This is the basic llava version of the original mol-instruct model, none tuned with only vision encoder of CLIP add on.
aea55e2 verified | { | |
| "auto_map": { | |
| "AutoProcessor": "processing_llava.LlavaProcessor" | |
| }, | |
| "image_processor_type": "CLIPImageProcessor", | |
| "processor_class": "LlavaProcessor", | |
| "patch_size": 14, | |
| "vision_feature_select_strategy": "default" | |
| } |