Gradient-based sensitivity scoring automatically allocates LoRA-MoE expert budget across parameter blocks; the module-separated variant matches or slightly beats prior methods with fewer trainable parameters.
Sketch-fusion: a gradient compression method with multi-layer fusion for communication-efficient distributed training,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
A Sensitivity-Driven Expert Allocation Method in LoRA-MoE for Efficient Fine-Tuning
Gradient-based sensitivity scoring automatically allocates LoRA-MoE expert budget across parameter blocks; the module-separated variant matches or slightly beats prior methods with fewer trainable parameters.