| task,model,n,acc,chance,kappa,acc spread,flip@0.25,flip slope,unstable,failure J @0.25,success J @0.25 |
| VSR,Qwen2.5-VL-3B,2195,0.76,0.5,0.521,0.023,0.109,0.039,0.195,0.631,0.866 |
| L1,Qwen2.5-VL-3B,821,0.405,0.25,0.202,0.012,0.194,0.045,0.223,0.811,0.729 |
| L2,Qwen2.5-VL-3B,764,0.374,0.25,0.164,0.018,0.211,0.072,0.228,0.816,0.702 |
| L3,Qwen2.5-VL-3B,836,0.255,0.25,-0.018,0.028,0.274,0.029,0.267,0.815,0.561 |
| VSR,Qwen2.5-VL-7B,2195,0.805,0.5,0.613,0.034,0.093,0.033,0.159,0.613,0.89 |
| L1,Qwen2.5-VL-7B,821,0.452,0.25,0.271,0.024,0.202,0.08,0.208,0.773,0.728 |
| L2,Qwen2.5-VL-7B,764,0.392,0.25,0.186,0.026,0.2,0.056,0.211,0.813,0.723 |
| L3,Qwen2.5-VL-7B,836,0.278,0.25,0.035,0.019,0.232,0.035,0.227,0.851,0.654 |
| VSR,Qwen3.5-4B,2195,0.813,0.5,0.622,0.017,0.076,0.031,0.126,0.662,0.911 |
| L1,Qwen3.5-4B,821,0.53,0.25,0.374,0.022,0.166,0.062,0.222,0.782,0.805 |
| L2,Qwen3.5-4B,764,0.386,0.25,0.191,0.012,0.187,0.065,0.174,0.826,0.739 |
| L3,Qwen3.5-4B,836,0.296,0.25,0.048,0.016,0.295,0.057,0.261,0.799,0.576 |
| VSR,Qwen3.5-4B-Base,2195,0.827,0.5,0.654,0.016,0.086,0.04,0.138,0.608,0.901 |
| L1,Qwen3.5-4B-Base,821,0.557,0.25,0.409,0.03,0.157,0.049,0.208,0.76,0.798 |
| L2,Qwen3.5-4B-Base,764,0.448,0.25,0.27,0.014,0.188,0.058,0.196,0.805,0.76 |
| L3,Qwen3.5-4B-Base,836,0.279,0.25,0.032,0.036,0.285,0.037,0.26,0.806,0.588 |
|
|
