Question Answering on AmbigQA (One Intent Accuracy, Any Intent Accuracy)
69.8Accuracy (One Intent)BAG2+
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| BAG2+Backbone=Gemini-2.5-Flash, Generation Setting=Augmented Generation, Strategy=BAG2+2026.05 | 69.8 | — | |
| BAG1+Backbone=Gemini-2.5-Flash, Generation Setting=Augmented Generation, Strategy=BAG1+2026.05 | 68.8 | — | |
| Disambig OracleBackbone=Gemini-2.5-Flash, Generation Setting=Direct Generation, Strategy=Disambig2026.05 | 68.7 | 74.2 | |
| BAG3+Backbone=Gemini-2.5-Flash, Generation Setting=Augmented Generation, Strategy=BAG3+2026.05 | 68.4 | — | |
| SAG+Backbone=Gemini-2.5-Flash, Generation Setting=Augmented Generation, Strategy=SAG+2026.05 | 67.1 | — | |
| BAG1+Backbone=Qwen3-14B, Generation Setting=Augmented Generation, Strategy=BAG1+2026.05 | 63.8 | — | |
| BAG3+Backbone=Qwen3-14B, Generation Setting=Augmented Generation, Strategy=BAG3+2026.05 | 60.4 | — | |
| BAG1+Backbone=Qwen3-8B, Generation Setting=Augmented Generation, Strategy=BAG1+2026.05 | 60.2 | — | |
| Standard Direct GenerationBackbone=Gemini-2.5-Flash, Generation Setting=Direct Generation, Strategy=Standard2026.05 | 59.1 | 75.6 | |
| BAG2+Backbone=Qwen3-14B, Generation Setting=Augmented Generation, Strategy=BAG2+2026.05 | 58.8 | — | |
| BAG2+Backbone=OLMo2-13B, Generation Setting=Augmented Generation, Strategy=BAG2+2026.05 | 55.6 | — | |
| BAG3+Backbone=OLMo2-13B, Generation Setting=Augmented Generation, Strategy=BAG3+2026.05 | 55.1 | — | |
| BAG1+Backbone=OLMo2-13B, Generation Setting=Augmented Generation, Strategy=BAG1+2026.05 | 54.9 | — | |
| BAG3+Backbone=OLMo3-7B, Generation Setting=Augmented Generation, Strategy=BAG3+2026.05 | 52.9 | — | |
| Disambig OracleBackbone=OLMo2-13B, Generation Setting=Direct Generation, Strategy=Disambig2026.05 | 49.6 | 55.9 | |
| SAG+Backbone=OLMo2-13B, Generation Setting=Augmented Generation, Strategy=SAG+2026.05 | 46 | — | |
| Standard Direct GenerationBackbone=OLMo2-13B, Generation Setting=Direct Generation, Strategy=Standard2026.05 | 45.3 | 59.2 | |
| BAG2+Backbone=Qwen3-8B, Generation Setting=Augmented Generation, Strategy=BAG2+2026.05 | 45 | — | |
| BAG1+Backbone=OLMo2-7B, Generation Setting=Augmented Generation, Strategy=BAG1+2026.05 | 44.7 | — | |
| SAG+Backbone=Qwen3-14B, Generation Setting=Augmented Generation, Strategy=SAG+2026.05 | 43.6 | — | |
| BAG2+Backbone=OLMo2-7B, Generation Setting=Augmented Generation, Strategy=BAG2+2026.05 | 42.8 | — | |
| Disambig OracleBackbone=Qwen3-14B, Generation Setting=Direct Generation, Strategy=Disambig2026.05 | 42.8 | 49.3 | |
| SAG+Backbone=OLMo2-7B, Generation Setting=Augmented Generation, Strategy=SAG+2026.05 | 42.5 | — | |
| BAG2+Backbone=OLMo3-7B, Generation Setting=Augmented Generation, Strategy=BAG2+2026.05 | 41.8 | — | |
| BAG3+Backbone=OLMo2-7B, Generation Setting=Augmented Generation, Strategy=BAG3+2026.05 | 41.7 | — | |
| BAG3+Backbone=Qwen3-8B, Generation Setting=Augmented Generation, Strategy=BAG3+2026.05 | 40.8 | — | |
| Standard Direct GenerationBackbone=Qwen3-14B, Generation Setting=Direct Generation, Strategy=Standard2026.05 | 40.1 | 52.4 | |
| BAG1+Backbone=OLMo3-7B, Generation Setting=Augmented Generation, Strategy=BAG1+2026.05 | 39.8 | — | |
| Disambig OracleBackbone=OLMo2-7B, Generation Setting=Direct Generation, Strategy=Disambig2026.05 | 36.3 | 41.2 | |
| SAG+Backbone=Qwen3-8B, Generation Setting=Augmented Generation, Strategy=SAG+2026.05 | 34.4 | — | |
| Standard Direct GenerationBackbone=OLMo2-7B, Generation Setting=Direct Generation, Strategy=Standard2026.05 | 33.9 | 44.8 | |
| Disambig OracleBackbone=Qwen3-8B, Generation Setting=Direct Generation, Strategy=Disambig2026.05 | 33.7 | 40.3 | |
| Standard Direct GenerationBackbone=Qwen3-8B, Generation Setting=Direct Generation, Strategy=Standard2026.05 | 33.1 | 43.2 | |
| SAG+Backbone=OLMo3-7B, Generation Setting=Augmented Generation, Strategy=SAG+2026.05 | 26.4 | — | |
| Disambig OracleBackbone=OLMo3-7B, Generation Setting=Direct Generation, Strategy=Disambig2026.05 | 25.2 | 29.9 | |
| Standard Direct GenerationBackbone=OLMo3-7B, Generation Setting=Direct Generation, Strategy=Standard2026.05 | 24.7 | 32.4 |