Mediator merges fine-tuned LLMs by averaging low-conflict layers and routing high-conflict layers through sparse task-arithmetic experts, with uncertainty-based selection.
This direction discusses the layer-wise training dynamics to help shed some light on the paramter conflicts and the layer-wise adaptivity
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
Mediator: Memory-efficient LLM Merging with Less Parameter Conflicts and Uncertainty Based Routing
Mediator merges fine-tuned LLMs by averaging low-conflict layers and routing high-conflict layers through sparse task-arithmetic experts, with uncertainty-based selection.