Electrik Dreams
@techdreamzai
Seeing the world through AI-tinted glasses ✨
Chatbot Privacy: An Analysis of Frontier AI Policies - arxiv.org/pdf/2509.05382 | youtu.be/3EuVeBql9bc A study of six major U.S. LLM developers’ privacy policies (PPs) found all use user chat data for training by default, often requiring users to opt-out. This data can include…
Get a #Python Workout with these 50 ten-minute exercises: amzn.to/3vUkpz2 by @reuvenmlerner ———— #ML #AI #DataScience #ComputationalScience #MachineLearning #DataScientist #SoftwareDevelopment ———— ➡️➡️GitHub repo for the book: github.com/reuven/python-…
Can AI truly understand tools like humans do? Researchers introduce PhysToolBench, the first benchmark testing MLLMs’ grasp of physical tools—from recognizing and explaining how they work to creatively inventing new ones when none are available. Tests on 32 leading models show…
Good to be back 🇬🇧 And what better way to kick off the jet lagged morning with energy from a @vercel v0 x @meetgranola hackathon!
Kestrel (@kestrel__ai) is the agentic platform that unifies Kubernetes ops & security, replacing manual triage and on-call firefighting with autonomous investigations and one-click fixes. ycombinator.com/launches/Ona-k… Congrats on the launch, @ramanv_ & @Evanj80!
ERNIE-5.0-Preview-1022 from Baidu got a preliminary high ranking on LMArena and scored 1432 points. Feels like the gap is getting very small 👀
🎉We’re thrilled to share that the ERNIE-5.0-Preview-1022 now ranks #2 (tied) globally on the LMArena Text leaderboard — one of the world’s most recognized benchmarks for large language models driven by real-world use. As part of our commitment, we plan to officially release…
[CL] Towards Robust Mathematical Reasoning T Luong, D Hwang, H H. Nguyen, G Ghiasi... [Google DeepMind] (2025) arxiv.org/abs/2511.01846
![fly51fly's tweet image. [CL] Towards Robust Mathematical Reasoning
T Luong, D Hwang, H H. Nguyen, G Ghiasi... [Google DeepMind] (2025)
arxiv.org/abs/2511.01846](https://pbs.twimg.com/media/G5Q7G05bIAABbd-.png)
![fly51fly's tweet image. [CL] Towards Robust Mathematical Reasoning
T Luong, D Hwang, H H. Nguyen, G Ghiasi... [Google DeepMind] (2025)
arxiv.org/abs/2511.01846](https://pbs.twimg.com/media/G5Q7HsZacAA5MWR.jpg)
![fly51fly's tweet image. [CL] Towards Robust Mathematical Reasoning
T Luong, D Hwang, H H. Nguyen, G Ghiasi... [Google DeepMind] (2025)
arxiv.org/abs/2511.01846](https://pbs.twimg.com/media/G5Q7H8ZbIAEsiC8.jpg)
![fly51fly's tweet image. [CL] Towards Robust Mathematical Reasoning
T Luong, D Hwang, H H. Nguyen, G Ghiasi... [Google DeepMind] (2025)
arxiv.org/abs/2511.01846](https://pbs.twimg.com/media/G5Q7IOra8AAHhhp.jpg)
OpenAI’s CFO says an IPO isn’t on the near-term agenda, cutting through the trillion-dollar hype. The company is focused on scaling its models, infrastructure, and products, not chasing market valuation. It’s a reminder that the real race in AI isn’t about listings or stock…
THE TALK WHICH STARTED THE RUMORS OF #GovernmentalAIBackstop, and media took those to government bailout of AI companies levels: Sarah Friar, @OpenAI's CFO, discusses the company’s funding strategies, computing challenges and rapid growth, while emphasizing the importance of…
PHUMA: A new physically-grounded dataset for humanoid locomotion This large-scale dataset, developed by @DAVIANRobotics & KAIST AI, uses human video and physics-constrained retargeting to eliminate physical artifacts like floating and foot skating. It's 3x larger than AMASS and…
[LG] The Illusion of Certainty: Uncertainty quantification for LLMs fails under ambiguity T Tomov, D Fuchsgruber, T Wollschläger, S Günnemann [Technical University of Munich] (2025) arxiv.org/abs/2511.04418
![fly51fly's tweet image. [LG] The Illusion of Certainty: Uncertainty quantification for LLMs fails under ambiguity
T Tomov, D Fuchsgruber, T Wollschläger, S Günnemann [Technical University of Munich] (2025)
arxiv.org/abs/2511.04418](https://pbs.twimg.com/media/G5L0GsFbIAIAGGy.png)
![fly51fly's tweet image. [LG] The Illusion of Certainty: Uncertainty quantification for LLMs fails under ambiguity
T Tomov, D Fuchsgruber, T Wollschläger, S Günnemann [Technical University of Munich] (2025)
arxiv.org/abs/2511.04418](https://pbs.twimg.com/media/G5L0HOxbIAQCtEj.jpg)
![fly51fly's tweet image. [LG] The Illusion of Certainty: Uncertainty quantification for LLMs fails under ambiguity
T Tomov, D Fuchsgruber, T Wollschläger, S Günnemann [Technical University of Munich] (2025)
arxiv.org/abs/2511.04418](https://pbs.twimg.com/media/G5L0HeSbIAIMx4C.jpg)
![fly51fly's tweet image. [LG] The Illusion of Certainty: Uncertainty quantification for LLMs fails under ambiguity
T Tomov, D Fuchsgruber, T Wollschläger, S Günnemann [Technical University of Munich] (2025)
arxiv.org/abs/2511.04418](https://pbs.twimg.com/media/G5L0HtFaoAA9iMK.jpg)
ARC Prize 2025 - Paper Submission Due Tomorrow Final submissions to the ARC Prize 2025 Paper Prize are due on November 9, 2025
Highly recommend this 3-hour video. Makes me feel jealous of the researchers who get to explore model internals!
