Prince Canuma comments

Results 572 comments of


                                            Prince Canuma

Request Multilingual TTS with Kokoro 1.0

Thank you very much! We need to investigate this further but it's hard because I only speak Portuguese, Spanish, English, and a bit of Polish and Hindi. Could you help...

Unrecognized image processor in mlx-community/Qwen2.5-VL-7B-Instruct-4bit

Hey @JoeJoe1313 @leoho0722 This is an issue on the transformers side It seems the Qwen2.5VLImageProcessor class was delete as it's identical with Qwen2VL. The fix is to either: 1. Change...

Unrecognized image processor in mlx-community/Qwen2.5-VL-7B-Instruct-4bit

Thanks @neilmehta24! It will definetly be.

Video2Text Inference is slow and high vram consumption

Could you share the specs of your machine?

Video2Text Inference is slow and high vram consumption

I would recommend: 1. Trying 8bit or 4bit quants. 2. Trying the 2B version. 3. Or lowering the resolution further to 512 or 224

Video2Text Inference is slow and high vram consumption

Awesome! It should work fine if you just lower the resolution. I have M3 Max with 96GB URAM. I can run this example in under a minute: https://github.com/Blaizzy/mlx-vlm/blob/62bb0ee2f57354de4cd27e42be593049269353a4/examples/video_generation.ipynb

Video2Text Inference is slow and high vram consumption

> Ok, Thanks My pleasure!

Video2Text Inference is slow and high vram consumption

Closing stale

Add Support for OS-Copilot/OS-Atlas-Base-7B

Hey, I just tried it. It works well on demo samples but fails with custom UIs Check the screen resolution they are using and the prompting strategy

Add Support for OS-Copilot/OS-Atlas-Base-7B

Qwen2vl needs to normalise their bbox to 1000