pull down to refresh

In other words, meek models (those with limited computation budget) shall inherit the earth, approaching the performance level of the best models overall.
I actually have a similar take. The diminishing returns combined with retail hardware advances are going to make it that in 2030 your average home computer will run a model that more or less performs similar to todays frontier models.
We've seen this same thing constantly. There was a time when an iPhone 4 could do a tremendous amount more than an iPhone 1....but that hasn't been true for a long time. Does the average phone user really need anything more advanced than say a iPhone 9? Text, Photos, Emails, Maps, etc....basically its all become a "solved problem".
I think the same is going to happen in AI. At a certain point, the day-to-day use cases are going to be solved by open source models running on commodity hardware.
I don't think we'll have to wait that long.
NVIDIA DGX Spark Arrives for World’s AI Developers #1256239
reply
33 sats \ 2 replies \ @optimism 5h
I want one but its sold out. Oh and its probably expensive haha
reply
33 sats \ 0 replies \ @freetx 4h
I'm waiting for more comparisons between AMD AI 395 and the DGX.
Main benefits of DGX is that NVidia tooling and ecosystem is so much better....however raw CPU of AMD is probably faster.
Preliminary testing on https://www.reddit.com/r/LocalLLaMA/comments/1o6izz2/dgx_spark_vs_ai_max_395/ seems to indicate that AMD AI 395 wins (also ~33% cheaper)
reply