List your company

Long-Context Generation on PG-19 20K context length

4Throughput Speedup (micro)SpecPV-4K
1.6082.2292.853.471Dec 2, 2025

Evaluation results

MethodLinks
SpecPV-4KBackbone=LLaMA3.1-8B-Instruct, Partial KV Budget=4K2025.1243.57
SpecPV-2KBackbone=LLaMA3.1-8B-Instruct, Partial KV Budget=2K2025.123.833.42
SpecPV-8KBackbone=LLaMA3.1-8B-Instruct, Partial KV Budget=8K2025.123.543.51
EAGLE3-YARNBackbone=LLaMA3.1-8B-Instruct2025.122.743.49
TriForceBackbone=LLaMA3.1-8B-Instruct, Partial KV Budget=4K2025.122.442.82
TokenSwiftBackbone=LLaMA3.1-8B-Instruct, Partial KV Budget=4K2025.121.71.73