# google-research/morph-net

Fast & Simple Resource-Constrained Learning of Deep Network Structure

Repository: https://github.com/google-research/morph-net
Canonical: https://ross.abutalabs.com/products/morph-net
Language: Python
License: Apache-2.0
License Family: permissive
Topics: neural-architecture-search, automl, tensorflow, deep-learning, python, machine-learning
Last push: 2026-07-02T11:28:00+00:00

## Health v2 (maintenance only)
Score: 73/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 90, release rhythm 35, longevity 100
- inputs: {"age_days": 2731, "days_push": 62, "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 1038, forks 151 (observed 2026-08-28T04:03:19.576075+00:00)

## What it is
MorphNet is a TensorFlow library for resource-constrained learning of deep network structure. It adds regularizers during training that induce activation sparsity, pruning channels to optimize targets like FLOPs, model size, or latency.

## Use cases
- shrink a convolutional neural network to fit memory or latency constraints
- prune filters from an existing seed network during training
- perform differentiable architecture search with FiGS
- reduce model size for mobile or embedded deployment
- optimize a network for a target cost such as FLOPs or device latency

## When to choose
- you have a working TensorFlow model and want to compress it under a resource budget
- you want structure learning without changing network topology

## When to avoid
- you need to change layer counts or connectivity patterns of the network
- you work primarily outside TensorFlow or need actively maintained NAS tooling

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning
- domain: machine-learning, deep-learning
- platform: python
- tags: neural-architecture-search, automl, pruning, tensorflow, model-compression, network-structure-learning, gpu

## Member repositories
- google-research/morph-net (main) score 73

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:19.576075+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 2026-08-30T07:04:15.466375+00:00, confidence not recorded.
  - readme: https://github.com/google-research/morph-net (fetched 2026-08-28T04:03:19.576075+00:00, sha e805783223ed)
- Data as of 2026-08-30T08:39:29.467469+00:00.