintains a static medium shot throughout the video, framing the man from approximately the waist up. The angle is eye-level, providing a direct and unobtrusive view of the subject and his immediate surroundings. There are no camera movements, zooms, or cuts, allowing for an uninterrupted observation of the man's actions and the intriguing backdrop.",
+  "avg_length": 371
+}

prediction_results/iter2000_sampled_85k/141827-Split-000-Start-10-End-369+cropped_split_0.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/141827-Split-000-Start-10-End-369+cropped_split_0.mp4",
+  "gt_caption": "The video opens with a medium shot of two young women, likely sisters or close friends in their late teens or early twenties, walking forward side-by-side against a modern urban backdrop. The scene exudes youthful camaraderie and joy, establishing a theme of friendship and lightheartedness on a sunny day. To the left, one woman wears a light grey and white striped long-sleeved crop top, light blue jeans secured with a black belt, and a simple silver pendant necklace; a black bra strap is visible. Her companion sports a plain white t-shirt tied at the front to create a cropped look, layered with a blue kimono-style cardigan adorned with vibrant pink, green, and white floral patterns, and dark-colored pants. Their long, straight light-brown or blonde hair sways gently as they walk at a steady, synchronized pace.\n\nThey are engaged in an animated conversation, frequently turning to face each other with wide, genuine smiles and shared glances. Their expressions are consistently joyful, with crinkles forming around their eyes as they laugh silently. Their heads occasionally tilt or nod, emphasizing their engagement. Their shoulders almost brush as they walk closely, their arms swinging freely, reflecting their ease and comfort. Behind them, a tall building with vertical concrete or stone panels provides a minimalist backdrop, punctuated by a small, well-maintained green tree. The bright and even lighting indicates a clear, sunny day, enhancing the cheerful atmosphere. The camera follows their forward movement at a consistent distance, maintaining focus and creating a fluid viewing experience, likely facilitated by a gimbal or track.",
+  "prediction": "The video captures a heartwarming moment between two young women, likely friends or sisters, enjoying a leisurely walk outdoors and sharing genuine laughter. The scene is set against a backdrop of modern architecture featuring tall, light grey concrete pillars with vertical grooves, suggesting an urban or contemporary park environment. A small, vibrant green tree provides a touch of nature to the otherwise minimalist setting.\n\nThe woman on the left has long, flowing blonde hair that gently sways as she walks. She wears a casual outfit consisting of a light grey and white striped long-sleeved crop top, revealing a dark bra strap underneath, paired with light blue denim jeans cinched at the waist by a dark belt. A delicate silver necklace with a small pendant adorns her neck. Her companion on the right also has long, straight blonde hair, styled similarly. She is dressed in a crisp white tank top tied at the front, layered beneath a striking floral kimono-style jacket in warm earthy tones, possibly beige or brown, adorned with pink and green blossoms. Both women have fair skin and appear to be in their late teens or early twenties.\n\nThroughout the video, both women are in constant motion, walking forward at a relaxed pace. Their expressions are consistently joyful, marked by wide, genuine smiles and frequent bursts of laughter. Their eyes are bright and engaged, often looking at each other, indicating active conversation and shared amusement. They maintain close proximity, walking side-by-side, their bodies angled towards one another, emphasizing their connection. Their arms swing naturally at their sides, contributing to their relaxed gait. The camera employs a smooth, continuous tracking shot, moving forward in sync with the subjects, maintaining a consistent medium shot framing them from the waist up. The eye-level angle creates an intimate and natural perspective, drawing the viewer into their shared experience. The overall atmosphere is bright and airy, with soft, diffused lighting, creating a cheerful and optimistic mood.",
+  "avg_length": 323
+}

prediction_results/iter2000_sampled_85k/1430987-Split-000-Start-10-End-365+cropped_split_0.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/1430987-Split-000-Start-10-End-365+cropped_split_0.mp4",
