List your company

Long-Context Generation on PG-19 40K context length

5.46Throughput Speedup (micro)SpecPV-4K
1.84082.78043.724.6596Dec 2, 2025

Evaluation results

MethodLinks
SpecPV-4KBackbone=LLaMA3.1-8B-Instruct, Partial KV Budget=4K2025.125.463.67
SpecPV-2KBackbone=LLaMA3.1-8B-Instruct, Partial KV Budget=2K2025.125.213.55
SpecPV-8KBackbone=LLaMA3.1-8B-Instruct, Partial KV Budget=8K2025.124.883.56
TriForceBackbone=LLaMA3.1-8B-Instruct, Partial KV Budget=4K2025.123.12.8
EAGLE3-YARNBackbone=LLaMA3.1-8B-Instruct2025.123.023.57
TokenSwiftBackbone=LLaMA3.1-8B-Instruct, Partial KV Budget=4K2025.121.981.77