Posts tagged: curves
I just pushed a new paper to arXiv. I realized that a lot of my previous work on robust losses and nerf-y things was dancing around something simpler: a slight tweak to the classic Box-Cox power transform that makes it much more useful and stable. It's this f(x, λ) here: pic.twitter.com/GZsA8BYbEe
— Jon Barron (@jon_barron) February 18, 2025
contributed by Andy on Feb 18, 2025 in curves, future fields medalists
Adam, a 9-yr old optimizer, is the go-to for training LLMs (eg, GPT-3, OPT, LLAMA).
— Tengyu Ma (@tengyuma) May 24, 2023
Introducing Sophia, a new optimizer that is 2x faster than Adam on LLMs. Just a few more lines of code could cut your costs from $2M to $1M (if scaling laws hold).https://t.co/GrMY600lLO 🧵⬇️ pic.twitter.com/bPLCOWcIHZ
contributed by Andy on May 24, 2023 in curves, small multiples
Another example: Take two glasses of equally sweet water and mix them together. Children of intermediate age think that the resulting solution will be sweeter than the originals! Similar results have been found for water temperature. pic.twitter.com/vf1o2ihwMj
— Chris Said (@Chris_Said) April 21, 2023
contributed by Andy on Apr 21, 2023 in curves
Fintechs tends to have worse delinquency rates than banks in personal loans. Source: @JPMorganAM pic.twitter.com/ROslktYnib
— Snippet Finance (@SnippetFinance) March 7, 2022
contributed by Andy on Apr 18, 2022 in curves
Isn't Japan somewhat unique since a significant amount of land (major cities esp) is "leasehold," held by the same families for generations. Lessors (users) build for short-term use, maybe a generation or 20 years before expected teardown, and
— Rob Frances (@RFrances2) April 7, 2022
Japan's population is shrinking: pic.twitter.com/2ApBe5OOfs
contributed by Andy on Apr 7, 2022 in curves