# catalyst-team/catalyst

Accelerated deep learning R&D

Repository: https://github.com/catalyst-team/catalyst
Canonical: https://ross.abutalabs.com/products/catalyst
Homepage: https://catalyst-team.com
Language: Python
License: Apache-2.0
License Family: permissive
Topics: deep-learning, reinforcement-learning, machine-learning, computer-vision, pytorch, python, distributed-computing, infrastructure, research, reproducibility, image-processing, image-classification, image-segmentation, object-detection, natural-language-processing, text-classification, text-segmentation, information-retrieval, recommender-system, metric-learning
Last push: 2026-07-08T21:40:01+00:00

## Health v2 (maintenance only)
Score: 64/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 91, release rhythm 8, longevity 100
- inputs: {"age_days": 2935, "days_push": 56, "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 3382, forks 397 (observed 2026-08-28T04:07:58.674049+00:00)

## What it is
Catalyst is a high-level PyTorch framework for deep learning research and development, focused on reproducibility, rapid experimentation, and codebase reuse. It provides configurable training loops, callbacks, and distributed training support so researchers avoid rewriting boilerplate training code.

## Use cases
- run pytorch training loops without boilerplate
- reproducible deep learning experiments
- distributed multi-gpu training in pytorch
- image classification and segmentation pipelines
- metric learning with triplet and contrastive losses
- config-driven experiment management for research
- train object detection models in pytorch

## When to choose
- you want a high-level, config-driven training framework on top of PyTorch
- reproducibility and rapid experimentation matter more than full control of the loop
- you need built-in distributed training, callbacks, and logging integrations

## When to avoid
- you prefer writing plain PyTorch or using more actively maintained alternatives like PyTorch Lightning
- you need the latest model architectures or ecosystem features out of the box
- your project depends on very recent PyTorch versions with tight compatibility

## Facets
- artifact type: framework
- maturity: maintenance
- function: deep-learning, machine-learning, llm-training, computer-vision, nlp, image-processing, gpu-computing
- domain: deep-learning, machine-learning, computer-vision, data-science
- platform: python
- tags: pytorch, training-loop, reproducibility, distributed-training, experiment-tracking, research, natural-language-processing, linux, macos, gpu

## Member repositories
- catalyst-team/catalyst (main) score 64

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:07:58.674049+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 2026-08-29T18:39:40.041407+00:00, confidence not recorded.
  - readme: https://github.com/catalyst-team/catalyst (fetched 2026-08-28T04:07:58.674049+00:00, sha 2557e262d0b7)
  - homepage: https://catalyst-team.com (fetched 2026-08-29T09:33:28.042125+00:00, sha 45cfdd29314e)
  - registry_pypi: https://pypi.org/pypi/catalyst/json (fetched 2026-08-29T09:33:28.044856+00:00, sha b2a83abd8b84)
- Data as of 2026-08-30T08:39:29.467469+00:00.