Issues

• int16 support

  google/aqt#781 · Filter by author EelcoHoogendoorn opened on Aug 9, 2025
  2

• Add support for native asymmetric quantization to AQTv2.

  google/aqt#725 · Filter by author phoenix-meadowlark opened on Sep 19, 2024

• Reduce checkpoint size for Flax models

  google/aqt#720 · Filter by author liamclarkza opened on Sep 17, 2024
  2

• Best practices to reduce memory usage in SERVE mode

  google/aqt#706 · Filter by author mar-muel opened on Sep 5, 2024

• Support Python versions <3.10

  google/aqt#703 · Filter by author mar-muel opened on Aug 28, 2024
  1

• Gradient quantization support

  google/aqt#691 · Filter by author lixcli opened on Aug 9, 2024

• generalized einsum or matmul api for pure jax

  google/aqt#600 · Filter by author sh0416 opened on Apr 27, 2024

• NormalFloat4 support

  google/aqt#599 · Filter by author sh0416 opened on Apr 27, 2024

• How to use it with jnp.einsum?

  google/aqt#595 · Filter by author sh0416 opened on Apr 23, 2024
  1

• Embedded inference or tflite compatibility on the roadmap?

  google/aqt#579 · Filter by author soren-halter opened on Apr 10, 2024

• Binary quantization?

  google/aqt#548 · Filter by author kishorenc opened on Mar 25, 2024

• Can AQT be used to calculate qk score?

  google/aqt#514 · Filter by author Lisennlp opened on Mar 4, 2024
  2

• Implement backprop quantization for convolution.

  feature
  google/aqt#285 · Filter by author lukaszlew opened on Nov 7, 2023

• Refactor config/code classes to follow Flax.

  feature
  google/aqt#281 · Filter by author lukaszlew opened on Nov 7, 2023

• How to use this package to quantize a pretrained model in huggingface, such as BERT / Roberta?

  documentation
  google/aqt#263 · Filter by author TrueNobility303 opened on Nov 2, 2023
  1