| { | |
| "architectures": [ | |
| "Qwen3VLForConditionalGeneration" | |
| ], | |
| "model_type": "designer-thinking", | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "4.44.2", | |
| "thinking_tokens": 2000000, | |
| "_name_or_path": "zenlm/zen-designer-235b-a22b-thinking", | |
| "_base_model": "Qwen/Qwen3-VL-235B-A22B-Thinking", | |
| "vision_config": { | |
| "hidden_size": 2048, | |
| "image_size": 2048, | |
| "num_hidden_layers": 48, | |
| "patch_size": 14 | |
| }, | |
| "text_config": { | |
| "vocab_size": 151936, | |
| "hidden_size": 8192, | |
| "num_hidden_layers": 80, | |
| "num_attention_heads": 64, | |
| "num_key_value_heads": 8, | |
| "max_position_embeddings": 131072 | |
| }, | |
| "num_experts": 64, | |
| "num_experts_per_tok": 4, | |
| "expert_interval": 1, | |
| "_total_params": "235B", | |
| "_active_params": "22B" | |
| } |