Multiple-type control summarization on NYT (test)
50.04R1GPT-L Prompt+filter
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| GPT-L Prompt+filterModel=GPT-L, Setting=Prompt+filter2024.06 | 50.04 | 32.65 | 44.35 | 69.48 | 4.82 | |
| GPT-L Prompt+RL+filterModel=GPT-L, Setting=Prompt+RL+filter2024.06 | 50.01 | 32.52 | 44.14 | 69.51 | 4.6 | |
| GPT-M Prompt+filterModel=GPT-M, Setting=Prompt+filter2024.06 | 49.73 | 32.65 | 44.55 | 69 | 6.75 | |
| GPT-M Prompt+RL+filterModel=GPT-M, Setting=Prompt+RL+filter2024.06 | 49.41 | — | 44.05 | 68.4 | 3.65 | |
| GPT-L PromptModel=GPT-L, Setting=Prompt2024.06 | 49.41 | 32.2 | 44.31 | 69.36 | 10.64 | |
| GPT-L Prompt+RLModel=GPT-L, Setting=Prompt+RL2024.06 | 49.35 | 32.24 | 44.31 | 69.27 | 8.52 | |
| GPT-M PromptModel=GPT-M, Setting=Prompt2024.06 | 48.34 | 30.74 | 43.64 | 68.75 | 13.17 | |
| GPT-M Prompt+RLModel=GPT-M, Setting=Prompt+RL2024.06 | 48.23 | 30.58 | 43.61 | 68.67 | 12.61 | |
| GPT-S Prompt+filterModel=GPT-S, Setting=Prompt+filter2024.06 | 47.88 | 30.55 | 42.5 | 67.87 | 8.06 | |
| GPT-S Prompt+RL+filterModel=GPT-S, Setting=Prompt+RL+filter2024.06 | 47.84 | 30.43 | 42.26 | 67.54 | 3.89 | |
| GPT-S Prompt+RLModel=GPT-S, Setting=Prompt+RL2024.06 | 47.3 | 29.84 | 42.36 | 67.81 | 10.53 | |
| GPT-S PromptModel=GPT-S, Setting=Prompt2024.06 | 47.22 | 29.47 | 42.01 | 67.76 | 17.62 |