List your company

Speculative Decoding on MBPP

6.75SpeedupTAPS
1.2382.6694.15.531May 30, 2026May 31, 2026Jun 1, 2026Jun 2, 2026

Evaluation results

MethodLinks
TAPSHardware=NVIDIA A40-48GB GPU, Target Model=Qwen3-4B2026.056.757.86
TreeFlashTarget model=Qwen3 4B, Temperature (T)=0, Draft budget (B)=642026.066.448.96
TreeFlashTarget model=Qwen3 8B, Temperature (T)=0, Draft budget (B)=642026.066.418.96
TAPSHardware=NVIDIA A800-SXM4-80GB GPU, Target Model=Qwen3-4B2026.056.257.97
TAPSHardware=NVIDIA A40-48GB GPU, Target Model=Qwen3-8B2026.056.117.89
TreeFlashTarget model=Qwen3 4B, Temperature (T)=1, Draft budget (B)=642026.065.958.34
DDTreeTarget model=Qwen3 4B, Temperature (T)=0, Draft budget (B)=642026.065.857.9
DDTreeTarget model=Qwen3 8B, Temperature (T)=0, Draft budget (B)=642026.065.737.68
DDTreeHardware=NVIDIA A800-SXM4-80GB GPU, Target Model=Qwen3-4B2026.055.689
TAPSHardware=NVIDIA A800-SXM4-80GB GPU, Target Model=Qwen3-8B2026.055.667.67
TreeFlashTarget model=Qwen3 8B, Temperature (T)=1, Draft budget (B)=642026.065.667.86
DDTreeTarget model=Qwen3 4B, Temperature (T)=1, Draft budget (B)=642026.065.397.29
DDTreeTarget model=Qwen3 8B, Temperature (T)=1, Draft budget (B)=642026.065.156.95
TreeFlashTarget model=Qwen3 4B, Temperature (T)=0, Draft budget (B)=162026.065.057.27
TreeFlashTarget model=Qwen3 8B, Temperature (T)=0, Draft budget (B)=162026.065.037.24
TAPSHardware=NVIDIA A800-SXM4-80GB GPU, Target Model=LLaMA-3.1-8B-Instruct2026.054.916.66
DDTreeTarget model=Qwen3 4B, Temperature (T)=0, Draft budget (B)=162026.064.916.8
DFlashHardware=NVIDIA A40-48GB GPU, Target Model=Qwen3-4B2026.054.785.68
DDTreeTarget model=Qwen3 8B, Temperature (T)=0, Draft budget (B)=162026.064.686.47
TreeFlashTarget model=Qwen3 4B, Temperature (T)=1, Draft budget (B)=162026.064.676.73
DFlashHardware=NVIDIA A40-48GB GPU, Target Model=Qwen3-8B2026.054.655.82
DDTreeTarget model=Qwen3 4B, Temperature (T)=1, Draft budget (B)=162026.064.526.22
TAPSHardware=NVIDIA A40-48GB GPU, Target Model=LLaMA-3.1-8B-Instruct2026.054.486.69
TreeFlashTarget model=Qwen3 8B, Temperature (T)=1, Draft budget (B)=162026.064.446.37
DFlashHardware=NVIDIA A800-SXM4-80GB GPU, Target Model=Qwen3-4B2026.054.425.66
DFlashTarget model=Qwen3 4B, Temperature (T)=0, Draft budget (B)=162026.064.375.83
DFlashTarget model=Qwen3 8B, Temperature (T)=0, Draft budget (B)=162026.064.365.87
DDTreeTarget model=Qwen3 8B, Temperature (T)=1, Draft budget (B)=162026.064.255.91
DFlashHardware=NVIDIA A800-SXM4-80GB GPU, Target Model=Qwen3-8B2026.054.045.21
DFlashTarget model=Qwen3 4B, Temperature (T)=1, Draft budget (B)=162026.063.965.31
DFlashTarget model=Qwen3 8B, Temperature (T)=1, Draft budget (B)=162026.063.644.93
DDTreeHardware=NVIDIA A800-SXM4-80GB GPU, Target Model=Qwen3-8B2026.053.598.79
DFlashHardware=NVIDIA A40-48GB GPU, Target Model=LLaMA-3.1-8B-Instruct2026.053.575.02
DDTreeHardware=NVIDIA A40-48GB GPU, Target Model=Qwen3-4B2026.053.548.73
DFlashHardware=NVIDIA A800-SXM4-80GB GPU, Target Model=LLaMA-3.1-8B-Instruct2026.053.525.02
DDTreeHardware=NVIDIA A800-SXM4-80GB GPU, Target Model=LLaMA-3.1-8B-Instruct2026.052.677.43
DDTreeHardware=NVIDIA A40-48GB GPU, Target Model=Qwen3-8B2026.052.659.02
EAGLE3Target model=Qwen3 8B, Temperature (T)=0, Draft budget (B)=642026.061.963.35
EAGLE3Target model=Qwen3 4B, Temperature (T)=0, Draft budget (B)=642026.061.943.38
EAGLE-3Hardware=NVIDIA A40-48GB GPU, Target Model=LLaMA-3.1-8B-Instruct2026.051.872.62
EAGLE3Target model=Qwen3 4B, Temperature (T)=1, Draft budget (B)=642026.061.873.32
EAGLE3Target model=Qwen3 8B, Temperature (T)=1, Draft budget (B)=642026.061.863.25
DDTreeHardware=NVIDIA A40-48GB GPU, Target Model=LLaMA-3.1-8B-Instruct2026.051.847.44
EAGLE-3Hardware=NVIDIA A40-48GB GPU, Target Model=Qwen3-4B2026.051.782.95
EAGLE3Target model=Qwen3 8B, Temperature (T)=0, Draft budget (B)=162026.061.742.97
EAGLE-3Hardware=NVIDIA A40-48GB GPU, Target Model=Qwen3-8B2026.051.692.82
EAGLE-3Hardware=NVIDIA A800-SXM4-80GB GPU, Target Model=Qwen3-4B2026.051.692.89
EAGLE3Target model=Qwen3 4B, Temperature (T)=0, Draft budget (B)=162026.061.692.94
EAGLE-3Hardware=NVIDIA A800-SXM4-80GB GPU, Target Model=Qwen3-8B2026.051.642.74
EAGLE3Target model=Qwen3 4B, Temperature (T)=1, Draft budget (B)=162026.061.642.88
EAGLE3Target model=Qwen3 8B, Temperature (T)=1, Draft budget (B)=162026.061.632.8
EAGLE-3Hardware=NVIDIA A800-SXM4-80GB GPU, Target Model=LLaMA-3.1-8B-Instruct2026.051.452.07