Mathematical Reasoning on Aggregate 6 Benchmarks Average
62.04Average AccuracyGRAIL
Evaluation results
| Method | Links | ||
|---|---|---|---|
| GRAILModel=Qwen3-8B, Sampling Temperature=0.6, Generations per problem=32026.06 | 62.04 | 67.29 | |
| GRAILModel=Qwen3-4B, Sampling Temperature=0.6, Generations per problem=32026.06 | 58.86 | 66.9 | |
| GRPOModel=Qwen3-8B, Sampling Temperature=0.6, Generations per problem=32026.06 | 57.72 | 65.08 | |
| GRPOModel=Qwen3-4B, Sampling Temperature=0.6, Generations per problem=32026.06 | 55.2 | 64.25 | |
| GRAILModel=R1-Distill-Llama-8B, Sampling Temperature=0.6, Generations per problem=32026.06 | 50.73 | 59.79 | |
| GRPOModel=R1-Distill-Llama-8B, Sampling Temperature=0.6, Generations per problem=32026.06 | 48.18 | 58.57 | |
| GRAILModel=OctoThinker-8B, Sampling Temperature=0.6, Generations per problem=32026.06 | 26.82 | 37.97 | |
| GRPOModel=OctoThinker-8B, Sampling Temperature=0.6, Generations per problem=32026.06 | 23 | 32.84 | |
| GRAILModel=OctoThinker-3B, Sampling Temperature=0.6, Generations per problem=32026.06 | 22.88 | 33.12 | |
| GRPOModel=OctoThinker-3B, Sampling Temperature=0.6, Generations per problem=32026.06 | 19.27 | 29.09 |