| { | |
| "logit_scale_init_value": 4.60515022277832, | |
| "model_type": "clip", | |
| "source_checkpoint": "D:/sakura-project/model_hubs/raw/SARCLIP/ViT-L-14/vit_l_14_model.pth", | |
| "text_config": { | |
| "hidden_act": "quick_gelu", | |
| "hidden_size": 768, | |
| "intermediate_size": 3072, | |
| "layer_norm_eps": 1e-05, | |
| "max_position_embeddings": 77, | |
| "model_type": "clip_text_model", | |
| "num_attention_heads": 12, | |
| "num_hidden_layers": 12, | |
| "vocab_size": 49408 | |
| }, | |
| "vision_config": { | |
| "image_size": 224, | |
| "model_type": "clip_vision_model", | |
| "num_channels": 3, | |
| "patch_size": 14 | |
| } | |
| } |