List your company

Semantic Stress Evaluation on S3E

96.7Original PerformanceQwen3VL-2B
83.49286.92190.3593.779Jun 7, 2026

Evaluation results

MethodLinks
Qwen3VL-2BL=24, N=9,6612026.0696.786.183.80.632.541.92.12
Qwen3VL-4BL=24, N=9,2262026.0693.38580.16.5113.326.814.28
InternVL3-2BL=23, N=9,7852026.0692.790.384.90.324.133.812.01
Qwen3VL-8BL=32, N=9,6692026.0690.491.383.913.6226.5512.932.85
Gemma3-4BL=27, N=9,0682026.068493.478.70.030.960.935.45