Spoken Question Answering on HotpotQA v1.0 (test)
79.16Exact Match (EM)AEG (with LFE)
Evaluation Results
| Method | Links | |
|---|---|---|
| AEG (with LFE)Model=GPT-4o Audio2026.03 | 79.16 | |
| baselineModel=GPT-4o Audio2026.03 | 79.07 | |
| AEG (w/o LFE)Model=GPT-4o Audio2026.03 | 78.03 | |
| AEG (with LFE)Model=Qwen3-Omni-30B-A3B2026.03 | 76.95 | |
| AEG (w/o LFE)Model=Qwen3-Omni-30B-A3B2026.03 | 76.69 | |
| AEG (with LFE)Model=LongCat-Flash-Omni2026.03 | 76.09 | |
| AEG (w/o LFE)Model=LongCat-Flash-Omni2026.03 | 75.34 | |
| baselineModel=Qwen3-Omni-30B-A3B2026.03 | 75.02 | |
| baselineModel=LongCat-Flash-Omni2026.03 | 74.36 | |
| AEG (with LFE)Model=Qwen3-Omni Flash2026.03 | 74.17 | |
| baselineModel=Qwen3-Omni Flash2026.03 | 73.44 | |
| AEG (w/o LFE)Model=Qwen3-Omni Flash2026.03 | 73.02 |