Difference between revisions of "2025:Music Reasoning QA Results"

From MIREX Wiki
(MMAR Results)
(OMniBench Results)
Line 46: Line 46:
 
|-
 
|-
 
| Baseline 1
 
| Baseline 1
| SAR-LM
+
| SAR-LM (w/ Qwen3)
 
| style="text-align:right;" | 31.26%
 
| style="text-align:right;" | 31.26%
 
| style="text-align:right;" | 41.50%
 
| style="text-align:right;" | 41.50%

Revision as of 16:23, 12 September 2025

MMAR Results

System Methods Used ACC music ACC mix-sound-music mix-music-speech mix-sound-music-speech
Baseline 1 SAR-LM (w/ Qwen3) 40.00% 33.98% 27.27% 48.78% 37.50%
Baseline 2 Qwen2.5-Omni 56.70% 40.78% 54.55% 67.07% 58.33%
Baseline 3 SAR-LM (w/ Gemini) TBA TBA TBA TBA TBA

OMniBench Results

System Methods Used ACC music ACC
Baseline 1 SAR-LM (w/ Qwen3) 31.26% 41.50%
Baseline 2 Qwen2-Audio-7B-Instruct 40.72% 38.68%