# Tencent/PocketFlow

An Automatic Model Compression (AutoMC) framework for developing smaller and faster AI applications.

Repository: https://github.com/Tencent/PocketFlow
Canonical: https://ross.abutalabs.com/products/tencent-pocketflow
Homepage: https://pocketflow.github.io
Language: Python
License: NOASSERTION
License Family: other
Topics: deep-learning, model-compression, mobile-app, automl, computer-vision
Last push: 2023-03-31T22:23:33+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 2863, "days_push": 1251, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2909, forks 489 (observed 2026-08-28T04:07:29.170221+00:00)

## What it is
PocketFlow is an open-source AutoML framework from Tencent AI Lab for automatically compressing and accelerating deep learning models. Developers specify a target compression ratio and it searches hyper-parameters to produce smaller, faster models suitable for mobile deployment.

## Use cases
- compress a deep learning model for mobile deployment
- prune channels of a CNN automatically
- quantize neural network weights
- speed up model inference with minimal accuracy loss
- run neural architecture search for compressed models
- deploy computer vision models on resource-limited devices

## When to choose
- you need to shrink a TensorFlow model for mobile or embedded inference
- you want automated pruning, sparsification, or quantization without manual tuning
- you can afford a search loop over compression hyper-parameters

## When to avoid
- you work primarily in PyTorch rather than TensorFlow
- you need actively maintained tooling - the project has seen few updates since 2023
- you want one-shot compression without an automated search process

## Facets
- artifact type: framework
- maturity: maintenance
- function: machine-learning, deep-learning, llm-training
- domain: deep-learning, machine-learning, computer-vision, mobile-development
- platform: python
- tags: model-compression, automl, quantization, pruning, neural-network-acceleration, tensorflow, linux, gpu

## Member repositories
- Tencent/PocketFlow (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:07:29.170221+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 2026-08-30T07:34:21.257615+00:00, confidence not recorded.
  - readme: https://github.com/Tencent/PocketFlow (fetched 2026-08-28T04:07:29.170221+00:00, sha 5cb851eb5389)
  - homepage: https://pocketflow.github.io (fetched 2026-08-29T09:49:32.467610+00:00, sha 5d2d5a57e00b)
- Data as of 2026-08-30T08:39:29.467469+00:00.