somerandomperson@lemmy.dbzer0.com to Privacy@lemmy.ml · 2 个月前Should i trust proton?message-squaremessage-square70fedilinkarrow-up168arrow-down17file-text
arrow-up161arrow-down1message-squareShould i trust proton?somerandomperson@lemmy.dbzer0.com to Privacy@lemmy.ml · 2 个月前message-square70fedilinkfile-text
minus-squaresunzu2@thebrainbin.orglinkfedilinkarrow-up2·2 个月前Well then please do educate the class on how it works
minus-squareSteve@communick.newslinkfedilinkEnglisharrow-up1·2 个月前They do text prediction based on the training data. If the training data is all encrypted gibberish, it’ll only output gibberish.
minus-squaresunzu2@thebrainbin.orglinkfedilinkarrow-up2·2 个月前I assumed you would need to let LLM to access your data for it to be any market advantage v generic llm. If this is just a generic llm that doesn’t have access to your data them my point above is not an issue.
That’s not how LLMs work.
Well then please do educate the class on how it works
They do text prediction based on the training data. If the training data is all encrypted gibberish, it’ll only output gibberish.
I assumed you would need to let LLM to access your data for it to be any market advantage v generic llm.
If this is just a generic llm that doesn’t have access to your data them my point above is not an issue.