Training Tricks - a Sasikanth Collection

Sasikanth 's Collections

Training Tricks

Training Tricks

updated 1 day ago

ZClip: Adaptive Spike Mitigation for LLM Pre-Training

Paper • 2504.02507 • Published 3 days ago • 71
Variance Control via Weight Rescaling in LLM Pre-training

Paper • 2503.17500 • Published 16 days ago • 5