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

Mathematical Reasoning on MATH500 (pass@k metrics)

91.3Pass@1

PieceHint-Nemotron-1.5B

Updated 1mo ago

Evaluation Results

Method	Links
PieceHint-Nemotron-1.5B 2026.04		91.3	-
PieceHint-Qwen3-1.7B 2026.04		89.8	-
Nemotron-1.5B 2026.04		88.5	-
PieceHint-DeepSeek-1.5B 2026.04		87.7	-
DeepSeek-R1-Distill-32B 2026.04		86	-
Qwen3-4B 2026.04		85.1	-
Qwen3-1.7B 2026.04		84.1	-
OPO 2025.10		82.2	-
STEER 2025.10		82.2	-
DeepSeek-R1-Distill-7B 2026.04		80.3	-
GRPO w/ Entro. Loss 2025.10		80	-
Eurus-PRIME 2025.10		79.8	-
Entro. Adv. 2025.10		79.6	-
SimpleRL-Zoo 2025.10		79.2	-
GRPO w/ Fork Tokens 2025.10		79.2	-
W-REINFORCE 2025.10		79.2	-
F-DAPO 2026.02		79.1	96.6
F-CISPO 2026.02		79	97.8
GRPO w/ clip-high 2025.10		79	-
KL–Cov 2025.10		78.8	-
CISPO 2026.02		78.7	97
DAPO 2026.02		78.6	95.2
GRPO 2025.10		78.6	-
Clip–Cov 2025.10		78	-
CISPO 2026.02		77.3	95.5
F-DAPO 2026.02		76.8	97
F-CISPO 2026.02		76.8	96.5
DAPO 2026.02		76.5	96.2
F-GRPO 2026.02		76.2	95.1
GRPO 2026.02		75.8	92.8
GRPO 2026.02		75.4	95.6
F-GRPO 2026.02		74.7	96.1
DeepSeek-R1-Distill-1.5B 2026.04		72.1	-
F-DAPO 2026.02		58.6	92
CISPO 2026.02		56.9	89.1
F-CISPO 2026.02		56.5	91
DAPO 2026.02		55.9	83.8
GRPO 2026.02		55	90.6
F-GRPO 2026.02		54.1	92.9
Qwen2.5-Math-7B 2025.10		39.6	-