FeralRobots on Nostr: "You can scale up your LLM and feed it even more data — but the result will be a ...
"You can scale up your LLM and feed it even more data — but the result will be a better pattern-matcher and not a better reasoner."
That is: it gets better at producing average results, which is the thing for which you absolutely need it least.
#genAI #AI #LLM
https://pivot-to-ai.com/2024/10/13/llms-cant-reason-they-just-crib-reasoning-like-steps-from-their-training-data/
That is: it gets better at producing average results, which is the thing for which you absolutely need it least.
#genAI #AI #LLM
https://pivot-to-ai.com/2024/10/13/llms-cant-reason-they-just-crib-reasoning-like-steps-from-their-training-data/