List your company

Mathematical Reasoning on GSM8K (Acc, Tp, Spd)

92.6Accuracy (GSM8K)Vanilla Decoding
91.45691.75392.0592.347Apr 15, 2026

Evaluation results

MethodLinks
Vanilla DecodingModel Series=Llama-3 (70B/1B), Shot=8-shot, Protocol=Pass@12026.0492.613.41
SWIFTModel Series=Llama-3 (70B/1B), Shot=8-shot, Protocol=Pass@12026.0492.612.90.96
LookaheadModel Series=Llama-3 (70B/1B), Shot=8-shot, Protocol=Pass@12026.0492.611.10.83
CSDModel Series=Llama-3 (70B/1B), Shot=8-shot, Protocol=Pass@12026.0492.624.51.83
SpecDecodeModel Series=Llama-3 (70B/1B), Shot=8-shot, Protocol=Pass@12026.0492.323.31.74
Lossy SDModel Series=Llama-3 (70B/1B), Shot=8-shot, Protocol=Pass@1, τ=0.62026.0492.324.21.8
CSDModel Series=Qwen-2.5 (72B/7B), Shot=8-shot, Protocol=Pass@12026.0492.225.21.81
Vanilla DecodingModel Series=Qwen-2.5 (72B/7B), Shot=8-shot, Protocol=Pass@12026.0491.713.91
SpecDecodeModel Series=Qwen-2.5 (72B/7B), Shot=8-shot, Protocol=Pass@12026.0491.522.21.59