Speech-to-Speech Question Answering on AlpacaEval
72.22AccuracyVocalNet
Evaluation results
| Method | Links | |
|---|---|---|
| VocalNetModel Parameters=8B2026.03 | 72.22 | |
| Qwen2.5-Omni-7BModel Parameters=7B2026.03 | 67.17 | |
| SPEECH-OMNI-LITEModel Parameters=8B2026.03 | 65.66 | |
| LLaMA-Omni2Model Parameters=7B2026.03 | 63.64 | |
| GLM-4-VoiceModel Parameters=9B2026.03 | 58.59 | |
| Freeze-OmniModel Parameters=7B2026.03 | 28.79 | |
| Baichuan-Omni-1.5Model Parameters=7B2026.03 | 13.92 |