Spaces:
Runtime error
Runtime error
Model Human Metric Auto Metric Identify (Binary Accuracy) | |
Humans 95 92 | |
Ground-truth Caption β Llama-2-7b (Oracle) 71 | |
Ground-truth Caption β Llama-2-13b (Oracle) 70 | |
Ground-truth Caption β GPT4 (Oracle) 69 | |
Ground-truth Caption β GPT3 (Oracle) 68 62 74 | |
BLIP2 FlanT5-XXL (Fine-tuned) 27 24 73 | |
BLIP2 FlanT5-XL (Fine-tuned) 15 18 60 | |
Predicted Caption β Llama-2-7b 36 | |
Predicted Caption β Llama-2-13b 36 | |
Predicted Caption β GPT4 36 | |
Predicted Caption β GPT3 33 42 59 | |
InstructBLIP 31 | |
LLaVA 31 | |
mPLUG-Owl 24 | |
BLIP2 FlanT5-XXL (Zero-shot) 0 0 50 |