FudanNLP/fitlog

Health v2 · maintenance only

Adoption not part of the score

1512 stars · 131 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

fitlog (fast + git + log) is a Python tool that helps deep learning practitioners record training logs and manage experiment code, combining automatic git commits with metric/loss logging. It includes a CLI and a local web server for browsing, searching, and comparing experiment runs.

Use cases

• log metrics and losses during deep learning training
• automatically commit code versions with each experiment run
• compare hyperparameter tuning results in a web dashboard
• filter and search past experiment logs by metric ranges
• notify a Feishu bot when a training run finishes or fails
• silence logging during debugging or in non-main distributed processes

When to choose

• you want a lightweight, git-integrated experiment logger for PyTorch/deep learning projects
• you need a local web UI to compare hyperparameter sweeps without cloud services
• you work on Linux/macOS (or Windows Git Bash) and want minimal setup

When to avoid

• you need a mature, actively developed experiment tracker with broad integrations (consider MLflow or Weights & Biases)
• you train on Windows outside Git Bash
• you require multi-process (multiprocessing) logging or extensive documentation and community support

Facets

library · maturity maintenance

logging developer-tools cli machine-learning deep-learning machine-learning deep-learning developer-tools windows python cli experiment-tracking hyperparameter-tuning git-integration training-logs web-dashboard fastnlp command-line linux macos

Member repositories

For agents

markdown · JSON · MCP: product_card(name="FudanNLP/fitlog")