hEzo
@hEzohEzo4
Interested in Cognitive Science, Philosophy of Science, Logic and Artificial Intelligence
ID:1447120569854349313
10-10-2021 08:43:33
485 Tweets
120 Followers
946 Following
Follow People
There's a pattern that predicts who does well on IQ tests: people who spend more time on hard problems.
New work w/ Sam Cheyette extends a 40 year old finding about the role of effort in IQ testing, showing why these tests aren't bare measures of 'intelligence.'
CLeaR conference was an absolute blast!
45 posters, 12 oral presentations, 5 keynotes...
1/n
#causality #causalAI #causalinference #machinelearning #conference #causaltwitter CLeaR-Conference on Causal Learning and Reasoning Judea Pearl
🚨Pre-print alert:🚨
Have we built machines that think like people?
In new work, led by Luca Schulze Buschoff and Elif Akata and together with Matthias Bethge, we assess multi-modal #LLMs reasoning abilities in three core domains: intuitive physics, causality, and intuitive psychology.
In case you are wondering, Karthik Valmeekam found that LLM's still can't plan even after the advent of Claude 3 Opus and Gemini Pro..😬
tldr; no AGI doom by all fools day! You can still tame the lot by throwing assorted blocks at 'em..😋
The much ballyhooed Claude 3 Opus does no…
Circuits are a hot topic in interpretability, but how do you find a circuit and guarantee it reflects how your model works?
We (Sandro Pezzelle, Yonatan Belinkov, and I) introduce a new circuit-finding method, EAP-IG, and show it finds more faithful circuits arxiv.org/abs/2403.17806 1/8
This is one of my favorite posts on CrossValidated because it clearly describes what each assumption in linear regression means and what it is necessary and sufficient for. A great antidote to people thinking residuals have to be normally distributed.
stats.stackexchange.com/a/16460/116195