Varun Mayya
RT @nearcyan: Making Deep Learning Go Brrrr From First Principles (2022)

Improving the speed of your training runs looks very different depending on if your performance is bottlenecked by compute, bandwidth, or overhead

blogpost: https://t.co/3BWYMRvE6R https://t.co/MGnzKWrRP9
tweet

🅹🅰︎🅱🅰︎🅻🅰︎🅽🅳.🅺🅴
Kila siku at 4:20 nitakuwa naachilia thread moja wazimu mnipee engagements
tweet

🅹🅰︎🅱🅰︎🅻🅰︎🅽🅳.🅺🅴
Chilling with my neighbors pussy https://t.co/iKldTEar6f
tweet

murat 🍥
one wild change in the recent OpenAI models is that temperature=0 is the optimal setting for a lot of use cases. especially gpt4

no token leads to mode collapse
tweet

🅹🅰︎🅱🅰︎🅻🅰︎🅽🅳.🅺🅴
RT @Brandboy_: Who is better?

Dax or NF? https://t.co/tHUZftfMCB
tweet

🅹🅰︎🅱🅰︎🅻🅰︎🅽🅳.🅺🅴
RT @MrBlackOG: Only smart people get this joke https://t.co/Y78JsJKO9v
tweet

Paul Graham
RT @scienceisstrat1: Cost of sequencing a human genome:

• 2001 - $95 million
• 2021 - $450
• 2023 - $200 with new Illumina machine

Cc: @fdesouza @Noahpinion @erikbryn @amcafee @rtnarch https://t.co/DxZEJzoSbR
tweet

Robin Hanson
"training [to] enable women to better convey their knowledge & preferences to their husbands … program increases spousal discussion about the family’s health, nutrition, & finances … does not increase overall adoption of health-promoting behaviors or https://t.co/QI13HtqfTk… https://t.co/7VN1sQunnF
tweet

Paul Graham
RT @JMannhart: This is an absolutely incredible video.

Hinton: “That's an issue, right. We have to think hard about how to control that.“

Reporter: “Can we?“

Hinton: “We don't know. We haven't been there yet. But we can try!“

Reporter: “That seems kind of concerning?“

Hinton: “Uh, yes!“ https://t.co/S6GJk3eWXW
tweet

proxima centauri b
RT @deepfates: gm https://t.co/SJzHzZpBKG
tweet