Artificial Intelligence (AI)

Serious question, how would one go about making an LLM have a very particular world view?

blocked it because of this. No hate on the timeline please! https://t.co/iqsxY4tR3c

you have no idea how hard it is to get an rlhf model to be even “centrist” much less right reactionary. they must have beat this guy up pretty hard

roon

Serious question, how would one go about making an LLM have a very particular world view? 

https://x.com/tszzl/status/1942696841528893920?s=46

@albertwenger thinks you can do this in pre-training

Metadata Conditioning Accelerates Language Model Pre-training

"MeCo first provides metadata (e.g., URLs like https://t.co/ATuAv8hVeA) alongside the text during training and later uses a cooldown phase with only the standard text, thereby enabling the model to function normally https://t.co/KIbnRkE1c4

This is why I believe that pre-training data should also be augmented with explicit moral values metadata

Albert Wenger 🌎🔥⌛

Pragmatism (Security) at OP Labs.
Prev: ConsenSys & Coinbase.

Attribs: Friendly, curious, introverted, mid-witted.
Chans: /eth-security /maurelian /stoicism




Hacker founder using ai to coordinate onchain economies /tenfold, nounish channel client /comint

More: rubinovitz.com

Re Elon a friend who specializes in this thinks it’s their training distribution 

https://x.com/nsaphra/status/1942796411751268565?s=46

He cites this in the tweet 

https://x.com/iscienceluvr/status/1876151047254602191?s=46

@albertwenger thinks you can do this in pre-training 

https://x.com/albertwenger/status/1876396693018538493