Re:Stanford Alpaca: An Instruction-following LLaMA model
by mlyle on Monday March 13, 2023 @02:59PM
(#35140713)
On the other hand, LLM's without some degree of safety work will frequently advocate for genocide/eugenics; perpetuate terrible stereotypes; mock ethnic dialects of language; veer randomly into erotica; etc. Not things in "trigger warning" territory but stuff that is broadly considered unacceptable.
[ Parent
| on HN ]
Re:Stanford Alpaca: An Instruction-following LLaMA model
by Traubenfuchs on Monday March 13, 2023 @03:08PM
(#35140844)
In private, you can ask it not to produce this kind of content.
If you offer a public API it‘s your responsibility to restrain the LLM or do an automated acceptability analysis before publishing content.
But the raw, open source code should not be constrained, castrated and sterilized.
Re:Stanford Alpaca: An Instruction-following LLaMA model
by mlyle on Monday March 13, 2023 @03:39PM
(#35141290)
> But the raw, open source code should not be constrained, castrated and sterilized.
Which is what we have now. But they are going to fine-tune it so that we can use it for various purposes without worrying too much it will go on a rant about "the blacks" again, which makes it a lot more useful for many use cases.
Re:Stanford Alpaca: An Instruction-following LLaMA model
by int_19h on Monday March 13, 2023 @11:53PM
(#35147242)
On the gripping hand, LLMs that were forced into being "safe" just end up regurgitating biases to that effect. One particular example that I ran into: https://news.ycombinator.com/item?id=34987074