{"categories":[{"label":"Artificial Intelligence","url":"https://skillfed.io/packages/category/scientific-engineering-artificial-intelligence/5"}],"enrichment":{"capability":"AQT provides quantization of tensor operations (matmul, einsum, conv) in JAX computations, enabling int8 and lower-bit model training with minimal configuration while maintaining training-serving consistency.","skillfed_tags":["quantization","jax","neural-networks"],"use_cases":["Train large language models or vision models with int8 quantization on TPU or GPU clusters to reduce memory and accelerate computation.","Conduct quantization research by configuring forward and backward bit widths independently and testing different calibration strategies.","Deploy models trained with AQT without retraining or post-training quantization, since quantized weights are identical during training and serving.","Integrate quantization into existing Flax models by replacing the dot_general operation without restructuring the model.","Benchmark lower-bit training on production workloads to evaluate accuracy-efficiency tradeoffs."],"what_it_does":"AQT is a quantization library for JAX that replaces standard tensor operations (dot_general, einsum, convolution) with quantized variants, enabling training of neural networks at reduced precision (int8, int4, or lower) without hand-tuning. It works by injecting quantized operations into JAX-based frameworks like Flax, Pax, and MaxText, and is designed to maintain bit-exact consistency between training and serving\u2014avoiding the training-serving bias common in post-training quantization.\n\nThe library provides flexible configuration of forward and backward pass quantization separately, with support for various numerics (int8, int4, bfloat16, float8), calibration algorithms, and stochastic rounding. It is built on top of jax, jaxlib, flax, and absl-py, and is actively maintained by Google with demonstrated production use at scale and research validation across multiple papers and frameworks.","worth_installing":"Yes, if you are training neural networks in JAX and need quantization. The library is actively maintained, has no known vulnerabilities, carries a permissive license, and is backed by Google's production use at scale. Install friction is low. The main constraint is a hard dependency on jax and jaxlib (which require compilation for your hardware), and the library is still in alpha (Development Status :: 3), so expect API changes. Best suited for researchers and practitioners already committed to the JAX ecosystem."},"id":"aqtp","links":{"html":"https://skillfed.io/packages/aqtp","md":"https://skillfed.io/packages/aqtp.md","pypi":"https://pypi.org/project/aqtp/"},"maintenance":{"status":"active"},"meta":{"latest_release":"2025-08-01","license_spdx":null,"license_treatment":"permissive","name":"aqtp","python_support":"supports_current","summary":"Accurate Quantized Training library."},"popularity":{"monthly_downloads":457805,"position":6547,"tier":"top_15000"},"security":{"n_vulnerabilities":0},"version":"0.9.0"}