very probable: AI manipulating democracies
for some weird(?) reason meta's LLMs are ranking higher on my leaderboard
my wallet is disconnected now 😞
what is the answer?
or maybe countries will finally realize BTC is uncontrollable and switch to BTC after stablecoin pains. maybe trojan horses are made in "stables" (pun intended).
knight h6
For ChatGPT, Grok, and Claude the censorship happens at multiple levels.
1. Censorship at the model training level
2. Censorship at the input box
3. Censorship in hidden system prompt
nostr:nprofile1qyjhwumn8ghj7en9v4j8xtnwdaehgu3wvfskuep0dakku62ltamx2mn5w4ex2ucpxpmhxue69uhkjarrdpuj6em0d3jx2mnjdajz6en4wf3k7mn5dphhq6rpva6hxtnnvdshyctz9e5k6tcqyp7u8zl8y8yfa87nstgj2405t2shal4rez0fzvxgrseq7k60gsrx6zeuh5t only has number 1 because we use models trained by others. Over time we are providing a variety of open models so people can choose the training bias that works best for them.
We do not do numbers 2 and 3.
I checked some of the fine tunings that are claiming censorship removal such as Nous Hermes and Dolphin.. They are usually worse in the human alignment compared to the base model. So there is a trade of there.
You could try ablations of base models, by the method found by Maxime Lebonne.
Mine are very human aligned but not caring about censorship at this point. I could do that in the future maybe it could be a selling point. But I do use lots of nostr notes, so my models are somewhat more brave which could be described as having balls, having spine or being based.
What kind of fine tunings would you like for maple AI?
thomas seyfried, thomas cowan
I used the word syntropy as an antonym to entropy. Getting lower in entropy means getting higher in syntropy. The tendency to struggle to beat chaotic ego can be the tendency to unite with Infinite Orderliness which can be described as syntropy.
Yes I built it. More info here
how we fix the AI thingy
yep. label each generation as something different and then grammas can't teach babies their belief systems because they are different (!).
nice episode! my work may be relevant to y'all.
faith is actually a column in my leadearboard :) i think when an LLM "is faithful" it is more human aligned.
https://huggingface.co/blog/etemiz/aha-leaderboard
lets build a beneficial AI https://huggingface.co/blog/etemiz/building-a-beneficial-ai
I need to make sure you have admirable life before taking this tip.
yes. nostr is a good start to find the best knowledge. been fine tuning LLMs based on my own curations and nostr, and making it available to users as nostr:nprofile1qyg8wumn8ghj7efwdehhxtnvdakz7qgwwaehxw309ahx7uewd3hkctcqyrzl4h44myxk3whlccc52ks8eg6qenc7xygsj40pd4z7khu8z37djfwamdn . we should bring more people into curation process. ping me if u want to get involved.
