Mlx-vlm

What's Changed
* Fix pixtral LM only and tokenizer loading for DS-VL2 by Blaizzy in https://github.com/Blaizzy/mlx-vlm/pull/165


**Full Changelog**: https://github.com/Blaizzy/mlx-vlm/compare/v0.1.8...v0.1.9

What's Changed
* Fix idefics3 vision model type checking by Blaizzy in https://github.com/Blaizzy/mlx-vlm/pull/164


**Full Changelog**: https://github.com/Blaizzy/mlx-vlm/compare/v0.1.7...v0.1.8

What's Changed
* Fix multi-image and 2x speed improvements (DS-VL2) by Blaizzy in https://github.com/Blaizzy/mlx-vlm/pull/157
* Refactor utils (model loading, inference and output processing) by Blaizzy in https://github.com/Blaizzy/mlx-vlm/pull/161
* Fix Llama-3.2-Vision (18x faster generation and 75% less memory usage) by Blaizzy in https://github.com/Blaizzy/mlx-vlm/pull/163

⚠️ Breaking Changes
This release introduces some breaking changes. If you encounter any issues, please open an [issue](https://github.com/Blaizzy/mlx-vlm/issues) or submit a [PR](https://github.com/Blaizzy/mlx-vlm/pulls).


**Full Changelog**: https://github.com/Blaizzy/mlx-vlm/compare/v0.1.6...v0.1.7

What's Changed
* Fixes DeepSeek quant loading by Blaizzy in https://github.com/Blaizzy/mlx-vlm/pull/156
* Fixes Florence-2 LM only by Blaizzy in https://github.com/Blaizzy/mlx-vlm/pull/156


**Full Changelog**: https://github.com/Blaizzy/mlx-vlm/compare/v0.1.5...v0.1.6

What's Changed
* Add support for Deepseek-vl2 by Blaizzy in https://github.com/Blaizzy/mlx-vlm/pull/153
* Add support for Language only inputs by Blaizzy in https://github.com/Blaizzy/mlx-vlm/pull/153


**Full Changelog**: https://github.com/Blaizzy/mlx-vlm/compare/v0.1.4...v0.1.5

What's Changed
* Add support for Paligemma-2 by Blaizzy in https://github.com/Blaizzy/mlx-vlm/pull/142
* Update prompt utils + bump version (Paligemma-2) by Blaizzy in https://github.com/Blaizzy/mlx-vlm/pull/143


**Full Changelog**: https://github.com/Blaizzy/mlx-vlm/compare/v0.1.3...v0.1.4