+  "gt_caption": "The video captures an intimate moment of dedicated strength training in a dimly lit, potentially industrial gym setting. Dramatic shafts of natural light cut through distant windows, illuminating particles and casting striking shadows that create a moody, almost cinematic atmosphere. The focus is on a powerfully built, shirtless man performing bicep curls with substantial, heavy, black, hexagonal dumbbells, showcasing the raw beauty of the human form sculpted through intense physical effort.\n\nHe wears dark athletic shorts with a light waistband and black wrist wraps for support. A sheen of sweat glistens across his chest and abdomen, highlighting the muscular definition beneath. The camera employs a tight medium shot, often from a slightly low angle, framing his torso and arms and emphasizing his physique. The shot is largely static, allowing full appreciation of the muscular detail.\n\nThe man executes controlled bicep curls, alternating arms. He begins with his right arm, lifting the dumbbell towards his shoulder, showcasing the powerful contraction of his biceps, before slowly lowering it. Then, he repeats the movement with his left arm. Each movement is deliberate and focused, emphasizing precise muscle engagement and a steady cadence. Throughout the exercise, he maintains an incredibly stable and erect posture, demonstrating excellent core strength and control. His movements are fluid yet powerful, with no wasted motion, highlighting a disciplined approach prioritizing form.\n\nWhile his face is largely obscured, the intensity of his physical exertion communicates profound focus and determination. There's no verbal communication; instead, his entire body language speaks of concentration and unwavering resolve, a quiet battle against gravity and fatigue. The scene embodies determination, the pursuit of peak physical condition, and the quiet intensity of an athlete deeply immersed in their workout.",
+  "prediction": "The video opens with a medium close-up, static shot of a muscular man's torso and arms, immediately establishing a theme of strength training and physical discipline. The man, shirtless and revealing well-defined abdominal muscles, is performing bicep curls with two dark-colored dumbbells, likely black or deep grey, each marked with \"12KG\" in white text. He wears dark athletic shorts with a light-colored waistband. The setting appears to be an indoor gym or training space, suggested by blurred shapes resembling gym equipment or structural elements in the background. The lighting is dim, creating a dramatic atmosphere with strong backlighting that highlights the contours of his muscles and casts him in silhouette.\n\nThe sequence begins with the man holding the dumbbells at shoulder height. He then smoothly lowers them to his sides before initiating a controlled bicep curl with his right arm, raising the dumbbell towards his shoulder. After reaching the peak contraction, he slowly lowers it back down. This deliberate, rhythmic motion is repeated on his left arm, showcasing a steady and powerful execution of the exercise. His posture remains upright and stable throughout, demonstrating excellent form and control. The camera maintains its focus on the man's upper body, emphasizing the muscle engagement and the repetitive nature of the exercise, conveying a sense of quiet determination and intense concentration.",
+  "avg_length": 218
+}

prediction_results/iter2000_sampled_85k/1443340-Split-001-Start-370-End-658+cropped_split_0.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/1443340-Split-001-Start-370-End-658+cropped_split_0.mp4",
+  "gt_caption": "The video captures the rhythmic motion of a person running on a verdant woodland path, emphasizing physical activity and connection with nature. The camera, positioned low to the ground, intimately tracks the runner's legs from behind as they move forward through the dappled sunlight. The runner wears grey, fitted athletic pants that taper around the ankles, paired with black running shoes featuring thick, dark soles with visible tread, hinting at good grip for the uneven terrain. Subtle white and purple accents are visible on the sides of the shoes and accompanying socks. The ground consists of lush green grass interspersed with scattered autumn leaves in shades of brown and yellow, suggesting a late summer or early fall setting. The runner's legs move in a steady, alternating rhythm; each footfall is firm yet agile as the black soles briefly contact the grass and leaves before lifting again. This continuous forward progression demonstrates a moderate to fast pace and efficient running form. Tall, mature trees create a soft, out-of-focus backdrop, their dark trunks forming vertical lines that suggest depth and contribute to the peaceful, inviting atmosphere of the park or forest setting. The steady, following shot emphasizes the repetitive and rhythmic nature of the run, highlighting the runner's determined stride and controlled leg movements as they navigate the natural terrain.",
