Question Answering on WikiQA (Accuracy/Normalized Accuracy)
96AccuracyFine-tuned
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Fine-tunedBackbone=T5-large2026.06 | 96 | — | |
| ReTeXBackbone=T5-large, Merging Strategy=Dynamic model merging2026.06 | 95.7 | 99.7 | |
| WEMoEBackbone=T5-large, Merging Strategy=Dynamic model merging2026.06 | 94.1 | 98 | |
| AdaMergingBackbone=T5-large, Merging Strategy=Static model merging2026.06 | 93.5 | 97.4 | |
| Consensus TABackbone=T5-large, Merging Strategy=Static model merging2026.06 | 90.3 | 94.1 | |
| Twin-MergingBackbone=T5-large, Merging Strategy=Dynamic model merging2026.06 | 77.1 | 80.3 | |
| DaWinBackbone=T5-large, Merging Strategy=Dynamic model merging2026.06 | 70.2 | 73.1 | |
| TIES-MergingBackbone=T5-large, Merging Strategy=Static model merging2026.06 | 64.9 | 67.6 | |
| Weight AveragingBackbone=T5-large, Merging Strategy=Static model merging2026.06 | 63.2 | 65.8 | |
| Task ArithmeticBackbone=T5-large, Merging Strategy=Static model merging2026.06 | 59.1 | 61.6 | |
| MoW-MergingBackbone=T5-large, Merging Strategy=Dynamic model merging2026.06 | 56.6 | 59 |