model,task,baseline,n,accuracy,num_correct,unparsed_rate,scoring,chance,floor,headroom_captured Qwen_Qwen2.5-VL-3B-Instruct,vsr,0.00,2195,0.7713,1693,0.0,generate,0.5,0.5,0.5426 Qwen_Qwen2.5-VL-3B-Instruct,vsr,0.05,2195,0.7677,1685,0.0,generate,0.5,0.5,0.5353 Qwen_Qwen2.5-VL-3B-Instruct,vsr,0.10,2195,0.7595,1667,0.0,generate,0.5,0.5,0.5189 Qwen_Qwen2.5-VL-3B-Instruct,vsr,0.15,2195,0.7581,1664,0.0,generate,0.5,0.5,0.5162 Qwen_Qwen2.5-VL-3B-Instruct,vsr,0.20,2195,0.7526,1652,0.0,generate,0.5,0.5,0.5052 Qwen_Qwen2.5-VL-3B-Instruct,vsr,0.25,2195,0.7481,1642,0.0,generate,0.5,0.5,0.4961 Qwen_Qwen2.5-VL-3B-Instruct,youtube_level1,0.00,821,0.3995,328,0.0,generate,0.25,0.25,0.1994 Qwen_Qwen2.5-VL-3B-Instruct,youtube_level1,0.05,821,0.4056,333,0.0,generate,0.25,0.25,0.2075 Qwen_Qwen2.5-VL-3B-Instruct,youtube_level1,0.10,821,0.4117,338,0.0,generate,0.25,0.25,0.2156 Qwen_Qwen2.5-VL-3B-Instruct,youtube_level1,0.15,821,0.4105,337,0.0,generate,0.25,0.25,0.214 Qwen_Qwen2.5-VL-3B-Instruct,youtube_level1,0.20,821,0.3995,328,0.0,generate,0.25,0.25,0.1994 Qwen_Qwen2.5-VL-3B-Instruct,youtube_level1,0.25,821,0.4007,329,0.0,generate,0.25,0.25,0.201 Qwen_Qwen2.5-VL-3B-Instruct,youtube_level2,0.00,764,0.3626,277,0.0,generate,0.25,0.25,0.1501 Qwen_Qwen2.5-VL-3B-Instruct,youtube_level2,0.05,764,0.3743,286,0.0,generate,0.25,0.25,0.1658 Qwen_Qwen2.5-VL-3B-Instruct,youtube_level2,0.10,764,0.3783,289,0.0,generate,0.25,0.25,0.171 Qwen_Qwen2.5-VL-3B-Instruct,youtube_level2,0.15,764,0.3796,290,0.0,generate,0.25,0.25,0.1728 Qwen_Qwen2.5-VL-3B-Instruct,youtube_level2,0.20,764,0.3809,291,0.0,generate,0.25,0.25,0.1745 Qwen_Qwen2.5-VL-3B-Instruct,youtube_level2,0.25,764,0.3704,283,0.0,generate,0.25,0.25,0.1606 Qwen_Qwen2.5-VL-3B-Instruct,youtube_level3,0.00,836,0.2667,223,0.0,generate,0.25,0.25,0.0223 Qwen_Qwen2.5-VL-3B-Instruct,youtube_level3,0.05,836,0.2452,205,0.0,generate,0.25,0.25,-0.0064 Qwen_Qwen2.5-VL-3B-Instruct,youtube_level3,0.10,836,0.2392,200,0.0,generate,0.25,0.25,-0.0144 Qwen_Qwen2.5-VL-3B-Instruct,youtube_level3,0.15,836,0.2524,211,0.0,generate,0.25,0.25,0.0032 Qwen_Qwen2.5-VL-3B-Instruct,youtube_level3,0.20,836,0.2596,217,0.0,generate,0.25,0.25,0.0128 Qwen_Qwen2.5-VL-3B-Instruct,youtube_level3,0.25,836,0.2656,222,0.0,generate,0.25,0.25,0.0207 Qwen_Qwen2.5-VL-7B-Instruct,vsr,0.00,2195,0.8223,1805,0.0,generate,0.5,0.5,0.6446 Qwen_Qwen2.5-VL-7B-Instruct,vsr,0.05,2195,0.8077,1773,0.0,generate,0.5,0.5,0.6155 Qwen_Qwen2.5-VL-7B-Instruct,vsr,0.10,2195,0.8027,1762,0.0,generate,0.5,0.5,0.6055 Qwen_Qwen2.5-VL-7B-Instruct,vsr,0.15,2195,0.8068,1771,0.0,generate,0.5,0.5,0.6137 Qwen_Qwen2.5-VL-7B-Instruct,vsr,0.20,2195,0.8023,1761,0.0,generate,0.5,0.5,0.6046 Qwen_Qwen2.5-VL-7B-Instruct,vsr,0.25,2195,0.7882,1730,0.0,generate,0.5,0.5,0.5763 