HyperAIHyperAI

Command Palette

Search for a command to run...

各模型在此基准测试上的性能结果

指标

Avg.
MATH
ASDiv
GSM-H
GSM8K
SVAMP
RuleTaker
ProofWriter
Agent MiniWob++
MiniWob++
Ours
1-shot
STAR+Env.
Distill GPT4
iter. SFT+DPO
Self-Rewarding
Distill Claude2
3 总计