Decoding Speedup on InfiniteBench (Long-context understanding)
1.48Decoding SpeedupPRR
Evaluation Results
| Method | Links | |
|---|---|---|
| PRRModel=DeepSeek-R1, DSA mechanism=InfLLM-v22026.06 | 1.48 | |
| PRRModel=Llama3, DSA mechanism=InfLLM-v22026.06 | 1.45 | |
| PRRModel=Qwen3-14B, DSA mechanism=InfLLM-v22026.06 | 1.43 | |
| PRRModel=DeepSeek-R1, DSA mechanism=Quest2026.06 | 1.42 | |
| PRRModel=GLM-4, DSA mechanism=Quest2026.06 | 1.4 | |
| PRRModel=Llama3, DSA mechanism=Quest2026.06 | 1.4 | |
| PRRModel=GLM-Z1, DSA mechanism=Quest2026.06 | 1.39 | |
| PRRModel=GLM-4, DSA mechanism=InfLLM-v22026.06 | 1.38 | |
| PRRModel=GLM-Z1, DSA mechanism=InfLLM-v22026.06 | 1.35 | |
| PRRModel=Qwen3-32B, DSA mechanism=Quest2026.06 | 1.34 | |
| PRRModel=Qwen3-14B, DSA mechanism=Quest2026.06 | 1.27 | |
| PRRModel=Qwen3-32B, DSA mechanism=InfLLM-v22026.06 | 1.27 |