GPT-Neo 125M + Self-Sampling

Reported on 2 benchmarks across 1 task · 1 paper

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Reasoning2 results

Arithmetic ReasoningonGSM8K
Accuracy· 2022-05-28
7.5
best: 97.72 (Claude 3.5 Sonnet (HPT))
Learning Math Reasoning from Self-Sampled Correct and Partially-Correct Solutions arXiv:2205.14318
Arithmetic ReasoningonGSM8K
Parameters (Billion)· 2022-05-28
0.125
best: 540 (PaLM 540B (Self Improvement, Self Consistency))
Learning Math Reasoning from Self-Sampled Correct and Partially-Correct Solutions arXiv:2205.14318