Casey Aylward (@caseyaylward) started following @jfischoff on Sep 28, 2024

You can make this easily with @fal APIs right now. If you’re a dev looking for a weekend project, make an iOS app to make these for families, companies, sports teams etc. Someone is gonna do it, why not you?

Did this in my dream once

What a remarkable letter

RT @repkarinabao: Intel dances alone TSMC dances with 400 partners Morris Chang quotes Jensen Huang to describe TSMC’s edge over Intel’s f…

RT @EpochAIResearch: 1/9 OpenAI's o1-preview outperforms previous models, including GPT-4o, on many reasoning tasks. But how much of this i…

cuda mode? what about triton mode thanks for hosting and for the cool prizes & swag :)

RT @mattdeitke: Introducing Molmo: a family of open state-of-the-art multimodal language models that push the frontier of vision by pointin…

RT @anikembhavi: Delighted to share Molmo! With an extreme focus on data quality over quantity, Molmo is fast to train, competitive with e…

RT @PrimeIntellect: https://t.co/krqsjl3cas

RT @PrimeIntellect: Contribute compute and power the future of decentralized training. Coming soon 🦋 https://t.co/M9g66PzNrF

Thanks for coming!

RT @charles_irl: serverless pgSQL 🤝 serverless GPUs

RT @YangsiboHuang: I'm super excited about the *CL workshop we're planning to organize on LLM memorization, & its implications for complian…

o1's chain of thought contains a lot of verbal expressions like 'Hmm', 'But how?', etc. Are they using lecture recordings to train this model...

RT @PhysInHistory: The difference between information and knowledge :✍️ https://t.co/jP6vwTJFgo

i love everything about this 🔥

The lepton team is incredible! Huge shoutout to @jiayq , JJ and the rest of the team!

Exciting opportunity to hang out with @qingyun_wu and the entire autogen team. Definitely attend this if you are at ICML.

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

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 @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

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.