+  "prediction": "The video opens with a low-angle, close-up shot of a person jogging through a serene park or forest during autumn. The camera, positioned just above ground level, tracks the runner's legs and feet as they move forward, emphasizing the rhythmic motion of their stride. The runner is wearing dark athletic shoes with visible laces and soles, paired with light grey sweatpants or joggers that fall loosely around their ankles. The ground is covered in vibrant green grass interspersed with numerous fallen leaves in shades of yellow, brown, and orange, indicating the season. The runner's legs move in a consistent, alternating pattern, lifting and swinging forward with each step, demonstrating a steady pace. The background features blurred tree trunks and foliage, suggesting a natural, wooded environment. Soft, diffused lighting, possibly from an overcast day or early morning/late afternoon sun, creates a calm and peaceful atmosphere. The focus remains sharp on the runner's legs and feet, while the background is gently blurred, drawing attention to the continuous, fluid motion of the run. The overall theme is one of active outdoor recreation and the simple pleasure of movement within a natural setting.",
+  "avg_length": 189
+}

prediction_results/iter2000_sampled_85k/1443847-Split-001-Start-386-End-674+cropped_split_1.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/1443847-Split-001-Start-386-End-674+cropped_split_1.mp4",
+  "gt_caption": "The video portrays a poignant scene inside a sterile hospital room, focusing on an elderly male patient resting in bed. The overarching theme centers on vulnerability, medical care, and quiet contemplation during illness and recovery, emphasized by the presence of two medical professionals at his bedside. The camera remains static in a medium, eye-level shot, offering a direct and intimate view of the scene.\n\nThe patient, with short, neat, receding grey hair and a fair complexion bearing the marks of age, lies supine on the hospital bed, covered up to his chest with a soft, light blue blanket. He wears a light-colored hospital gown with subtle vertical stripes. His posture is relaxed, with his hands gently interlocked on his abdomen, contributing to the peaceful atmosphere. The bed itself is neatly made with light blue linens and a crisp white pillow. His eyes are open, directed slightly upwards and to the left, conveying a sense of deep thought or internal reflection, with no visible signs of discomfort. He remains completely still and silent.\n\nThe background reinforces the clinical environment. The light-colored walls, likely white or off-white, and the bright, uniform lighting enhance the clean, somewhat austere atmosphere. Visible medical equipment includes clear IV bags hanging from poles, thin tubes extending from them, and a white, rectangular medical monitor with dials and buttons. The medical staff near the bed also maintain minimal movement, their presence suggesting quiet observation or consultation. The overall mood is serious and contemplative, reflecting the solemnity and care inherent in the setting. There are no camera movements or transitions, allowing the viewer to focus on the patient's state.",
+  "prediction": "The video depicts a quiet, contemplative scene within a modern hospital room, focusing on an elderly male patient resting in his bed. The overall theme emphasizes patient care and the quiet dignity of recovery.\n\nThe patient lies supine in a standard hospital bed with a white pillow and a light yellow blanket covering him from the waist down. He is dressed in a light-colored, possibly striped, hospital gown. His short, neatly combed grey hair and fair complexion suggest advanced age. His hands are clasped together at chest level, fingers intertwined, conveying a sense of calm or introspection. His eyes are open but unfocused, gazing slightly upwards and to the right, suggesting he is lost in thought rather than actively observing his surroundings. His facial expression is neutral, reflecting a state of rest or mild contemplation. He remains still throughout the clip, reinforcing the tranquil atmosphere.