| { | |
| "model_cfg": { | |
| "embed_dim": 1152, | |
| "init_logit_bias": -10, | |
| "custom_text": true, | |
| "vision_cfg": { | |
| "image_size": 256, | |
| "timm_model_name": "vit_so400m_patch16_siglip_256", | |
| "timm_model_pretrained": false, | |
| "timm_pool": "map", | |
| "timm_proj": "none" | |
| }, | |
| "text_cfg": { | |
| "context_length": 64, | |
| "vocab_size": 250000, | |
| "hf_tokenizer_name": "timm/ViT-SO400M-16-SigLIP-i18n-256", | |
| "tokenizer_kwargs": { | |
| "clean": "canonicalize" | |
| }, | |
| "width": 1152, | |
| "heads": 16, | |
| "layers": 27, | |
| "mlp_ratio": 3.7362, | |
| "no_causal_mask": true, | |
| "pool_type": "last", | |
| "proj_type": "none", | |
| "norm_kwargs": { | |
| "eps": 1e-06 | |
| } | |
| } | |
| }, | |
| "preprocess_cfg": { | |
| "mean": [ | |
| 0.5 | |
| ], | |
| "std": [ | |
| 0.5 | |
| ], | |
| "interpolation": "bicubic", | |
| "resize_mode": "squash" | |
| } | |
| } |