loss-spike

Here are 2 public repositories matching this topic...

bluorion-com / ZClip

Official implementation of the paper: "ZClip: Adaptive Spike Mitigation for LLM Pre-Training".

loss gradient-clipping adaptive-gradient-clipping loss-spike adaptive-clipping zclip stable-llm-pretraining enable-high-learning-rate traning-stability pre-training-stability stable-training llm-stable-training gradient-norm-clipping

Updated Nov 20, 2025
Python

kaelvalen / trainscope

Star

Post-mortem debugger for LLM training loss spikes. Records per-layer gradients, activations, and weight distributions — scrub back to the exact step that caused the divergence.

debugging machine-learning llm loss-spike

Updated May 18, 2026
Python

Improve this page

Add a description, image, and links to the loss-spike topic page so that developers can more easily learn about it.

Curate this topic

Add this topic to your repo

To associate your repository with the loss-spike topic, visit your repo's landing page and select "manage topics."

Learn more

Provide feedback

Saved searches

Use saved searches to filter your results more quickly