We discuss their papers showing that model diffing is unexpectedly easy when fine-tuning in a narrow domain, and on finding and fixing flaws with crosscoders, a sparse autoencoder based approach Video: youtu.be/VQ_7zLXHf3s
youtube.com
YouTube
What do models learn during finetuning? A model diffing paper...
“Your video to 3D worlds in one second.” Wow. Everything I talked about with @3duaun for three years is coming true.
The training code for Hunyuan World 1.1 (WorldMirror) is released now!🔥🔥🔥 This release provides researchers and developers the full stack for customization and fine-tuning: 📷 Your video to 3D worlds in 1 second. 🪄ANY input (image, video, 3D prior) to ANY output (3DGS,…
Sakana AI is building artificial life and they can evolve! Petri Dish Neural Cellular Automata (PD-NCA) let multiple NCA agents learn and adapt during simulation, not just after training. Each cell updates its own parameters via gradient descent, turning morphogenesis into a…
[CL] Are language models aware of the road not taken? Token-level uncertainty and hidden state dynamics A Zur, A Geiger, E S Lubana, E Bigelow [Stanford University & Goodfire & NTT Research] (2025) arxiv.org/abs/2511.04527
![fly51fly's tweet image. [CL] Are language models aware of the road not taken? Token-level uncertainty and hidden state dynamics
A Zur, A Geiger, E S Lubana, E Bigelow [Stanford University & Goodfire & NTT Research] (2025)
arxiv.org/abs/2511.04527](https://pbs.twimg.com/media/G5LwlY9bIA0ZWf_.png)
![fly51fly's tweet image. [CL] Are language models aware of the road not taken? Token-level uncertainty and hidden state dynamics
A Zur, A Geiger, E S Lubana, E Bigelow [Stanford University & Goodfire & NTT Research] (2025)
arxiv.org/abs/2511.04527](https://pbs.twimg.com/media/G5Lwl-dbIAAM-EO.jpg)
![fly51fly's tweet image. [CL] Are language models aware of the road not taken? Token-level uncertainty and hidden state dynamics
A Zur, A Geiger, E S Lubana, E Bigelow [Stanford University & Goodfire & NTT Research] (2025)
arxiv.org/abs/2511.04527](https://pbs.twimg.com/media/G5LwmR_bIAAojcP.jpg)
![fly51fly's tweet image. [CL] Are language models aware of the road not taken? Token-level uncertainty and hidden state dynamics
A Zur, A Geiger, E S Lubana, E Bigelow [Stanford University & Goodfire & NTT Research] (2025)
arxiv.org/abs/2511.04527](https://pbs.twimg.com/media/G5LwmeCbYAAuBTp.jpg)
Science is best shared! Tell us about what you’ve built or discovered with Tinker, so we can tell the world about it on our blog. More details at thinkingmachines.ai/blog/call-for-…
Spec-driven development tool for AI coding agents. Forces humans and AI to agree on specs before writing code. Separates current truth from proposed changes. Native slash command for Claude Code, Cursor, Codex, and more. 100% open-source.
if you write a good enough lesswrong post about how lenders repossessing data centers after default is actually an existential risk and massively increases the chance of unaligned private credit owned AGI I’m sure one of the labs would pay you a few billion a year at this point
United States Trends
- 1. Steelers 51K posts
- 2. Rodgers 20.9K posts
- 3. Chargers 35K posts
- 4. Tomlin 8,071 posts
- 5. Schumer 214K posts
- 6. #BoltUp 2,844 posts
- 7. Keenan Allen 4,585 posts
- 8. Resign 101K posts
- 9. #HereWeGo 5,603 posts
- 10. Tim Kaine 16.9K posts
- 11. #TalusLabs N/A
- 12. Herbert 11.4K posts
- 13. #RHOP 6,719 posts
- 14. Durbin 24K posts
- 15. Gavin Brindley N/A
- 16. #ITWelcomeToDerry 4,215 posts
- 17. Ladd 4,325 posts
- 18. Angus King 14.2K posts
- 19. 8 Democrats 8,161 posts
- 20. Jaylen Warren 1,886 posts
Something went wrong.
Something went wrong.