Edited Visual Question Answering on MMEdit E-VQA
100ReliabilityFT
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| FTMLLM Backbone=BLIP22026.05 | 100 | 100 | 93.94 | 64.79 | |
| FTMLLM Backbone=MiniGPT-42026.05 | 100 | 100 | 97.5 | 40.85 | |
| WISEMLLM Backbone=MiniGPT-42026.05 | 100 | 98.86 | 100 | 56.48 | |
| ASAMWISEMLLM Backbone=MiniGPT-42026.05 | 100 | 99 | 100 | 56.75 | |
| IKEMLLM Backbone=MiniGPT-42026.05 | 99.95 | 99.9 | 50.02 | 3.31 | |
| IKEMLLM Backbone=BLIP22026.05 | 99.71 | 99.62 | 47.74 | 2.53 | |
| WISEMLLM Backbone=BLIP22026.05 | 99 | 84.8 | 100 | 8.99 | |
| ASAMWISEMLLM Backbone=BLIP22026.05 | 99 | 93.97 | 100 | 9.73 | |
| SERACMLLM Backbone=BLIP22026.05 | 97.6 | 97.3 | 100 | 3.21 | |
| ASAMMENDMLLM Backbone=MiniGPT-42026.05 | 96.6 | 95.8 | 99.62 | 93.6 | |
| MENDMLLM Backbone=MiniGPT-42026.05 | 96.13 | 95.41 | 99.61 | 93.92 | |
| ASAMMENDMLLM Backbone=BLIP22026.05 | 95.99 | 94.65 | 99.66 | 93.8 | |
| MENDMLLM Backbone=BLIP22026.05 | 95.22 | 94.17 | 99.53 | 92.35 | |
| UniKEMLLM Backbone=BLIP22026.05 | 94.32 | 87.18 | 95.98 | 93.15 | |
| SERACMLLM Backbone=MiniGPT-42026.05 | 91.7 | 98.6 | 99.99 | 3.72 | |
| UniKEMLLM Backbone=MiniGPT-42026.05 | 84.32 | 81.29 | 78.45 | 85.81 | |
| T-PatcherMLLM Backbone=BLIP22026.05 | 80.35 | 77.82 | 87.14 | 85.28 | |
| T-PatcherMLLM Backbone=MiniGPT-42026.05 | 70.56 | 68.79 | 64.45 | 81.77 | |
| Pre-editedMLLM Backbone=BLIP22026.05 | 25.85 | 26.37 | 99.38 | 92.83 | |
| Pre-editedMLLM Backbone=MiniGPT-42026.05 | 19.21 | 24.08 | 99.44 | 91.56 |