Similar User
@misskecupbung
@tantaman
@mattrickard
@benbjohnson
@TitzerBL
@reneeshah123
@eatonphil
@RobustaDev
@KateGoldenring
@dxhuang
@MarkCallaghanDB
@brooksmtownsend
@BadCodeOnly
@Avinesh__T
@mattomata
that “what did you do this week” email responses are gonna be awkward when it’s a collective company expertly nuked by @sama
7 years and counting, and still on my top 7 books to learn System Design. 👇
please please PLEASE post the post-mortem @NetflixEng I have no doubt it will be so comprehensive that therapists can use it for continuing education credits.
Coding interviews should be using ai. If you disagree you literally don’t understand what ai is for Sure you can use it to fill gaps & find easy answers More importantly you can use it to accelerate and do much bigger things Failing to select for this skill will ruin tech co’s
yep people seem to get confused and think subsequent pre-training is purely additive. GPT-4o isn’t merely 3.5 with _more_ data and parameters.
the first alien species to crack how to make llama’s weights go brrr is in for a trip. Especially if they somehow didn’t unlock that part of the tech tree yet. it’s like the theory of leaking scientific advancements to your adversary to nerd snipe at nation state scales.
lol at all of these ai consulting practices that slowly turns into build your own consulting business consulting, like everything else under the sun.
I’m starting to dial in the magic of scaling inference time with o1, how to get it to think for longer by guiding how it should spend its reasoning time. Vibes check out.
a tool use SLM that is as tiny but accurate as Silero VAD is for VAD tasks would be world changing. I just can’t get SmolLM2 135m to behave - tool use seems to really start becoming minimally viable at that high 1B parameter mark.
constantly surprised by people citing evals without having read anything in the dataset then wonder why their out of set task performs wildly different. it’s like you’ve read that an amnesiac toddler answers 80% of soup questions while asking how they liked their enchiladas.
🎨 Introducing Prompt Canvas — a novel UX for prompt engineering Building LLM applications requires new and dedicated tools for prompt engineering. With Prompt Canvas in LangSmith, you can: • Collaborate with an AI agent to draft, refine, and edit your prompts • Define custom…
United States Trends
- 1. Brian Kelly 5.918 posts
- 2. Gators 9.346 posts
- 3. Louisville 4.938 posts
- 4. Feds 33,1 B posts
- 5. Nuss 3.029 posts
- 6. #UFC309 35,6 B posts
- 7. Stanford 8.173 posts
- 8. Billy Napier 1.528 posts
- 9. Nebraska 8.982 posts
- 10. Mizzou 3.918 posts
- 11. Lagway 3.995 posts
- 12. Brohm N/A
- 13. #Huskers 1.346 posts
- 14. Tyler Warren 1.948 posts
- 15. Raiola N/A
- 16. Ron English N/A
- 17. #MostRequestedLive 4.582 posts
- 18. Baylor 3.282 posts
- 19. Heisman 13,2 B posts
- 20. Moura 6.773 posts
Who to follow
-
Ananda Dwi Ae
@misskecupbung -
Matt Wonlaw
@tantaman -
Matt Rickard
@mattrickard -
Ben Johnson
@benbjohnson -
Ben L. Titzer
@TitzerBL -
Renee Shah
@reneeshah123 -
Phil Eaton
@eatonphil -
Robusta.dev
@RobustaDev -
Kate Goldenring
@KateGoldenring -
Ed Huang
@dxhuang -
Mark Callaghan
@MarkCallaghanDB -
Brooks Townsend 🦀
@brooksmtownsend -
Jeff McCoy
@BadCodeOnly -
Avinesh Tripathi
@Avinesh__T -
Matt Moore ⛓🚀
@mattomata
Something went wrong.
Something went wrong.