Hallucination Evaluation on CHAIR MSCOCO
59.4CHAIR_SVCD
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| VCDBase Model=LLaVA-1.5, Max New Tokens=2562026.03 | 59.4 | 16 | |
| DoLABase Model=Shikra, Max New Tokens=2562026.03 | 55.7 | 14.8 | |
| GreedyBase Model=Shikra, Max New Tokens=2562026.03 | 55.4 | 15.2 | |
| VCDBase Model=Shikra, Max New Tokens=2562026.03 | 54.9 | 15.1 | |
| GreedyBase Model=InstructBLIP, Max New Tokens=2562026.03 | 54.7 | 20.2 | |
| DoLABase Model=LLaVA-1.5, Max New Tokens=2562026.03 | 54 | 14.2 | |
| DoLABase Model=InstructBLIP, Max New Tokens=2562026.03 | 52.8 | 19.1 | |
| GreedyBase Model=LLaVA-1.5, Max New Tokens=2562026.03 | 51.8 | 13.7 | |
| FarSightBase Model=Shikra, Max New Tokens=2562026.03 | 51.6 | 12.2 | |
| HALCBase Model=LLaVA-1.5, Max New Tokens=2562026.03 | 51.4 | 13.1 | |
| VCDBase Model=InstructBLIP, Max New Tokens=2562026.03 | 50.1 | 19.4 | |
| FarSightBase Model=InstructBLIP, Max New Tokens=2562026.03 | 48.3 | 18.7 | |
| OPERABase Model=InstructBLIP, Max New Tokens=2562026.03 | 48.2 | 14.3 | |
| HALCBase Model=InstructBLIP, Max New Tokens=2562026.03 | 48 | 15.3 | |
| HALCBase Model=Shikra, Max New Tokens=2562026.03 | 47.2 | 14.5 | |
| OPERABase Model=LLaVA-1.5, Max New Tokens=2562026.03 | 44.1 | 12.8 | |
| FarSightBase Model=LLaVA-1.5, Max New Tokens=2562026.03 | 43.9 | 12.5 | |
| GreedyBase Model=MiniGPT4, Max New Tokens=2562026.03 | 43 | 13.4 | |
| DoLABase Model=MiniGPT4, Max New Tokens=2562026.03 | 42 | 13.5 | |
| VCDBase Model=MiniGPT4, Max New Tokens=2562026.03 | 40 | 14.6 | |
| FarSightBase Model=MiniGPT4, Max New Tokens=2562026.03 | 39.2 | 12.4 | |
| OPERABase Model=Shikra, Max New Tokens=2562026.03 | 38.2 | 14.2 | |
| HALCBase Model=MiniGPT4, Max New Tokens=2562026.03 | 37.4 | 12.1 | |
| OPERABase Model=MiniGPT4, Max New Tokens=2562026.03 | 37.3 | 13.5 | |
| AD-HHBase Model=Shikra, Max New Tokens=2562026.03 | 36.9 | 13.7 | |
| AD-HHBase Model=InstructBLIP, Max New Tokens=2562026.03 | 36 | 10.3 | |
| AD-HHBase Model=LLaVA-1.5, Max New Tokens=2562026.03 | 35.2 | 8.8 | |
| VCDBase Model=InstructBLIP, Max New Tokens=642026.03 | 34.1 | 15.8 | |
| AD-HHBase Model=MiniGPT4, Max New Tokens=2562026.03 | 32.8 | 11.5 | |
| AIRBase Model=Shikra, Max New Tokens=2562026.03 | 30.3 | 9.3 | |
| AIRBase Model=InstructBLIP, Max New Tokens=2562026.03 | 30.1 | 8.6 | |
| GreedyBase Model=InstructBLIP, Max New Tokens=642026.03 | 29.8 | 14.2 | |
| GreedyBase Model=MiniGPT4, Max New Tokens=642026.03 | 28.8 | 11.4 | |
| AIRBase Model=LLaVA-1.5, Max New Tokens=2562026.03 | 28.8 | 8.6 | |
| DoLABase Model=MiniGPT4, Max New Tokens=642026.03 | 28.6 | 11.5 | |
| DoLABase Model=InstructBLIP, Max New Tokens=642026.03 | 28.1 | 14.2 | |
| VCDBase Model=MiniGPT4, Max New Tokens=642026.03 | 28 | 11.3 | |
| OPERABase Model=MiniGPT4, Max New Tokens=642026.03 | 26.6 | 10.4 | |
| HALCBase Model=MiniGPT4, Max New Tokens=642026.03 | 24.2 | 9.7 | |
| HALCBase Model=InstructBLIP, Max New Tokens=642026.03 | 24.2 | 14.1 | |
| VCDBase Model=Shikra, Max New Tokens=642026.03 | 23.5 | 8.3 | |
| VCDBase Model=LLaVA-1.5, Max New Tokens=642026.03 | 22.8 | 7.1 | |
| FarSightBase Model=MiniGPT4, Max New Tokens=642026.03 | 22.5 | 8.9 | |
| GreedyBase Model=Shikra, Max New Tokens=642026.03 | 22.3 | 8.2 | |
| DoLABase Model=LLaVA-1.5, Max New Tokens=642026.03 | 22.2 | 6.7 | |
| AD-HHBase Model=MiniGPT4, Max New Tokens=642026.03 | 22 | 8.5 | |
| AIRBase Model=MiniGPT4, Max New Tokens=2562026.03 | 21.3 | 8.9 | |
| HALCBase Model=LLaVA-1.5, Max New Tokens=642026.03 | 21.2 | 6.6 | |
| GreedyBase Model=LLaVA-1.5, Max New Tokens=642026.03 | 20.8 | 6.2 | |
| FarSightBase Model=InstructBLIP, Max New Tokens=642026.03 | 20.2 | 9.6 | |
| DoLABase Model=Shikra, Max New Tokens=642026.03 | 20.2 | 9.6 | |
| AD-HHBase Model=InstructBLIP, Max New Tokens=642026.03 | 19.6 | 8.5 | |
| FarSightBase Model=LLaVA-1.5, Max New Tokens=642026.03 | 18.6 | 5.9 | |
| OPERABase Model=InstructBLIP, Max New Tokens=642026.03 | 18.4 | 8.7 | |
| OPERABase Model=LLaVA-1.5, Max New Tokens=642026.03 | 17.8 | 6.5 | |
| AIRBase Model=MiniGPT4, Max New Tokens=642026.03 | 17.2 | 7.5 | |
| AIRBase Model=InstructBLIP, Max New Tokens=642026.03 | 16.3 | 8.1 | |
| AD-HHBase Model=LLaVA-1.5, Max New Tokens=642026.03 | 15.6 | 5.7 | |
| FarSightBase Model=Shikra, Max New Tokens=642026.03 | 15.4 | 7.6 | |
| AIRBase Model=LLaVA-1.5, Max New Tokens=642026.03 | 14.8 | 5.1 | |
| OPERABase Model=Shikra, Max New Tokens=642026.03 | 14.6 | 7.4 | |
| HALCBase Model=Shikra, Max New Tokens=642026.03 | 14.4 | 7.8 | |
| AD-HHBase Model=Shikra, Max New Tokens=642026.03 | 13.8 | 7 | |
| AIRBase Model=Shikra, Max New Tokens=642026.03 | 12.9 | 7 |