Releases: google/tf-shell
v0.1.13 Release
Performance improvements in certain add and mul kernels.
Full Changelog: v0.1.12...v0.1.13
v0.1.12
Reduce memory footprint of tf_shell_ml models.
Full Changelog: v0.1.11...v0.1.12
v0.1.11 Release
Highlights:
- tf-shell depends on tensorflow instead of tensorflow-cpu.
- DpSgdSequential and PostScaleSequential support using accelerators to compute jacobian.
Full Changelog: v0.1.10...v0.1.11
v0.1.10 Release
Enable pfor in Jacobian computation of tf_shell_ml models.
Full Changelog: v0.1.9...v0.1.10
v0.1.9 Release
Embedding layers are compatible with tf_shell_ml models.
Full Changelog: v0.1.8...v0.1.9
v0.1.8 Release
Highlights:
Update TensorFlow dependency to 2.18
Full Changelog: v0.1.7...v0.1.8
v0.1.7 Release
Full Changelog: v0.1.6...v0.1.7
v0.1.6 Release
Highlights:
- Better support for distributed execution across multiple machines.
- Key caching to avoid regeneration.
- Initial support for Keras models with DP-SGD and post-scale training algorithms (WIP).
Full Changelog: v0.1.5...v0.1.6
v0.1.5 Release
Downgrade TensorFlow dependency to 2.15.0 for use with tensorflow-privacy python package.
Full Changelog: v0.1.4...v0.1.5
v0.1.4 Release
New features:
- Encrypted segment sum and embedding layer.
- Faster rotation protocol.
- Plaintext-plaintext graph compiler optimization pass.
- Automated parameter generation optimization pass.
Disable builds with Python 3.12 due to TensorFlow dependency on distutils.
Full Changelog: v0.1.2...v0.1.4