| { |
| "architectures": [ |
| "Qwen3VLForConditionalGeneration" |
| ], |
| "dtype": "float32", |
| "image_token_id": 151655, |
| "model_type": "qwen3_vl", |
| "text_config": { |
| "attention_bias": false, |
| "attention_dropout": 0.0, |
| "head_dim": 8, |
| "hidden_act": "silu", |
| "hidden_size": 16, |
| "initializer_range": 0.02, |
| "intermediate_size": 16, |
| "max_position_embeddings": 128000, |
| "model_type": "qwen3_vl_text", |
| "num_attention_heads": 2, |
| "num_hidden_layers": 1, |
| "num_key_value_heads": 2, |
| "pad_token_id": null, |
| "rms_norm_eps": 1e-06, |
| "rope_parameters": { |
| "mrope_section": [ |
| 1, |
| 1, |
| 2 |
| ], |
| "rope_theta": 1000000.0, |
| "rope_type": "default", |
| "type": "default" |
| }, |
| "use_cache": true, |
| "vocab_size": 1024, |
| "rope_scaling": { |
| "mrope_section": [ |
| 1, |
| 1, |
| 2 |
| ], |
| "rope_type": "default", |
| "type": "default" |
| } |
| }, |
| "tie_word_embeddings": false, |
| "transformers_version": "5.6.2", |
| "video_token_id": 151656, |
| "vision_config": { |
| "deepstack_visual_indexes": [ |
| 0 |
| ], |
| "depth": 1, |
| "hidden_act": "gelu_pytorch_tanh", |
| "hidden_size": 16, |
| "in_channels": 3, |
| "initializer_range": 0.02, |
| "intermediate_size": 16, |
| "model_type": "qwen3_vl_vision", |
| "num_heads": 2, |
| "num_position_embeddings": 2304, |
| "out_hidden_size": 16, |
| "patch_size": 14, |
| "spatial_merge_size": 2, |
| "temporal_patch_size": 2 |
| }, |
| "vision_end_token_id": 151653, |
| "vision_start_token_id": 151652 |
| } |
