Share your thoughts, 1 month free Claude Pro on usSee more

Mathematical Reasoning on Math tasks (Accuracy Breakdown)

81.6Easy Accuracy

SFT-80K (no RL)

Updated 1d ago

Evaluation Results

Method	Links
SFT-80K (no RL) 2026.05		81.6	43.3	70.7
DPO (no RL) 2026.05		80.1	45.8	70.3
CPT + RL 2026.05		79.8	43.3	69.4
CPT 2026.05		79.1	44.2	69.1
Abs-RL 2026.05		78.7	39.4	67.5
SFT (no RL) 2026.05		78.4	45.2	68.9
SFT+RL 2026.05		78.3	39.6	67.2
DPO+RL 2026.05		77.2	42.7	67.3
SFT-80K+RL 2026.05		76.5	39.4	65.9
CPT + RL 2026.05		72.9	29.2	60.6
Abs-RL 2026.05		72.7	27.9	60
SFT-80K+RL 2026.05		72.4	28.9	60
SFT+RL 2026.05		69.8	29.8	58.4
DPO+RL 2026.05		69.7	30.2	58.5
CPT + RL 2026.05		67.5	26.5	55.8
Abs-RL 2026.05		67.3	23.1	54.7
SFT-80K+RL 2026.05		66.7	23	54.2
SFT+RL 2026.05		66	24.8	54.2
DPO+RL 2026.05		63.4	18.3	50.5
Base 2026.05		60.2	12.9	46.6
Base 2026.05		54	13.6	42.4
Base 2026.05		43.9	11.7	41
SFT-80K+RL 2026.05		39.1	7	29.9
CPT + RL 2026.05		29.2	3.3	21.8
SFT+RL 2026.05		28.4	4.6	21.6
Abs-RL 2026.05		28.1	1	20.4
DPO+RL 2026.05		24.5	2.7	18.3
Base 2026.05		22.8	2.5	17