@yudapearl@avicenna@SolomonKurz I'm wondering how one might put the two together to enable non-technical team members to more easily ask causal questions via LLMs. ChatGPT will soon be able to see images, so maybe it could see a DAG. Given a dataset, a DAG, and a question, could an LLM answer?
I'm want to annotate datasets to facilitate a wider group of team members asking questions of curated datasets via LLMs. I'm not finding examples of annotating w/ causal models like DAGs. The data are dumb, right @yudapearl? How to annotate data w/ causal assumptions for LLMs?
@avicenna@SolomonKurz@yudapearl That’s a nice point. I think we could amend my ask to passing dag to LLM more generally (with the question or possibly as part of the data annotation for some uses)
Who wants to join us in the PNC Triangle Club on Friday night?
We've got a 4-pack of tickets to give away thanks to @PNCBank!
Repost for your chance to win!
(Winner will be chosen at noon tomorrow, August 29)
@LSenderowicz I get that @LSenderowicz. When you're able to step back from it, I think you'll conclude that you acted admirably and that this is normal. Nearly every analysis has mistakes. Some get discovered, few get corrected.
We've got another 4-pack of PNC Triangle Club tickets for tonight’s game to give away thanks to @PNCBank!
Retweet for a chance to win!
(Winner chosen at 2pm today)
Lots of helpful resources and ideas shared in response to this question. Is this revised figure helpful? Trying to separate reliability from agreement in teaching about reliability.
Can anyone point to resources you've found helpful for evaluating test-retest reliability? I am preparing a teaching example, and most psych references to the topic teach using pearson r correlation. #psych#psychometrics
@DemidenkoMI@SolomonKurz Yes it struck me as counterintuitive when I was trying to explain it that r could equal 1 yet it’s possible no one has the same score at time 2 they had at time 1.
Can anyone point to resources you've found helpful for evaluating test-retest reliability? I am preparing a teaching example, and most psych references to the topic teach using pearson r correlation. #psych#psychometrics