madroidmaq/mlx-omni-server v0.4.4

on GitHub

What's Changed

• Support speculative-decoding/draft-model by @madroidmaq in #52
• Refine model logic by utilizing the include_details field to display additional information by @madroidmaq in #61
• Clip logprobs to -100 to avoid JSON encoder errors by @cbowdon in #56
• [Break] Remove mlx-audio, see Blaizzy/mlx-audio#207 (comment)
• Upgrade mlx-lm to 0.26.3, supporting more models, see https://github.com/ml-explore/mlx-lm/releases

New Contributors

• @cbowdon made their first contribution in #56

Full Changelog: v0.4.3...v0.4.4