Qwen_Qwen2.5-VL-7B-Instruct,youtube_level1,0.00,821,0.447,367,0.0,generate,0.25,0.25,0.2627 Qwen_Qwen2.5-VL-7B-Instruct,youtube_level1,0.05,821,0.4482,368,0.0,generate,0.25,0.25,0.2643 Qwen_Qwen2.5-VL-7B-Instruct,youtube_level1,0.10,821,0.4555,374,0.0,generate,0.25,0.25,0.2741 Qwen_Qwen2.5-VL-7B-Instruct,youtube_level1,0.15,821,0.4665,383,0.0,generate,0.25,0.25,0.2887 Qwen_Qwen2.5-VL-7B-Instruct,youtube_level1,0.20,821,0.4421,363,0.0,generate,0.25,0.25,0.2562 Qwen_Qwen2.5-VL-7B-Instruct,youtube_level1,0.25,821,0.4519,371,0.0,generate,0.25,0.25,0.2692 Qwen_Qwen2.5-VL-7B-Instruct,youtube_level2,0.00,764,0.4031,308,0.0,generate,0.25,0.25,0.2042 Qwen_Qwen2.5-VL-7B-Instruct,youtube_level2,0.05,764,0.394,301,0.0,generate,0.25,0.25,0.192 Qwen_Qwen2.5-VL-7B-Instruct,youtube_level2,0.10,764,0.3979,304,0.0,generate,0.25,0.25,0.1972 Qwen_Qwen2.5-VL-7B-Instruct,youtube_level2,0.15,764,0.4018,307,0.0,generate,0.25,0.25,0.2024 Qwen_Qwen2.5-VL-7B-Instruct,youtube_level2,0.20,764,0.377,288,0.0,generate,0.25,0.25,0.1693 Qwen_Qwen2.5-VL-7B-Instruct,youtube_level2,0.25,764,0.377,288,0.0,generate,0.25,0.25,0.1693 Qwen_Qwen2.5-VL-7B-Instruct,youtube_level3,0.00,836,0.2799,234,0.0,generate,0.25,0.25,0.0399 Qwen_Qwen2.5-VL-7B-Instruct,youtube_level3,0.05,836,0.2871,240,0.0,generate,0.25,0.25,0.0494 Qwen_Qwen2.5-VL-7B-Instruct,youtube_level3,0.10,836,0.2679,224,0.0,generate,0.25,0.25,0.0239 Qwen_Qwen2.5-VL-7B-Instruct,youtube_level3,0.15,836,0.2823,236,0.0,generate,0.25,0.25,0.0431 Qwen_Qwen2.5-VL-7B-Instruct,youtube_level3,0.20,836,0.2763,231,0.0,generate,0.25,0.25,0.0351 Qwen_Qwen2.5-VL-7B-Instruct,youtube_level3,0.25,836,0.2739,229,0.0,generate,0.25,0.25,0.0319 Qwen_Qwen3.5-4B,vsr,0.00,2195,0.8214,1803,0.0,generate,0.5,0.5,0.6428 Qwen_Qwen3.5-4B,vsr,0.05,2195,0.8173,1794,0.0,generate,0.5,0.5,0.6346 Qwen_Qwen3.5-4B,vsr,0.10,2195,0.8187,1797,0.0,generate,0.5,0.5,0.6374 Qwen_Qwen3.5-4B,vsr,0.15,2195,0.8114,1781,0.0,generate,0.5,0.5,0.6228 Qwen_Qwen3.5-4B,vsr,0.20,2195,0.8073,1772,0.0,generate,0.5,0.5,0.6146 Qwen_Qwen3.5-4B,vsr,0.25,2195,0.8046,1766,0.0,generate,0.5,0.5,0.6091 Qwen_Qwen3.5-4B,youtube_level1,0.00,821,0.5274,433,0.0037,generate,0.25,0.25,0.3699 Qwen_Qwen3.5-4B,youtube_level1,0.05,821,0.5213,428,0.0073,generate,0.25,0.25,0.3618 Qwen_Qwen3.5-4B,youtube_level1,0.10,821,0.5189,426,0.0012,generate,0.25,0.25,0.3585 Qwen_Qwen3.5-4B,youtube_level1,0.15,821,0.5408,444,0.0049,generate,0.25,0.25,0.3877 Qwen_Qwen3.5-4B,youtube_level1,0.20,821,0.5347,439,0.0012,generate,0.25,0.25,0.3796 Qwen_Qwen3.5-4B,youtube_level1,0.25,821,0.5347,439,0.0,generate,0.25,0.25,0.3796 Qwen_Qwen3.5-4B,youtube_level2,0.00,764,0.3914,299,0.0,generate,0.25,0.25,0.1885 Qwen_Qwen3.5-4B,youtube_level2,0.05,764,0.3901,298,0.0,generate,0.25,0.25,0.1867 Qwen_Qwen3.5-4B,youtube_level2,0.10,764,0.3822,292,0.0,generate,0.25,0.25,0.1763 