# qualcomm/aimet

AIMET is a library that provides advanced quantization and compression techniques for trained neural network models.

Repository: https://github.com/qualcomm/aimet
Canonical: https://ross.abutalabs.com/products/aimet
Homepage: https://quic.github.io/aimet-pages/index.html
Language: Python
License: NOASSERTION
License Family: other
Topics: quantization, deep-learning, compression, open-source, machine-learning, pruning, auto-ml, network-compression, deep-neural-networks, network-quantization, opensource
Last push: 2026-08-26T19:01:20+00:00

## Health v2 (maintenance only)
Score: 99/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 99, release rhythm 99, longevity 100
- inputs: {"age_days": 2325, "days_push": 7, "days_rel": 9, "gap_med": 13.0, "n_releases_24m": 47}
- flags: no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2688, forks 460 (observed 2026-08-28T04:07:10.847379+00:00)

## What it is
AIMET (AI Model Efficiency Toolkit) is a Python library from Qualcomm providing advanced quantization and compression techniques for trained neural network models. It supports PyTorch and ONNX models, using post-training and fine-tuning techniques to minimize accuracy loss while reducing memory footprint and compute load for edge deployment.

## Use cases
- quantize a trained PyTorch model to 8-bit integers
- compress a deep learning model for mobile deployment
- reduce memory footprint of an ONNX model for edge devices
- apply post-training quantization without losing accuracy
- prune channels from a neural network to speed up inference
- prepare a model to run fast on Qualcomm Hexagon DSP

## When to choose
- you need to quantize or compress PyTorch or ONNX models for edge/mobile inference
- you want advanced techniques like data-free quantization or cross-layer equalization to preserve accuracy
- you target Qualcomm hardware such as Hexagon DSP

## When to avoid
- you need quantization-aware training for frameworks other than PyTorch or ONNX, such as TensorFlow
- you just want a simple one-line quantization API without tuning options
- your project is unrelated to model efficiency or deployment optimization

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, deep-learning, llm-training
- domain: machine-learning, deep-learning, gpu-computing
- platform: python
- tags: quantization, model-compression, pruning, post-training-quantization, pytorch, onnx, edge-deployment, model-optimization, linux, gpu

## Member repositories
- qualcomm/aimet (main) score 99

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:07:10.847379+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 2026-08-30T02:17:44.690645+00:00, confidence not recorded.
  - readme: https://github.com/qualcomm/aimet (fetched 2026-08-28T04:07:10.847379+00:00, sha be14b1c6599c)
  - homepage: https://quic.github.io/aimet-pages/index.html (fetched 2026-08-29T09:59:48.755938+00:00, sha 2593f6bc9344)
- Data as of 2026-08-30T08:39:29.467469+00:00.