Creative Writing on ROCStories + WritingPrompts Llama-3-70B (test)
63.3QD-ScoreQD-LLM
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| QD-LLMBackbone=Llama-3-70B-Instruct, Budget=5002026.05 | 63.3 | 63.1 | 0.52 | 0.28 | 0.96 | |
| CMA-ME (ad.)Backbone=Llama-3-70B-Instruct, Budget=5002026.05 | 46.2 | 46 | 0.39 | 0.38 | — | |
| QDAIFBackbone=Llama-3-70B-Instruct, Budget=5002026.05 | 44.8 | 44.6 | 0.37 | 0.4 | — | |
| Best-of-N+MMRBackbone=Llama-3-70B-Instruct, Budget=5002026.05 | 40.1 | 39.9 | 0.31 | 0.46 | — | |
| Diverse BeamBackbone=Llama-3-70B-Instruct, Budget=5002026.05 | 38.2 | 38 | 0.28 | 0.52 | — | |
| NucleusBackbone=Llama-3-70B-Instruct, p=0.95, Budget=5002026.05 | 36.4 | 36.2 | 0.26 | 0.56 | — |