Aggregating preference deltas from several weak-weaker model pairs via LoRA adapters and geometric alignment merging improves strong-model performance on reasoning and search benchmarks beyond any single delta.
and Cattell, Raymond B
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
From "Weak" Signals to Strong Models: Preference Delta Aggregation with LoRA Merging
Aggregating preference deltas from several weak-weaker model pairs via LoRA adapters and geometric alignment merging improves strong-model performance on reasoning and search benchmarks beyond any single delta.