1.1.1

Changed the license to MIT

What else has Changed

• Fix safety_checker dtype by @ZachNagengast in #276
• Fix missing parameters ('mixed_bit_compression_apply', ) & import pkg (mixed_bit_compression_pre_analysis) issues by @jiangdi0924 in #279
• Fix pipeline.py for Python 3.8 by @TobyRoseman in #287
• Compiled Models in Python by @TobyRoseman in #296
• Use sRGB if available by @stuartjmoore in #301
• Applying mixed bit compression using new optimize API by @TobyRoseman in #311
• Update swift-argument-parser to latest minor version by @godly-devotion in #318

New Contributors

• @stuartjmoore made their first contribution in #301

Full Changelog: 1.1.0...1.1.1

1.1.0

• SDXL iOS and iPadOS Benchmarks
• SDXL refiner Swift inference by @ZachNagengast
• SDXL base Python inference by @lopez-hector
• CUDA RNG in Swift by @liuliu
• Karras schedule for DPMSolver by @pcuenq
• Other improvements

1.0.0

• 6-bit weight compression using coremltools
• Improved attention implementation (SPLIT_EINSUM_V2) which yields up to 30% improved Neural Engine performance
• Multilingual text encoder support
• New benchmarks for iPhone, iPad and Mac

0.4.0

• ControlNet
• Speed improvements to Swift.package scheduler and tokenizer code

0.3.0

Fix: Missing a nil check at StableDiffusionCLI (#138)

0.2.0

Image2image - python (#115)

* Add Encoder model to torch2coreml for image2image

and later for in-paining

* diagonal test with randn

* Revert "diagonal test with randn"

This reverts commit 270afe1bb970c9413d3c3c9d09c0bd4f2dcbfa19.

* readme updates for encoder

* pr comments

0.1.0

Initial release