Privacy-conscious delegation on PAPILLON
86.5Composite ScoreBetterTogether(PO, mmGRPO)
Evaluation Results
| Method | Links | |
|---|---|---|
| BetterTogether(PO, mmGRPO)Strategy=Module-level mmGRPO, Backbone=llama3.12025.08 | 86.5 | |
| MIPROv2 (PO)Strategy=Baseline, Backbone=llama3.12025.08 | 83.9 | |
| mmGRPOStrategy=Module-level mmGRPO, Backbone=llama3.12025.08 | 83.9 | |
| mmGRPOStrategy=Module-level mmGRPO, Backbone=qwen32025.08 | 83.3 | |
| BetterTogether(PO, mmGRPO)Strategy=Module-level mmGRPO, Backbone=qwen32025.08 | 81.1 | |
| Vanilla CoTStrategy=Baseline, Backbone=qwen32025.08 | 78.3 | |
| MIPROv2 (PO)Strategy=Baseline, Backbone=qwen32025.08 | 78.1 | |
| Vanilla CoTStrategy=Baseline, Backbone=llama3.12025.08 | 76.2 |