# julvo/reloading

Change Python code while it's running without losing state

Repository: https://github.com/julvo/reloading
Canonical: https://ross.abutalabs.com/products/reloading
Language: Python
License: MIT
License Family: permissive
Topics: python, pytorch, utility, interactive, deep-learning, hot-reloading, hot-reload, machine-learning, keras
Last push: 2024-06-15T18:50:01+00:00

## Health v2 (maintenance only)
Score: 23/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 8, longevity 100
- inputs: {"age_days": 2609, "days_push": 809, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1127, forks 33 (observed 2026-08-28T04:03:41.558880+00:00)

## What it is
A small Python library that reloads the source code of loops or functions on each iteration or invocation without losing runtime state. It is primarily used to edit code while long-running deep learning training jobs are in progress.

## Use cases
- edit training loop code without restarting a pytorch training run
- hot reload a python function while a long job is running
- add logging or statistics to a running training loop
- keep training state while changing model saving code
- create an endless loop whose body can be edited live

## When to choose
- you run long deep learning training loops and want to tweak code mid-run
- you want live-editable loop bodies or functions with state preserved

## When to avoid
- you need full module or web app hot reloading
- your loop is extremely tight and reload overhead matters
- you need to reload code outside loop bodies or decorated functions

## Facets
- artifact type: library
- maturity: maintenance
- function: developer-tools, workflow-automation
- domain: developer-tools, deep-learning, machine-learning
- platform: python, cross-platform
- tags: hot-reload, live-coding, training-loops, decorator, pytorch, keras

## Member repositories
- julvo/reloading (main) score 23

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:41.558880+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T06:38:48.027619+00:00, confidence not recorded.
  - readme: https://github.com/julvo/reloading (fetched 2026-08-28T04:03:41.558880+00:00, sha f7d4b384495a)
  - registry_pypi: https://pypi.org/pypi/reloading/json (fetched 2026-08-29T12:43:25.701229+00:00, sha a1328ea84a8a)
- Data as of 2026-08-30T08:39:29.467469+00:00.