v0.2.7

What's Changed

• Create Dockerfile_311_cuda by @valsdav in #19
• Docker images with multi-platform support by @riga in #17
• Fix model preparation for custom signatures by @Bogdan-Wiederspan in #22
• Fix model preparation for custom signatures (#22) by @riga in #23
• Add test for tf 2.18.0. by @riga in #24

Full Changelog: v0.2.6...v0.2.7

v0.2.6

What's Changed

• Fix XLA flag encoding. by @riga in #18

Full Changelog: v0.2.5...v0.2.6

v0.2.5

• Allow passing arbitrary flags to the aot compiler. #15

v0.2.4

• #14 Forward XLA flags for compilation

v0.2.3

• Fix provisioning imports in tf tools. #13 from @riga

v0.2.2

• bugfixes for TensorFlow XLA/AOT
• added pytorch to docker images

v0.2.1

• Add missing entry points. (4ae8f0c)

v0.2.0

Breaking changes

• Require Python 3.7 or higher (#7) (f5ec832)

Features and updates

• Add AOT tools (#10) (403d57e)
• Update project packaging (fda3989, 6cc7c71, 0e06944, 9435360)
• Update docker images (1d07dda, 345b3d4, d07bc05)

Fixes

None

v0.1.2

• Adapt to changes in TF 2.5 (#6) (27a1dd1)
• [keras] Log stats of GPUStatsLogger after eopch. (4c41fd1)
• Update required six version. (17cafe3)
• Add tf to docs requirements. (925f3f4)
• Add option to silence logs in import_tf. (c4f9951)
• Rename lazy_importer -> lazy loader, add tests. (56265b9)
• Make stats in GPUStatsLogger configurable. (55be4ea)
• Make submodule imports lazy. (5cf9c21)

v0.1.1

• Fix packaging. (e869392)
• Simplify documentation. (d8c81cb)
• Add initial installation to images. (90444ce)