# lessw2020/Ranger-Deep-Learning-Optimizer

Ranger - a synergistic optimizer using RAdam (Rectified Adam), Gradient Centralization and LookAhead in one codebase

Repository: https://github.com/lessw2020/Ranger-Deep-Learning-Optimizer
Canonical: https://ross.abutalabs.com/products/ranger-deep-learning-optimizer
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2023-12-22T13:24:12+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 2571, "days_push": 985, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1207, forks 176 (observed 2026-08-28T04:03:59.539312+00:00)

## What it is
Ranger is a PyTorch deep learning optimizer that synergistically combines RAdam (Rectified Adam), LookAhead, and Gradient Centralization in a single codebase. It is a drop-in optimizer intended to improve training stability and final accuracy compared to standard Adam/AdamW.

## Use cases
- optimize neural network training in pytorch
- improve image classifier accuracy over adamw
- train deep learning models with radam and lookahead
- apply gradient centralization to conv and fc layers
- beat fastai leaderboard scores with a better optimizer

## When to choose
- you are training PyTorch models and want a stronger optimizer than AdamW
- you want to experiment with RAdam, LookAhead, and Gradient Centralization together
- you are following fastai-style training recipes with flat LR plus cosine annealing

## When to avoid
- you need actively developed optimizers - consider Ranger21 or torch.optim built-ins
- you are not using PyTorch
- you need guaranteed state-of-the-art results on your specific task without benchmarking

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning
- domain: deep-learning, machine-learning
- platform: python
- tags: optimizer, radam, lookahead, gradient-centralization, pytorch

## Member repositories
- lessw2020/Ranger-Deep-Learning-Optimizer (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:59.539312+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 2026-08-30T06:18:59.314694+00:00, confidence not recorded.
  - readme: https://github.com/lessw2020/Ranger-Deep-Learning-Optimizer (fetched 2026-08-28T04:03:59.539312+00:00, sha b1c7b76f55b7)
- Data as of 2026-08-30T08:39:29.467469+00:00.