felixcantcode's profile picture. generating values for stakeholders, internet plumber, http://mu2mi.com, blog: https://felixng.me

opinions are of my own

Felix Nguyen

@felixcantcode

generating values for stakeholders, internet plumber, http://mu2mi.com, blog: https://felixng.me opinions are of my own

Felix Nguyen 님이 재게시함

This is a nice overview. Some more thought: - recursion, the idea, is excellent and underpins lots of algorithms - recursion, implemented as recursive function calls, should be only reserved for lightweight/toy use-cases. In production scenarios you end up with either stack…

react fiber is honestly one of those things that people think they get until they actually look under the hood and then suddenly the whole thing feels like some alien technology living inside javascript. everyone parrots those same shallow lines like “fiber is the new diffing…



Felix Nguyen 님이 재게시함

you can now login to astra via: > claude > openai > gemini link below


Felix Nguyen 님이 재게시함

Bun is joining Anthropic! bun.com/blog/bun-joins…


Felix Nguyen 님이 재게시함

NEW: Lambda FINALLY adds multiconcurrency support with Lambda Managed Instances! Sandboxes process multiple concurrent requests when deployed into EC2 clusters managed by AWS. You pick the instances, and you can set the max concurrency per sandbox. MEANING: No more cold…

astuyve's tweet image. NEW: Lambda FINALLY adds multiconcurrency support with Lambda Managed Instances!

Sandboxes process multiple concurrent requests when deployed into EC2 clusters managed by AWS.

You pick the instances, and you can set the max concurrency per sandbox.

MEANING: No more cold…

Felix Nguyen 님이 재게시함

opus 4.5 is really good at GPU programming, but somehow it’s even better at GPU programming jokes (h/t @Si_Boehm)

jekbradbury's tweet image. opus 4.5 is really good at GPU programming, but somehow it’s even better at GPU programming jokes (h/t @Si_Boehm)

til about floating point fuckery

felixcantcode's tweet image. til about floating point fuckery

Felix Nguyen 님이 재게시함

Baldur's Gate 3 publishing chief Michael Douse says the Steam Machine isn't super high-spec because there's "genuinely no point" when people who want demanding PC performance on TV "know how to turn any PC into a Steam Machine"

Pirat_Nation's tweet image. Baldur's Gate 3 publishing chief Michael Douse says the Steam Machine isn't super high-spec because there's 

"genuinely no point" when people who want demanding PC performance on TV "know how to turn any PC into a Steam Machine"
Pirat_Nation's tweet image. Baldur's Gate 3 publishing chief Michael Douse says the Steam Machine isn't super high-spec because there's 

"genuinely no point" when people who want demanding PC performance on TV "know how to turn any PC into a Steam Machine"

how i learn stuff these days. such a useful feature @OfficialLoganK

felixcantcode's tweet image. how i learn stuff these days. such a useful feature @OfficialLoganK

Felix Nguyen 님이 재게시함

Announcing: New @Steam Hardware, coming in 2026: Steam Controller Steam Machine Steam Frame Watch our jazzy announcement video and wishlist now: steampowered.com/hardware

valvesoftware's tweet image. Announcing: New @Steam Hardware, coming in 2026:

Steam Controller
Steam Machine
Steam Frame

Watch our jazzy announcement video and wishlist now: steampowered.com/hardware

"let me break it down". how about u go fuck yourself instead


Felix Nguyen 님이 재게시함

Shopify teams are using AI to announce their internal tools is one of the most delightful side effects of the AI age. This one is too fun not to share


til soumith worked on dcgan with alec and luke


Felix Nguyen 님이 재게시함

Training LLMs end to end is hard. Very excited to share our new blog (book?) that cover the full pipeline: pre-training, post-training and infra. 200+ pages of what worked, what didn’t, and how to make it run reliably huggingface.co/spaces/Hugging…

eliebakouch's tweet image. Training LLMs end to end is hard. Very excited to share our new blog (book?) that cover the full pipeline: pre-training, post-training and infra. 200+ pages of what worked, what didn’t, and how to make it run reliably

huggingface.co/spaces/Hugging…

Felix Nguyen 님이 재게시함

Please steal my AI research ideas. This is a list of research questions and concrete experiments I would love to see done, but don't have bandwidth to get to. If you are looking to break into AI research (e.g. as an undergraduate, or a software engineer in industry), these are…


Felix Nguyen 님이 재게시함

New blog post: The bug that taught me more about PyTorch than years of using it started with a simple training loss plateau... ended up digging through optimizer states, memory layouts, kernel dispatch, and finally understanding how PyTorch works!

ElanaPearl's tweet image. New blog post: The bug that taught me more about PyTorch than years of using it

started with a simple training loss plateau... ended up digging through optimizer states, memory layouts, kernel dispatch, and finally understanding how PyTorch works!

Felix Nguyen 님이 재게시함

Our latest post explores on-policy distillation, a training approach that unites the error-correcting relevance of RL with the reward density of SFT. When training it for math reasoning and as an internal chat assistant, we find that on-policy distillation can outperform other…

thinkymachines's tweet image. Our latest post explores on-policy distillation, a training approach that unites the error-correcting relevance of RL with the reward density of SFT. When training it for math reasoning and as an internal chat assistant, we find that on-policy distillation can outperform other…

Felix Nguyen 님이 재게시함

one of the often overlooked factors behind deepseek’s success lies in how the team built their distributed infrastructure from the ground up (despite severe gpu constraints!). their custom communication library hfreduce replaced nccl and delivered substantially higher bandwidth…

novasarc01's tweet image. one of the often overlooked factors behind deepseek’s success lies in how the team built their distributed infrastructure from the ground up (despite severe gpu constraints!). their custom communication library hfreduce replaced nccl and delivered substantially higher bandwidth…
novasarc01's tweet image. one of the often overlooked factors behind deepseek’s success lies in how the team built their distributed infrastructure from the ground up (despite severe gpu constraints!). their custom communication library hfreduce replaced nccl and delivered substantially higher bandwidth…
novasarc01's tweet image. one of the often overlooked factors behind deepseek’s success lies in how the team built their distributed infrastructure from the ground up (despite severe gpu constraints!). their custom communication library hfreduce replaced nccl and delivered substantially higher bandwidth…
novasarc01's tweet image. one of the often overlooked factors behind deepseek’s success lies in how the team built their distributed infrastructure from the ground up (despite severe gpu constraints!). their custom communication library hfreduce replaced nccl and delivered substantially higher bandwidth…

United States 트렌드

Loading...

Something went wrong.


Something went wrong.