\n\nOverhead, medical equipment is visible, including a blue blood pressure cuff attached to his left arm and a clear plastic bag connected by tubing, likely for oxygen or intravenous fluids. To the right, a medical monitor or pump unit stands on a stand, connected by various wires and hoses. The background consists of a clean, sterile environment with light-colored walls and a transparent glass partition, typical of a hospital room. The lighting is bright and even, contributing to a clinical yet reassuring atmosphere.\n\nThe camera maintains a static medium shot, framing the patient from the waist up and capturing the surrounding medical equipment. The angle is slightly low, looking up at the patient, which subtly emphasizes his presence. There are no camera movements, cuts, or transitions, allowing the viewer to observe the scene without interruption. The patient's posture and stillness, combined with the absence of any movement or interaction, create a sense of suspended animation, emphasizing the quiet nature of the scene.",
+  "avg_length": 307
+}

prediction_results/iter2000_sampled_85k/1557739-Split-000-Start-10-End-279+cropped_split_0.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/1557739-Split-000-Start-10-End-279+cropped_split_0.mp4",
+  "gt_caption": "In a serene, agricultural landscape, a man takes a moment of quiet contemplation atop a large, rectangular hay bale densely packed with golden-yellow dried straw. Dressed in a collared, long-sleeved shirt and trousers of muted, earthy tones like beige or light brown, suggesting sturdy, practical work attire, the man initially sits with his back to the viewer. He has dark hair styled in a short bun or top knot, visible when he removes his hat. The camera, in a stable, medium shot, frames the man and his immediate surroundings, capturing the scene's contemplative mood.\n\nThe sequence begins with him already seated, his right hand raised, first adjusting and then deliberately removing a dark, possibly black or navy, baseball cap with a white inner lining or mesh. He carefully places the cap on the hay bale beside him, showing its curved brim and structured form. Afterwards, he brings his right hand to his forehead, perhaps shielding his eyes from the bright, partly cloudy sky or wiping away sweat, before settling into a still posture. His face, now more visible, reveals a focused and possibly thoughtful gaze directed towards the distant horizon.\n\nThe setting is a sprawling, golden field recently harvested, evidenced by stubble and scattered straw bales, under the warm, sunny sky. Distant, undulating hills define the horizon, adding depth to the rural, pastoral atmosphere. Other similar hay bales are scattered across the expansive field. The lighting is natural and clear, casting soft shadows that enhance the tranquil and expansive mood of freedom and open space. The man maintains a relaxed, cross-legged posture throughout, indicating comfort and ease. His actions \u2013 removing and placing the hat \u2013 are performed with calm, unhurried grace. The gesture of raising his hand to his forehead suggests either shielding his eyes from the sun or a moment of quiet reflection, conveying a sense of calm and being at peace in the moment. There is no verbal communication, and the overall demeanor is one of quiet introspection, absorbing the vastness of the landscape. The camera remains mostly static, contributing to the unhurried feel and maintaining focus on the man against the sweeping backdrop.",
+  "prediction": "In a serene and expansive rural landscape, a young man finds solace on a large, rectangular hay bale, its golden-brown texture contrasting with the dry, golden-green grass of the field. The scene is bathed in bright, natural light under a partly cloudy sky, creating an atmosphere of tranquility and openness. Distant, rolling hills or mountains form a hazy horizon line, adding depth to the vista.\n\nThe man, seen from behind, has short, dark hair and wears a light blue, long-sleeved button-up shirt and light-colored trousers, suggesting practical attire for the farm environment. He sits comfortably on the hay bale, his posture relaxed yet attentive. Initially, he raises both hands to adjust his hair, then lowers them to place a dark, possibly black or deep brown, baseball cap with a visible brim onto the bale beside him. His right hand briefly touches the cap before resting on his lap, while his left hand remains on his leg.\n\nAfter placing the cap, the man slowly turns his head and upper torso to his right, his gaze sweeping across the vast field. This deliberate movement suggests a moment of quiet contemplation as he takes in the panoramic view. The camera maintains a static, medium-long shot throughout, positioned at eye-level or slightly below, emphasizing the man's presence within the expansive landscape and allowing the viewer to share in his peaceful observation. There is no verbal communication; his actions convey a sense of calm and contentment, inviting the viewer into this moment of quiet reflection amidst the beauty of nature.",
