Conversational Response Generation on DiscussLLM (test)
2.54Response PerplexityDecoupled
Evaluation Results
| Method | Links | |
|---|---|---|
| DecoupledZero-shot=false, Latency (ms/turn)=5.90, GPU Memory (GB)=0.472025.08 | 2.54 | |
| End-to-endZero-shot=false, Latency (ms/turn)=30.12, GPU Memory (GB)=15.472025.08 | 2.57 |