I chose last years Mac Studio M4 Max 128gb for my inference needs mainly because it was more power efficient than its competitors. I guess I lucked out with the mlx models making them more appealing to run everything local. What’s my best bet on models on 128gb?
I finally have a self hosted inference server I am proud to share setup on. I still love efficiency overlapped with performance over just raw performance though. Mac Studio setup migrated from ollama/lm studio over to omlx and its impressive benchmark differences.
@theo It helps to work backwards from your needs. For devs we’re asking a lot for a smaller model to be able to accomplish this. Whatcha working with? Framework? Mac? I can help with Mac and omlx.
It’s been a long time since I’ve given my open source project(s) some attention. I had time to update prisma-erd-generator to support @prisma v6 as well as some much needed updates for the developers who offer up time to help out.
Changes: https://t.co/pIWdn4xtsT
@EmmaBostian@michael_webhead was a great fit for me. I had a good idea where I wanted to progress and he always made things happen for me. Good challenges and easy conversations.
@leeerob Love for next.js and prisma together. Take generators and auto create apis in a spec and you’re prototyping in no time. Huge head start for my team. https://t.co/JWIQ9sxTwq
@prisma Love this question! I’m happy getting to build with software. It’s been a lot of fun. That being said, I am really attached to hiking and being outdoors. So many choices there it’s hard to pick. I do think software will give me that future flexibility to spend more time outdoors
@stories_of_ren I’ve been liking the M1 MacBook Air a lot. Most of the apps I use are ported over. There was that 5 month or so period of no difference though. Good time to get one!