Time-to-first-token (TTFT) on Llama 8B Instruct 112K input length 3.1
12.7Time-to-First-Token (s)
TriangleMix + MInference
Updated 1mo ago
Evaluation Results
| Method | Links | |
|---|---|---|
2025.07 | 12.7 | |
2025.07 | 13.6 | |
2025.07 | 15 | |
2025.07 | 15.1 | |
2025.07 | 15.9 | |
2025.07 | 19.1 | |
2025.07 | 19.1 | |
2025.07 | 19.3 | |
2025.07 | 27.5 |