@adrgrondin @lmstudio hell yeah

Hmmm, interesting...🤔, I think EXO Labs is still the best solution for [2, 3, 4, 5, 6] homogeneous and heterogeneous local/network-isolated clusters via Thunderbolt connections, and LM Link will be useful for remote access to a high-performance cluster (headless server) from any computing machine treated as a terminal. However, I could be wrong. 🤔

@adrgrondin @lmstudio connect multiple Macs? that sounds clumsy. just use one.

@adrgrondin @yagilb @lmstudio Huge, congrats yall!!

@ozzy_ag47 @adrgrondin @lmstudio How do you plan on running K2.6 on 128gb? 🤨

@adrgrondin @lmstudio What a team y’all are.

@adrgrondin @lmstudio This seems like it's a direct alternative to Exo. Will it be able to do both pipeline and TP?

@adrgrondin @lmstudio 4 Mac Studios distributed for inference is honestly insane

who finna afford that rig tho?

@adrgrondin @lmstudio i need it now! 🤣

@adrgrondin @lmstudio Oooh my two M2 Ultra 192gb Mac Studios gonna be liking this. But I have been hooked on oMLX features recently for single machine MLX inference

@adrgrondin @lmstudio Nice 👍

@adrgrondin @lmstudio I have light-years to go to get to something like that!

Until then I built a "studio" for testing (@ @googlegemma ) models on device , to be on a stage demoing something like that would be epic, congrats Adrien! https://github.com/AndrewVoirol/edge-ai-lab

@adrgrondin @lmstudio 🫪

@adrgrondin @lmstudio sweet! do you have the full video link?

@adrgrondin @lmstudio This is awesome

@adrgrondin @lmstudio that setup looks insane for running local models

gonna need a bigger desk though

@adrgrondin @lmstudio Nice!

@adrgrondin @lmstudio incredible work yall

@adrgrondin @lmstudio Congratulations mate 👏

@adrgrondin @lmstudio 楽しみだ Mac studio買い足さなきゃ！！！買えないよ😢