Kyle Casterline
@KyleACasterline
Followers
444
Following
887
Media
51
Statuses
425
CTO & Co-founder Barricade AI | professional entropy reducer
San Francisco, Bay Area
Joined July 2013
I don't have too too much to add on top of this earlier post on V3 and I think it applies to R1 too (which is the more recent, thinking equivalent). I will say that Deep Learning has a legendary ravenous appetite for compute, like no other algorithm that has ever been developed
DeepSeek (Chinese AI co) making it look easy today with an open weights release of a frontier-grade LLM trained on a joke of a budget (2048 GPUs for 2 months, $6M). For reference, this level of capability is supposed to require clusters of closer to 16K GPUs, the ones being
371
2K
14K
“Somewhere, something incredible is waiting to be known.” - Carl Sagan
145
3K
18K
Introducing Forkable Chat in LobeChat 🥳- Your Conversations, Unlimited Possibilities! 💡 Start a new chat from ANY message 🔄 Seamlessly switch between related topics 🌳 Build your knowledge tree naturally 💭 Never lose context of your ideas Try it now in LobeChat v1.34.0!
1
3
7
This was one of the most important events I’ve ever organized: the Defense Tech Hackathon, dedicated to Ukrainian challenges - Over 400 hackers collaborated on challenges provided by the Ministry of Defense of Ukraine - 37 teams presented their projects on the final day for
7
9
31
Hacking 👨💻 ✅ Shipping ⚓ ✅✅ Pitching 🎤 ✅ Sleeping 💤 ❌ Incredibly honored to win at the Defense Tech Hackathon for Ukraine 🇺🇦! Building tech that matters was worth every sleepless minute. Thanks for organizing @neffko - this wasn't just a hackathon, it was a chance to
This was one of the most important events I’ve ever organized: the Defense Tech Hackathon, dedicated to Ukrainian challenges - Over 400 hackers collaborated on challenges provided by the Ministry of Defense of Ukraine - 37 teams presented their projects on the final day for
1
0
2
🥇place @ uss defense tech hackathon. Thank you all organizers, mentors and judges! @Andercot @JvNixon @neffko
@newsaddiction Great job dropmesh team! @KyleACasterline @edch03 @osdotsystem @lukehollis @marshallbdoyle
0
5
12
Just got a free scratch n sniff sticker and it smells like freedom 🇺🇸
0
0
1
It’s possible that NotebookLM podcast episode generation is touching on a whole new territory of highly compelling LLM product formats. Feels reminiscent of ChatGPT. Maybe I’m overreacting.
333
395
6K
Let's bring llama.cpp to the clouds! You can now run llama.cpp-powered inference endpoints through Hugging Face with just a few clicks. Simply select a GGUF model, pick your cloud provider (AWS, Azure, GCP), a suitable node GPU/CPU and you are good to go. For more info, check
Wanna see something cool? You can now deploy GGUF models directly onto Hugging Face Inference Endpoints! Powered by llama.cpp @ggerganov Try it now --> https://t.co/fV9QMQhjNx
9
46
308
An insane number of people were just in Singapore for F1, Milken Institute, Token 2049, Break Point, and Balaji’s The Network State. Do not fade Singapore. This is obvious.
21
14
143