Yan: Pairwise evaluation fails for objective metrics like factuality
Eugene Yan · [Paper Club] Who Validates the Validators? Aligning LLM-Judges with Humans (w/ Eugene Yan) · Sep 28, 2024 · at 49:36
Eugene Yan discusses why pairwise comparison models are ill-suited for evaluating objective criteria compared to binary threshold classification.
0:00 / 0:14exact quote · 14.4s
▶ Watch the full episode on YouTube → 720p mp4 · rendered on demand · StarZero watermark
“The reason why pairwise preferences cannot work is that if you give two things that are both factual or if you give two things that are both non-factual, you would say that one is better than the other, but it still doesn't meet the bar of being factual enough.”
quote is from the automated transcript, cleaned for reading: filler sounds and stutters are removed, nothing is rephrased. names can be misheard (the analysis reads context, assessments check outside sources). how →
More from Eugene Yan
Eugene Yan: LLM ranking should use pairwise comparisons for stability
“I'm actually strongly convinced that it should be pairwise and we can debate that and see how it works. And I also think that... I think it's just more reliable and stable that way.”
Eugene Yan: Verifying LLM outputs is often harder than generating them
“And Shreya also has an interesting point, that it's much easier to verify the output and generate it. I actually observe the opposite. Or maybe it depends on the task. Like, for classification tasks, yes, it's easy. For, like, factuality, or comprehensiveness,…”
Eugene Yan: LLM pipeline validation still requires seed human-labeled data
“I'm of a slightly different take. I feel like we do need some set of seed human labeled data.”
Eugene Yan: LLM-as-a-Judge works reliably when reduced to binary classification
“I think when we simplify it to binary classification metrics, I think it can work. And I think a lot of things can be simplified, like Shreya mentioned, I think a lot of things can be simplified to binary classification metrics. And I've seen evidence of it wo…”
Yan: Using LLMs as evaluators is the only way to scale
“I know that we have to use an LLM as an evaluator. There's no way around it. If we want to scale, I think that's the only way.”
Meta used stepwise reward models and Monte Carlo Tree Search for Llama 3.1
“They actually went the extra step to, no pun intended, to actually train stepwise reward models. That's kind of crazy, no? I mean, they wanted each step in the chain of thought to be so good that they actually took the extra effort to train step, to train step…”