Continual Reinforcement Learning on DMC6
1.04Normalized ScoreFT-MT-pre-trained
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| FT-MT-pre-trainedTraining=Multi-task pre-trained2023.06 | 1.04 | 764.38 | — | |
| FT-MT-pre-trainedSetup=Multi-task pre-trained2023.06 | 1.04 | 764.38 | — | |
| FT-MT-scratchTraining=Multi-task from scratch2023.06 | 0.92 | 648.49 | — | |
| FT-MT-scratchSetup=Multi-task scratch2023.06 | 0.92 | 648.49 | — | |
| L2M-oracleMode=Oracle2023.06 | 0.7 | 549.29 | -0.02 | |
| L2M-oracleVariant=Oracle2023.06 | 0.7 | 549.29 | -0.02 | |
| L2MMode=Standard2023.06 | 0.56 | 401.65 | 0.11 | |
| L2M2023.06 | 0.56 | 401.65 | 0.11 | |
| L2P-Pv2Version=Pv22023.06 | 0.32 | 271.39 | 0.02 | |
| L2P-Pv2Variant=Pv22023.06 | 0.32 | 271.39 | 0.02 | |
| EWC2023.06 | 0.3 | 244.7 | 0.49 | |
| EWC2023.06 | 0.3 | 244.7 | 0.49 | |
| FTType=Standard Fine-tuning2023.06 | 0.27 | 225.81 | 0.83 | |
| FTMode=Fine-tuning2023.06 | 0.27 | 225.81 | 0.83 | |
| FT-last+headType=Fine-tuning last layer and head2023.06 | 0.25 | 204.49 | 0.51 | |
| FT-last+headFine-tuned components=last layer and head2023.06 | 0.25 | 204.49 | 0.51 | |
| L2P-PreTVersion=Pre-trained2023.06 | 0.2 | 182.33 | -0.03 | |
| L22023.06 | 0.2 | 192.8 | -0.01 | |
| L2P-PreTVariant=PreT2023.06 | 0.2 | 182.33 | -0.03 | |
| L22023.06 | 0.2 | 192.8 | -0.01 | |
| L2P-PTVersion=PT2023.06 | 0.18 | 158.09 | -0.06 | |
| L2P-PTVariant=PT2023.06 | 0.18 | 158.09 | -0.06 | |
| FT-headType=Fine-tuning head only2023.06 | 0.06 | 58.84 | -0.02 | |
| FT-headFine-tuned components=head only2023.06 | 0.06 | 58.84 | -0.02 |