+  "avg_length": 255
+}

prediction_results/iter2000_sampled_85k/1601654-Split-000-Start-10-End-341+cropped_split_1.mp4.txt

@@ -0,0 +1,6 @@
+{
+  "video_path": "/data/hongwei/xzh_workspace/vdc_benchmark_construction/benchmark_videos/1601654-Split-000-Start-10-End-341+cropped_split_1.mp4",
+  "gt_caption": "The video depicts a scene of modern digital finance, focusing on an individual poised to initiate a transaction using their smartphone and credit card. The scene conveys themes of convenience and the increasing reliance on mobile technology in everyday commerce.\n\nA sleek, black smartphone, protected by a dark case with a subtly glinting camera array, is held firmly in the left hand. In the right hand, a standard-sized, dark-hued credit card is held, its details blurred but its form clearly identifiable as a payment instrument. The smooth glass of the phone contrasts with the rigid plastic of the card. The hands remain remarkably still, each maintaining a steady grip, conveying a sense of focused anticipation. There is no movement or speech; instead, the scene captures a moment of ready intent, suggesting the user is about to input card details or confirm a transaction on the phone. This subtle stillness implies concentration before the interaction begins.\n\nThe individual is dressed in a light-colored, possibly pale blue or white, collared shirt. The fabric appears soft and slightly wrinkled, suggesting comfort. The skin tone of the hands is natural and unadorned. The face is not visible, ensuring the focus remains on the hands and the objects.\n\nThe hands are positioned centrally, in a relaxed but purposeful manner. The left hand cradles the smartphone with fingers gently wrapped around its sides, while the right hand holds the credit card horizontally, ready for input, with the fingers slightly splayed for a stable grip. The overall posture, though only partially visible, suggests a seated individual, comfortably engaged and demonstrating a practiced ease with handling these tools for financial tasks.\n\nThe background is a soft, warm bokeh, creating a shallow depth of field. The blurred hues suggest an indoor setting, possibly a living room or a quiet home office, illuminated by diffused, natural light. The atmosphere is calm and undistracted, reinforcing the private and personal nature of the financial interaction.\n\nThe camera utilizes a static close-up shot, framing the character's hands and the objects from a slightly elevated, eye-level perspective. The focus is sharp on the credit card and smartphone, while the character's body and the background are intentionally blurred to isolate the key elements. This shallow depth of field guides the viewer's attention, creating an intimate view of the pending transaction.",
+  "prediction": "The video depicts a person engaged in an online transaction, likely a purchase or payment, using a smartphone and a credit card. The scene is tightly framed in a static medium close-up shot, focusing on the hands and upper torso of the individual. The background is intentionally blurred with a soft bokeh effect, suggesting an indoor setting like a home or office, and the lighting is soft and even, contributing to a calm and focused atmosphere.\n\nThe individual's hands are the primary focus. In their right hand, they hold a sleek, black smartphone encased in a matte protective cover, possibly silicone. The phone features a prominent camera lens array at the top left corner. In their left hand, they hold a standard-sized black credit card, its surface smooth and unadorned. The individual is wearing a light-colored, possibly beige or cream, collared shirt, suggesting casual yet neat attire.\n\nThe individual's posture appears relaxed but attentive, with both hands held steadily. Their fingers are gently curled around the devices, indicating familiarity and ease with handling them. The right hand holds the smartphone with a natural grip, while the left hand cradles the credit card with a similar, secure hold. The stillness of the hands suggests a moment of concentration as they process information from the phone screen, presumably inputting details from the credit card. There are no discernible gestures or fidgeting, reinforcing the impression of a deliberate and composed approach to the task. The overall behavior conveys a sense of quiet diligence and efficiency.",
+  "avg_length": 252
+}