AVB
MLX bros and sises - DON’T miss this guy’s next post! Youll be able to do parallel and async requests to mlx vlm server after this gets merged. All your asyncio.gather calls to openai can be routed to localhost and shits gonna just work Continuous batching is a very… https://twitter.com/neural_avb/status/2044914277459656912/photo/1
中文: MLX 兄弟和大西邋關——别错过这家伙的下一篇帖子! 合并后，您将能够对 mlx vlm 服务器进行并行和同步请求。 所有对 openai 的 asyncio.gather 调用都可以路由到 localhost，而 shits 将只起作用 连续批次非常非常...
2026-04-16 23:02:11
AVB
Got to try out this VoxCPM2 model locally. Was trying out some voice cloning with the Pytorch as well as the 4-bit MLX version. For me, this speech model is competing with Qwen3-TTS-1.7B and there are some advantages and disadvantages with both. 1. Qwen is smaller, runs… https://twitter.com/neural_avb/status/2043572819326636361/photo/1
中文: 可以在本地试用这款VoxCPM2型号。 正在尝试使用 Pytorch 以及 4 位 MLX 版本进行语音克隆。 对我来说，这种语音模型与Qwen3-TTS-1.7B竞争，两者都有其优点和缺点。 1。Qwen 更小，运行 . . .
2026-04-13 06:11:42