Casey Aylward (@caseyaylward) started following @__tensorcore__ on Sep 27, 2024

RT @hyhieu226: Checkout our newest CUDA tutorial. https://t.co/aDDsDNqI5I The topic is software pipelining: overlap mem copying with compu…

RT @cudagdb: Comin with @__tensorcore__ , we ball https://t.co/Y110TGKChM

RT @charles_irl: incredible, incredibly dense talk on @ApacheDataFusio from @andrewlamb1111, ranging from context & overview to deep dive o…

RT @zhyncs42: Recently we did a sharing about SGLang DeepSeek MLA, slides can be found at https://t.co/eew6Ofg0br N/N

RT @sh_reya: LLMs have made exciting progress on hard tasks! But they still struggle to analyze complex, unstructured documents (including…

This is a great blog post from @sh_reya on data flywheels for LLM applications! It has some great thinking from first principles on designing systems that continuously improve over time, including many things I wish I had from the start in production systems I've worked on.

👀

RT @andrey_cheptsov: Kudos to the @LambdaAPI team for putting together this tutorial! Looking forward to seeing even more examples featurin…

You can finetune Llama-3.2 for free on Colab now! Unsloth makes finetuning 2x faster and uses 60% less VRAM with no accuracy degradation. Llama 3.2 (1B) QLoRA fits on a 4GB GPU, and (3B) fits on 7GB. Vision support coming soon. Finetuning Colab: https://t.co/ko1xReM90u

Llama 3.2 versions including GGUF's + bnb 4 bit versions + reuploaded versions are now on @HuggingFace! See all versions of Llama 3.2 here: https://t.co/lI0nNf5Wu0 We are actively working on supporting Vision models and 1B and 3B.

RT @soldni: Olmo goes multimodal! We are launching Molmo, a open family of multimodal models that rival the best closed VLMs out there 🤯…

This. This isn’t a Cursor shill but rather a reflection on how a tightly integrated copilot makes me feel like I can implement anything given my debugging time asymptotically approaching 0 and in turn I can simply just *try* more things and iterate faster

This was a great time!

RT @BenSteiner00: At the World Relays in the spring, a reporter asked USA’s Noah Lyles if he saw Canada as a threat in the 4x100m in Paris.…

RT @AlecStapp: Insane week for permitting reform (really feels like we've reached a tipping point): 1. NEPA exemption for chip fabs passed…

RT @zebulgar: In 2007 @karimatiyeh immigrated to the US from Lebanon to study at Harvard 14 years later, he's the co-founder & CTO of Ramp…

14/ We had so much fun bringing this special community together IRL for an amazing day of technical talks/hacking. Big thanks to everyone who turned out and our planning committee (here’s an “after” photo taken at midnight). Full videos of the talks to follow soon #cudamodeirl. https://t.co/h5htokUT6y

13/ We had 40+ teams submit projects for judging and 10 projects demoed for the entire group. The top five won prizes. The top three were also awarded 4080s signed by Jensen Huang :). https://t.co/0RFFvMf0rp

1/ A few months back, some friends from the GPU Mode (fka CUDA Mode) Discord server and I started talking about bringing the community IRL to plan a first-of-its-kind hackathon 🧵 https://t.co/HuHU7cRCru

RT @cambridgemike: https://t.co/B9katAVBD8

RT @KTmBoyle: Absolutely loving the Article publishing experience on @X. Going to go back and publish evergreen pieces in article format. G…

RT @deno_land: Deno 2 Release Candidate has dropped: ‣ Add `process` global variable, remove `window` ‣ Improve dep management ‣ Permissio…

Safari I despise you with a passion