Comment on It Only Takes A Handful Of Samples To Poison Any Size LLM, Anthropic Finds
hoppolito@mander.xyz 2 days agoAs far as I know that’s generally what is often done, but it’s a surprisingly hard problem to solve ‘completely’ for two reasons:
The more obvious one - how do you define quality? When you’re working with the amount of data LLMs require as input and need to be checked for on output you’re going to have to automate these quality checks, and in one way or another it comes back around to some system having to define and judge against this score.
There’s many different benchmarks out there nowadays, but it’s still virtually impossible to just have ‘a’ quality score for such a complex task.
Perhaps the less obvious one - you generally don’t want to ‘overfit’ your model to whatever quality scoring system you set up. If you get too close to it, your model typically won’t be generally useful anymore, rather just always outputting things which exactly satisfy the scoring principle, nothing else.
If it reaches a theoretical perfect score, it would just end up being a replication of the quality score itself.
Good points. What’s novel information vs. wrong information? (And subtly wrong is harder to understand than very wrong)
At some point it’s hitting a user who is giving feedback, but I imagine data lineage once it gets to the end user its tricky to understand.
WhiteOakBayou@lemmy.world 2 days ago
like the LLM that was finding cancers and people were initially impressed but then they figured out the LLM had just correlated a DR’s name on the scan to a high likelihood of cancer. Once the complicating data point was removed, the LLM no longer performed impressively. Point #2 is very Goodhart’s law adjacent.
bitjunkie@lemmy.world 1 day ago
I never knew the name for this law, but it’s basically how SEO ruined traditional search. I think it’s also a big reason that a LOT of software engineers put way too much emphasis on passing unit tests and not nearly enough on examining what they’re actually testing.
phutatorius@lemmy.zip 3 hours ago
It’s a special case of the buiness-school dictum that a metric that is made into a performance measure immediately becomes useless, since there are now incentives to game it.