Peter Bui
@peterbuiCS
CS @LifeAtPurdue AI Engineer, Entrepreneur. Aspiring to make Jarvis a real thing!
After months of work, I’ve open-sourced Nova Voice — a fully local, real-time speech-to-text and translation system built on Faster-Whisper. This is my first non-toy project — it’s a full distributed pipeline built for scalability and low-latency: • Real-time STT & translation…
Do you think our world is going to be like the movie "Her"? Obviously the current LLM is not going to help. But, could it?
Worth-reading
<Rant> I spent 25 years in the defense industry (with 8+ in uniform, 2+ in war zones). I have no love for the CCP, but no matter how I view China's govt, their AI research company's are doing a lot of good and deserve some credit. To anyone that thinks Deepseek is some kind of…
Be ready for image-input-only model.
BOOOOOOOM! CHINA DEEPSEEK DOES IT AGAIN! An entire encyclopedia compressed into a single, high-resolution image! — A mind-blowing breakthrough. DeepSeek-OCR, unleashed an electrifying 3-billion-parameter vision-language model that obliterates the boundaries between text and…
Instagram is too stupid that it rots my brain. X is too smart that it hurts my brain. Probably just go to sleep instead ;)
The way I understand it is that they turn a text document into a pixel and eventually multiple documents into an image, thus reducing token input size. The LLM now "sees" the text instead of actually reading it. I don't even know what to think rn :)
🚨 DeepSeek just did something wild. They built an OCR system that compresses long text into vision tokens literally turning paragraphs into pixels. Their model, DeepSeek-OCR, achieves 97% decoding precision at 10× compression and still manages 60% accuracy even at 20×. That…
This might sound so corny, but after months of depression, happiness comes from finally being able to roll your app out to production.
the recursive idea is exactly what I proposed in Recursive Omni, guess I need more execution to provide valuable results and metrics to my ideas.
Meta-agent framework for building high-performance multi-agent systems! ROMA is an open-source meta-agent framework for building agents with hierarchical task execution. It adopts a recursive hierarchical architecture where tasks are decomposed into subtasks, agents handle the…
Holy shit...Google just built an AI that learns from its own mistakes in real time. New paper dropped on ReasoningBank. The idea is pretty simple but nobody's done it this way before. Instead of just saving chat history or raw logs, it pulls out the actual reasoning patterns,…
Computer-Use agent is the future and this repo is proving it. #1 Github trending
just woke up to find we're #1 trending on GitHub! what a wild ride - and awesome to see 3 YC teams up there together!
Basic context-enginerring when vibe-coding: "Read relevant files first before implementing this request: [Your request]".
United States Trends
- 1. Northern Lights 28.3K posts
- 2. #DWTS 47.4K posts
- 3. #Aurora 5,849 posts
- 4. Justin Edwards 1,814 posts
- 5. Louisville 15.9K posts
- 6. Andy 60.1K posts
- 7. #RHOSLC 5,713 posts
- 8. Lowe 12.2K posts
- 9. #OlandriaxHarpersBazaar 3,375 posts
- 10. Elaine 42.6K posts
- 11. Kentucky 24.7K posts
- 12. Celtics 11.9K posts
- 13. Oweh 1,828 posts
- 14. JT Toppin N/A
- 15. Robert 99.4K posts
- 16. #WWENXT 15.9K posts
- 17. Dylan 30.8K posts
- 18. Whitney 8,750 posts
- 19. Jordan Walsh N/A
- 20. Pope 26.4K posts
Something went wrong.
Something went wrong.