This might very well be my last tweet.
Consider this account dead, it’ll be deactivated soon.
Just thought of giving the few real people who follow me here this notice in advance.
Bye!
How can we adapt and finetune large language models from an efficiency standpoint?
Yesterday, I discussed the recent LLaMA-Adapters. Many of you were curious about how this approach compares to the alternatives (eg low-rank adaption finetuning, prefix finetuning & others).
1/5
1. Get LLaMA
2. Future AI systems that are factual (do not hallucinate), can use tools, have physical intuition, can reason and plan, will have a very different architecture from the current crop of Auto-Regressive LLMs.
Find out what it is.
Breaking News: The Pentagon is blocking the U.S. from sharing evidence on Russian atrocities in Ukraine with the International Criminal Court, officials said. Military leaders fear setting a precedent that might pave the way for it to prosecute Americans. https://t.co/xnHCjkkZnK
Some people on my TL are imagining that we can slow or "press pause" on AI... y'know, for safety reasons. Let's be clear on what that would mean: https://t.co/twCFhpg6m0
@_cun__ @alexiajokanovic Nixon Shock: Transition to pure credit money with no anchor to the physical world. Absolutely no constraints on money creation and credit expansion (mostly to those more creditworthy). And ultra financialization of the economy and all assets.
I'm speechless.
Not peer-reviewed yet but a submitted paper.
The 'presented images' were shown to a group of humans. The 'reconstructed images' were the result of an fMRI output to Stable Diffusion.
In other words, #stablediffusion literally read people's minds.
Source 👇