| { |
| "_name_or_path": "/mnt/cschlarmann37/project_bimodal-robust-clip/checkpoints/hf-hub_laion_CLIP-ViT-g-14-laion2B-s12B-b42K__imagenet_l2_FARE2_6I8fe/checkpoints/hf_model_combined", |
| "architectures": [ |
| "CLIPModel" |
| ], |
| "initializer_factor": 1.0, |
| "logit_scale_init_value": 2.6592, |
| "model_type": "clip", |
| "projection_dim": 1024, |
| "text_config": { |
| "hidden_act": "gelu", |
| "hidden_size": 1024, |
| "intermediate_size": 4096, |
| "model_type": "clip_text_model", |
| "num_attention_heads": 16, |
| "num_hidden_layers": 24, |
| "torch_dtype": "float32" |
| }, |
| "torch_dtype": "float32", |
| "transformers_version": "4.49.0", |
| "vision_config": { |
| "hidden_act": "gelu", |
| "hidden_size": 1408, |
| "intermediate_size": 6144, |
| "model_type": "clip_vision_model", |
| "num_attention_heads": 16, |
| "num_hidden_layers": 40, |
| "patch_size": 14, |
| "torch_dtype": "float32" |
| } |
| } |
|
|
