| { |
| "model_name": "STAR_Qwen2.5-7B_VQGAN", |
| "model_type": "STARMultiModalityConfig", |
| "language_model": { |
| "model_name": "Qwen2.5-VL", |
| "model_path": "checkpoints/Qwen2.5-VL-7B-Instruct" |
| }, |
| "pixel_encoder": { |
| "model_name": "VQ_Model", |
| "model_path": "checkpoints/VQ-Model.pt", |
| "image_token_size": 65536, |
| "n_embed": 512, |
| "num_tokens": 576, |
| "num_heads": 8 |
| }, |
| "pixel_adapter": { |
| "model_name": "MLP_GELU", |
| "depth": 4, |
| "input_dim": 512, |
| "n_embed": 3584 |
| }, |
| "stacked_ar": { |
| "num_layers": 14 |
| }, |
| "pixel_output_head": { |
| "image_token_embed": 4096, |
| "image_token_size": 65536, |
| "n_embed": 3584 |
| }, |
| "pixel_decoder": { |
| "model_name": "LUMINA2", |
| "model_path": "checkpoints/lumina-image2" |
| }, |
| "torch_dtype": "bfloat16" |
| } |
|
|
