4 papers · 1 filter
One LR Doesn't Fit All: Heavy-Tail Guided Layerwise Learning Rates for LLMs
Di He, Songjun Tu, Keyu Wang +2
Learning rate configuration is a fundamental aspect of modern deep learning. The prevailing practice of applying a uniform learning rate across all layers overlooks the structural…
Chain-of-Experts: Unlocking the Communication Power of Mixture-of-Experts Models
Zihan Wang, Rui Pan, Jiarui Yao +7
We propose Chain-of-Experts (CoE), a new Mixture-of-Experts (MoE) architecture that introduces sequential expert communication within each layer. Unlike traditional MoE models, whe…
LIFT the Veil for the Truth: Principal Weights Emerge after Rank Reduction for Reasoning-Focused Supervised Fine-Tuning
Zihang Liu, Tianyu Pang, Oleg Balabanov +5
Recent studies have shown that supervised fine-tuning of LLMs on a small number of high-quality datasets can yield strong reasoning capabilities. However, full fine-tuning (Full FT…
NeuroTrails: Training with Dynamic Sparse Heads as the Key to Effective Ensembling
Bram Grooten, Farid Hasanov, Chenxiang Zhang +9
Model ensembles have long been a cornerstone for improving generalization and robustness in deep learning. However, their effectiveness often comes at the cost of substantial compu…