Curious to hear your experience with immich after you spend some time with it. I've found darktable very capable for my needs after some learning curve, but obviously "suitable" depends a lot on the specific task

Depending on how you see it that might have little effect

"Factorio is like crack, just more addictive and less social"

In that case she probably put the emphasis on "think about the implications", rather than the "fire horse"

And a different narrator 👀 (or AI generated, idc)

"Why worry about missing data if we can just make assumptions about their joint distribution with observables?" 💡

2.75

Is it known what they use for the synthetic training data? From my read I'd guess they construct 'random' causal graphs w/ hyperparameters so that the distribution of synthetic datasets implied by those graphs somehow resembles the distn of relationships between variables in some reference datasets

Now this is the real "students these days"

Some claim it's Tokio Hotel

👋

...one particular mark, more like

Great talk!

Is this residualized (in a weird way)? Don't see how NE would be lower than Midwest in absolute terms otherwise

Neat! Any plans (or thoughts on feasibility) to add some RAG/semantic search capabilities to it? Like, "Papers that use two-way fixed effects"?

"Danke" 🧑‍🍳😘👌