List your company

Multimodal Speech Understanding on MMSU Paralinguistic subset (test)

65.18AccuracyQwen2-Audio + PCLM
4.506420.258236.0151.7618May 26, 2026

Evaluation results

MethodLinks
Qwen2-Audio + PCLMPCLM=true2026.0565.18
Qwen2-Audio + PCLM + DPOPCLM=true, Training=DPO2026.0563.26
AF3 + PCLM + DPOPCLM=true, Training=DPO2026.0554.78
Step-Audio-R12026.0554.51
AF3 + PCLMPCLM=true2026.0554.06
Gemini 2.52026.0551.05
Qwen2-Audio + SFT w/o PCLMTraining=SFT, PCLM=false2026.0549.41
Qwen3-Omni2026.0549.13
Qwen2-Audio + SFT + DPO w/o PCLMTraining=SFT + DPO, PCLM=false2026.0547.86
AF3 + SFT + DPO w/o PCLMTraining=SFT + DPO, PCLM=false2026.0545.58
AF3 + SFT w/o PCLMTraining=SFT, PCLM=false2026.0544.76
Kimi-Audio2026.0541.48
AF32026.0537.74
GPT-4o Audio2026.0536.55
MiMo-Audio2026.0535.64
Qwen2-Audio2026.0534.37
Qwen2.5-Omni2026.0533.45
VITA-Audio2026.0529.54
AF22026.0527.44
SALMONN2026.056.84