mastodon
4.6.6
Professor at Indiana University Bloomington. Develops network, data, and machine learning methods and applies them to (mostly) social problems.
@yy@twitter is NOT my account.
Charlottesville is getting an urban grocery store. I hope they prioritize ample bike parking instead of spending a lot of money on encouraging driving and increasing traffic. Urban stores can never outpark Costco.
https://yyahn.com/wiki/Bike-parking-at-urban-grocery-stores/
Another incredibly informative video about why speed limit doesn't work (but better road design works) by Justine Underhill (I can't understand how she only has 3k followers... 🤯)
https://www.youtube.com/watch?v=v6LIYQRglnM
https://yyiki.org/wiki/articles/2024/neovim_zenmode/
Neovim + zen mode + twilight for writing
Check out our paper "Network community detection via neural embeddings" led by @sadamori@datasci.social (w Filippo Radicchi and Santo Fortunato)!
https://www.nature.com/articles/s41467-024-52355-w
We argue that n2v can encode community structure down to the detectability limit due to its connection to spectral embedding.
🔬 We're hiring a post-doctoral scholar!
Join us to study knowledge space, scientific funding, and scientific and technological progress! To see the details or to submit CV, research statement & references:
This paper looks like a nice guide for causal inference with instrumental variables. I haven't read whole thing closely yet, but it seems to go through each assumption and what happens if they are violated (not often exposed & explained well).
https://journals.sagepub.com/doi/full/10.1177/00491241241235900
“Typing” effect of LLM interface increases stereo”typing” via stronger anthropomorphism?
Cerebras inference can generate 450 tokens per second for Llama3.1 70B model 😵: https://cerebras.ai/blog/introducing-cerebras-inference-ai-at-instant-speed
How can they do the LLM inference so quickly?
They are making a HUGE chip ("wafer-scale engine" - one chip from one wafer) that contains close to 1M (!) cores specialized to perform sparse linear algebra operations.