What's/'ll be the solution around the faulty nonfunctional VGA output on the second server?
That is amazing lol. Artisan AI indeed. I look forward to the base model you make along with new Dolphins. So cool π keep up the good work.
Nice work on your AI servers Eric the configuration sounds great. The original dolphin-13b was a flop! I disagree with you. No I don't think it was a flop. ChatGTP 4.0 is using the Multiple Agent Model in order to to get high performance.
ts2.space/en/the-use-of-multi-agent-systems-for-aβ¦
Your model is running on a completely different architecture. In my opinion you did a nice job.
Jay
Amit Caleechurn
How have the MI100s fared compared to the 4090s? I'm about to build a node and wondering if the ROCM support is mature enough and performs well in comparison to well supported CUDA builds? Are you satisfied with the Gigabyte server after running it for a while, in terms of performance and thermal? You are one of the very few people who have blogged or tried the MI100s, so anything additional you are willing to share is most welcome.