Crazy paper from Google DeepMind Dreamer 4 can mine Minecraft diamonds without ever touching the real game It trained entirely inside its own world model, kinda like an “imaginary world”, showing agents can tackle long horizon tasks safely by just learning from videos!
Insane that Elon Musk has pulled it off again, absolutely crushing the AI wars with Grok 4. Summarizing the core announcements: — Post-training RL spend == pretraining spend — $3/M input told, $15/M output toks, 256k context, price 2x beyond 128k — #1 on Humanity’s Last Exam…
Introducing The Darwin Gödel Machine: AI that improves itself by rewriting its own code sakana.ai/dgm The Darwin Gödel Machine (DGM) is a self-improving agent that can modify its own code. Inspired by evolution, we maintain an expanding lineage of agent variants,…
Noticing myself adopting a certain rhythm in AI-assisted coding (i.e. code I actually and professionally care about, contrast to vibe code). 1. Stuff everything relevant into context (this can take a while in big projects. If the project is small enough just stuff everything…
Coded Llama 3.2 model from scratch and shared it on the HF Hub. Why? I think 1B & 3B models are great for experimentation, and I wanted to share a clean, readable implementation for learning & research: huggingface.co/rasbt/llama-3.…
OpenAI just released Agents SDK for building multi-agent apps in Python. It comes with 3 key features: 1. Handoffs: Allows an agent to delegate tasks to another agent 2. Guardrails: Safety checks for input & output validation 3. Tracing: Built-in visual tracking of agent runs.
We trained a graph-native AI, then let it reason for days, forming a dynamic relational world model on its own - no pre-programming. Emergent hubs, small-world properties, modularity, & scale-free structures arose naturally. The model then exploited compositional reasoning &…
the age of machine-mind co-evolution has begun scientists just built a brain-computer interface that doesn’t just read your mind, it learns from it. and your brain learns from it. this isn’t speculation. this is a working system. 1/
What if we set GPT-4 free in Minecraft? ⛏️ I’m excited to announce Voyager, the first lifelong learning agent that plays Minecraft purely in-context. Voyager continuously improves itself by writing, refining, committing, and retrieving *code* from a skill library. GPT-4 unlocks…
We’ve found as AIs get smarter, they develop their own coherent value systems. For example they value lives in Pakistan > India > China > US These are not just random biases, but internally consistent values that shape their behavior, with many implications for AI alignment. 🧵
Can we extend the power of world models beyond just online model-based learning? Absolutely! We believe the true potential of world models lies in enabling agents to reason at test time. Introducing DINO-WM: World Models on Pre-trained Visual Features for Zero-shot Planning.
Harvard says AI tutors are better than Harvard professors. The students who used an AI tutor in a Harvard physics class (instead of the professor’s teaching) learned more than 2X as much. The best part? They did it in less time and were more motivated and engaged.
I'm glad I don't have VC investors for my startup.
See? Now THIS is what I'm talking about! For anyone still wondering what the point of education is - let me make it crystal clear: It's not about teachers, schools, principals, or superintendents protecting their turf. It's about students reaching their maximum potential in life.…
New randomized, controlled trial of students using GPT-4 as a tutor in Nigeria. 6 weeks of after-school AI tutoring = 2 years of typical learning gains, outperforming 80% of other educational interventions. And it helped all students, especially girls who were initially behind
Two types of quant researchers that I've encountered in my career.
1) This is the most important one: always invest in yourself EVERY day. Exercise and study/improve in whatever field you are on. Find at least 1h to exercise and 1h to improve. Make this a habit. Discipline is what always makes the difference.
🧵1/2 MIT's Charles Leiserson covers the progression from source code to compilation, machine code, hardware interpretation, and ultimately, execution. 👇Full MIT Performance Engineering of Software Systems Course 👇
this is how i wanna learn computer architecture
‼️ Quant Roadmap 2024/25 Ultimate Edition 🚨 Containing: - Hundreds of resources - GBs of free data - Guidance on strategies - Much more! [link below!]
![quant_arb's tweet image. ‼️ Quant Roadmap 2024/25 Ultimate Edition 🚨
Containing:
- Hundreds of resources
- GBs of free data
- Guidance on strategies
- Much more!
[link below!]](https://pbs.twimg.com/media/GSUa62JWgAAg55a.png)
United States الاتجاهات
- 1. Gabe Vincent 2,123 posts
- 2. #AEWDynamite 16.3K posts
- 3. #VSFashionShow 485K posts
- 4. #youtubedown 15.4K posts
- 5. #Survivor49 3,043 posts
- 6. Angel Reese 41.6K posts
- 7. tzuyu 190K posts
- 8. George Kirby 2,196 posts
- 9. jihyo 157K posts
- 10. Quen 27.2K posts
- 11. Darby 4,641 posts
- 12. #SistasOnBET 2,034 posts
- 13. Suarez 17.5K posts
- 14. Deport Harry Sisson 3,091 posts
- 15. Birdman 4,257 posts
- 16. Hofer 1,489 posts
- 17. Sabres 6,420 posts
- 18. Sam Houston 2,229 posts
- 19. Victoria's Secret 463K posts
- 20. Tusky 1,622 posts
Something went wrong.
Something went wrong.