List your company

Mathematical Reasoning on Minerva Math (Pass@1)

38.2Pass@1SFT
8.76816.40924.0531.691Oct 9, 2025Nov 8, 2025Dec 9, 2025Jan 9, 2026Feb 8, 2026Mar 11, 2026Apr 11, 2026

Evaluation results

MethodLinks
SFTBase Model=Qwen2.5-Math-7B-Instruct, Evaluation Protocol=Zero-shot2025.1038.2
LightRBase Model=Qwen2.5-Math-7B-Instruct, Evaluation Protocol=Zero-shot2025.1034.2
BaselineBase Model=Qwen2.5-Math-7B-Instruct, Evaluation Protocol=Zero-shot2025.1033.8
FREIABackbone=Qwen2.5-Math-1.5B-Instruct2026.0432
LightRBase Model=Qwen2.5-Math-1.5B-Instruct, Evaluation Protocol=Zero-shot2025.1032
SFTBase Model=Qwen2.5-Math-1.5B-Instruct, Evaluation Protocol=Zero-shot2025.1031.6
GRPO (Supervised)Backbone=Qwen2.5-Math-1.5B-Instruct2026.0431.3
FREIABackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.0431.3
TTRLBackbone=Qwen2.5-Math-1.5B-Instruct2026.0430.9
TTRLBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.0430.9
GRPO (Supervised)Backbone=DeepSeek-R1-Distill-Qwen-1.5B2026.0430.5
EntropyBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.0429.8
EntropyBackbone=Qwen2.5-Math-1.5B-Instruct2026.0429.4
IntuitorBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.0429.4
BaselineBase Model=Qwen2.5-Math-1.5B-Instruct, Evaluation Protocol=Zero-shot2025.1029.4
BaseBackbone=Qwen2.5-Math-1.5B-Instruct2026.0428.7
IntuitorBackbone=Qwen2.5-Math-1.5B-Instruct2026.0428.7
BaseBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.0427.6
FREIABackbone=Qwen2.5-3B-Instruct2026.0426.1
TTRLBackbone=Qwen2.5-3B-Instruct2026.0425.8
GRPO (Supervised)Backbone=Qwen2.5-3B-Instruct2026.0425.4
EntropyBackbone=Qwen2.5-3B-Instruct2026.0425
BaseBackbone=Qwen2.5-3B-Instruct2026.0424.3
IntuitorBackbone=Qwen2.5-3B-Instruct2026.0424.3
SFTBase Model=DeepSeek-R1-Distill-Qwen-1.5B, Evaluation Protocol=Zero-shot2025.1018.4
LightRBase Model=DeepSeek-R1-Distill-Qwen-1.5B, Evaluation Protocol=Zero-shot2025.1018
BaselineBase Model=DeepSeek-R1-Distill-Qwen-1.5B, Evaluation Protocol=Zero-shot2025.1016.2
SFTBase Model=Qwen2.5-Math-1.5B, Evaluation Protocol=Zero-shot2025.1015.1
BaselineBase Model=Qwen2.5-Math-7B, Evaluation Protocol=Zero-shot2025.1014
SFTBase Model=Qwen2.5-Math-7B, Evaluation Protocol=Zero-shot2025.1012.1
LightRBase Model=Qwen2.5-Math-7B, Evaluation Protocol=Zero-shot2025.1012.1
LightRBase Model=Qwen2.5-Math-1.5B, Evaluation Protocol=Zero-shot2025.1011.4
BaselineBase Model=Qwen2.5-Math-1.5B, Evaluation Protocol=Zero-shot2025.109.9