List your company

Speculative Decoding on SQL

8.06MATHedgeSpec
0.71762.62384.536.4362Oct 22, 2025

Evaluation results

MethodLinks
HedgeSpecBackbone=LLaMA-3.1-8B-IT, Inference Setting=Greedy2025.108.06103.31
HedgeSpecTarget Model=LLaMA-3.1-8B-IT, Inference Mode=Non-greedy2025.107.8995.81
HedgeSpecBackbone=Qwen-3-8B, Inference Setting=Greedy2025.107.5281.94
HedgeSpecBackbone=Qwen-3-32B, Inference Setting=Greedy2025.107.1645.3
HedgeSpecTarget Model=Qwen-3-8B, Inference Mode=Non-greedy2025.107.1276.75
HedgeSpecTarget Model=Qwen-3-32B, Inference Mode=Non-greedy2025.106.5846.24
EagleBackbone=LLaMA-3.1-8B-IT, Inference Setting=Greedy2025.105.9980.41
Vanilla EagleTarget Model=LLaMA-3.1-8B-IT, Inference Mode=Non-greedy2025.105.9377.99
UCBSpecBackbone=LLaMA-3.1-8B-IT, Inference Setting=Greedy2025.105.7176.58
UCBSpecTarget Model=LLaMA-3.1-8B-IT, Inference Mode=Non-greedy2025.105.3966.85
EXP3SpecBackbone=LLaMA-3.1-8B-IT, Inference Setting=Greedy2025.105.3773.29
UCBSpecBackbone=Qwen-3-8B, Inference Setting=Greedy2025.105.2861.28
UCBSpecTarget Model=Qwen-3-8B, Inference Mode=Non-greedy2025.105.1956.55
EXP3SpecTarget Model=LLaMA-3.1-8B-IT, Inference Mode=Non-greedy2025.105.0764.06
UCBSpecBackbone=Qwen-3-32B, Inference Setting=Greedy2025.105.0737.36
EXP3SpecBackbone=Qwen-3-8B, Inference Setting=Greedy2025.104.9756.74
EXP3SpecTarget Model=Qwen-3-8B, Inference Mode=Non-greedy2025.104.9554.09
EXP3SpecBackbone=Qwen-3-32B, Inference Setting=Greedy2025.104.8835.29
UCBSpecTarget Model=Qwen-3-32B, Inference Mode=Non-greedy2025.104.7834.9
EXP3SpecTarget Model=Qwen-3-32B, Inference Mode=Non-greedy2025.104.5433.54
EagleBackbone=Qwen-3-8B, Inference Setting=Greedy2025.104.244.6
Vanilla EagleTarget Model=Qwen-3-8B, Inference Mode=Non-greedy2025.104.1848.06
Vanilla EagleTarget Model=Qwen-3-32B, Inference Mode=Non-greedy2025.103.0623.04
EagleBackbone=Qwen-3-32B, Inference Setting=Greedy2025.10321.34
LLaMA-3.1-8B-ITTarget Model=LLaMA-3.1-8B-IT, Inference Mode=Non-greedy2025.10118.25
Qwen-3-8BTarget Model=Qwen-3-8B, Inference Mode=Non-greedy2025.10115.14
Qwen-3-32BTarget Model=Qwen-3-32B, Inference Mode=Non-greedy2025.1018.82
LLaMA-3.1-8B-ITBackbone=LLaMA-3.1-8B-IT, Inference Setting=Greedy2025.10117.86
Qwen-3-8BBackbone=Qwen-3-8B, Inference Setting=Greedy2025.10114.65
Qwen-3-32BBackbone=Qwen-3-32B, Inference Setting=Greedy2025.1018.55