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

Mathematical Reasoning on AIME 24 (accuracy, delta)

14Accuracy

Berr. Latent

Updated 1mo ago

Evaluation Results

Method	Links
Berr. Latent 2026.02		14	-
Qwen2.5-Math-7B w/ CADFT 2026.04		10.41	-
Qwen2.5-Math-1.5B w/ CADFT 2026.04		8.76	-
Qwen2.5-Math-7B w/ DFT 2026.04		8.56	-
Qwen2.5-Math-1.5B w/ DFT 2026.04		6.87	-
Qwen2.5-Math-7B 2026.04		6.68	-
Qwen2.5-Math-1.5B 2026.04		4.16	-
LLaDA + Sonnet Plan 2026.02		3.3	3.3
Qwen2.5-Math-7B w/ SFT 2026.04		2.48	-
Qwen2.5-Math-1.5B w/ SFT 2026.04		1.87	-
DeepSeekMath-7B w/ CADFT 2026.04		1.65	-
Berr. BL 2026.02		1.5	-
Berr. Text 2026.02		1.5	-
DeepSeekMath-7B w/ DFT 2026.04		1.24	-
LLaMA-3.2-3B w/ CADFT 2026.04		1.05	-
LLaMA-3.2-3B w/ DFT 2026.04		0.83	-
LLaMA-3.1-8B w/ CADFT 2026.04		0.65	-
LLaMA-3.2-3B 2026.04		0.41	-
LLaMA-3.1-8B w/ DFT 2026.04		0.41	-
DeepSeekMath-7B w/ SFT 2026.04		0.41	-
LLaMA-3.1-8B 2026.04		0.21	-
DeepSeekMath-7B 2026.04		0.21	-
LLaDA 2026.02		0	-
LLaMA-3.2-3B w/ SFT 2026.04		0	-
LLaMA-3.1-8B w/ SFT 2026.04		0	-