Qwen_Qwen3.5-4B,youtube_level2,0.15,764,0.3874,296,0.0,generate,0.25,0.25,0.1832 Qwen_Qwen3.5-4B,youtube_level2,0.20,764,0.3796,290,0.0,generate,0.25,0.25,0.1728 Qwen_Qwen3.5-4B,youtube_level2,0.25,764,0.3848,294,0.0,generate,0.25,0.25,0.1798 Qwen_Qwen3.5-4B,youtube_level3,0.00,836,0.2931,245,0.0,generate,0.25,0.25,0.0574 Qwen_Qwen3.5-4B,youtube_level3,0.05,836,0.299,250,0.0,generate,0.25,0.25,0.0654 Qwen_Qwen3.5-4B,youtube_level3,0.10,836,0.2967,248,0.0,generate,0.25,0.25,0.0622 Qwen_Qwen3.5-4B,youtube_level3,0.15,836,0.3038,254,0.0,generate,0.25,0.25,0.0718 Qwen_Qwen3.5-4B,youtube_level3,0.20,836,0.2883,241,0.0,generate,0.25,0.25,0.051 Qwen_Qwen3.5-4B,youtube_level3,0.25,836,0.2931,245,0.0,generate,0.25,0.25,0.0574 Qwen_Qwen3.5-4B-Base,vsr,0.00,2195,0.8319,1826,0.0,likelihood,0.5,0.5,0.6638 Qwen_Qwen3.5-4B-Base,vsr,0.05,2195,0.8314,1825,0.0,likelihood,0.5,0.5,0.6629 Qwen_Qwen3.5-4B-Base,vsr,0.10,2195,0.8323,1827,0.0,likelihood,0.5,0.5,0.6647 Qwen_Qwen3.5-4B-Base,vsr,0.15,2195,0.8278,1817,0.0,likelihood,0.5,0.5,0.6556 Qwen_Qwen3.5-4B-Base,vsr,0.20,2195,0.8241,1809,0.0,likelihood,0.5,0.5,0.6483 Qwen_Qwen3.5-4B-Base,vsr,0.25,2195,0.8164,1792,0.0,likelihood,0.5,0.5,0.6328 Qwen_Qwen3.5-4B-Base,youtube_level1,0.00,821,0.542,445,0.0,likelihood,0.25,0.25,0.3894 Qwen_Qwen3.5-4B-Base,youtube_level1,0.05,821,0.5457,448,0.0,likelihood,0.25,0.25,0.3942 Qwen_Qwen3.5-4B-Base,youtube_level1,0.10,821,0.5603,460,0.0,likelihood,0.25,0.25,0.4137 Qwen_Qwen3.5-4B-Base,youtube_level1,0.15,821,0.5725,470,0.0,likelihood,0.25,0.25,0.43 Qwen_Qwen3.5-4B-Base,youtube_level1,0.20,821,0.5664,465,0.0,likelihood,0.25,0.25,0.4218 Qwen_Qwen3.5-4B-Base,youtube_level1,0.25,821,0.5554,456,0.0,likelihood,0.25,0.25,0.4072 Qwen_Qwen3.5-4B-Base,youtube_level2,0.00,764,0.4424,338,0.0,likelihood,0.25,0.25,0.2565 Qwen_Qwen3.5-4B-Base,youtube_level2,0.05,764,0.4529,346,0.0,likelihood,0.25,0.25,0.2705 Qwen_Qwen3.5-4B-Base,youtube_level2,0.10,764,0.4503,344,0.0,likelihood,0.25,0.25,0.267 Qwen_Qwen3.5-4B-Base,youtube_level2,0.15,764,0.4542,347,0.0,likelihood,0.25,0.25,0.2723 Qwen_Qwen3.5-4B-Base,youtube_level2,0.20,764,0.4503,344,0.0,likelihood,0.25,0.25,0.267 Qwen_Qwen3.5-4B-Base,youtube_level2,0.25,764,0.4398,336,0.0,likelihood,0.25,0.25,0.2531 Qwen_Qwen3.5-4B-Base,youtube_level3,0.00,836,0.3026,253,0.0,likelihood,0.25,0.25,0.0702 Qwen_Qwen3.5-4B-Base,youtube_level3,0.05,836,0.2715,227,0.0,likelihood,0.25,0.25,0.0287 Qwen_Qwen3.5-4B-Base,youtube_level3,0.10,836,0.2775,232,0.0,likelihood,0.25,0.25,0.0367 Qwen_Qwen3.5-4B-Base,youtube_level3,0.15,836,0.2667,223,0.0,likelihood,0.25,0.25,0.0223 Qwen_Qwen3.5-4B-Base,youtube_level3,0.20,836,0.2727,228,0.0,likelihood,0.25,0.25,0.0303 Qwen_Qwen3.5-4B-Base,youtube_level3,0.25,836,0.2823,236,0.0,likelihood,0.25,0.25,0.0431