Training Monitoring Python Packages

neuraldbg

Causal debugger for PyTorch training: tells you WHY it failed (vanishing/exploding gradients, NaN, dead neurons) with layer+step root cause. One line in your loop.

neural-dsl

Causal debugger for PyTorch training: tells you WHY it failed (vanishing/exploding gradients, NaN, dead neurons) with layer+step root cause. One line in your loop.

gpualert

Monitor GPU jobs, detect failures, and stay informed during long-running ML experiments.

tb-query

A CLI tool and MCP (Model Context Protocol) server for querying and analyzing TensorBoard event files without requiring a running TensorBoard server.

moewatch

👻 MoEWatch: A lightweight diagnostic and audit library for Mixture-of-Experts (MoE) models in HuggingFace Transformers — the pytest for MoE models.