diwank

6 ms·

31.
Effective harnesses for long-running agents(anthropic.com)
125 points by diwank 11mo ago|37 comments
32.
by diwank 1y ago
Just in time UI is incredibly promising direction. I don't expect (in the near term) that entire apps would do this but many small parts of them would really benefit. For instance, website/app tours could be just generated atop th
33.
Cursor 2.0 – A step in the wrong direction(forum.cursor.com)
3 points by diwank 1y ago|1 comments
34.
Stress-testing model specs reveals character differences among language models(alignment.anthropic.com)
2 points by diwank 1y ago|1 comments
35.
Thinking Sparks: Emergent Attention Heads in Reasoning Models(arxiv.org)
1 points by diwank 1y ago|0 comments
36.
by diwank 1y ago
I am a bit mind boggled by the pricing lately, especially since the cost increased even further. Is this driven by choices in model deployment (unquantized etc) or simply by perceived quality (as in 'hey our model is crazy good and we
37.
The contextvars and the chain of asyncio tasks in Python(valarmorghulis.io)
2 points by diwank 1y ago|0 comments
38.
AI Village: Watch frontier AIs interact with each other and the world(theaidigest.org)
3 points by diwank 1y ago|0 comments
39.
Discovery of Unstable Singularities in Navier Stokes (DeepMind)(arxiv.org)
2 points by diwank 1y ago|0 comments
40.
Llama.cpp: Deterministic Inference Mode (CUDA): RMSNorm, MatMul, Attention(github.com)
6 points by diwank 1y ago|0 comments
41.
by diwank 1y ago
Remarkable is really lagging behind on this. I was thinking of finally biting the bullet and writing an app for the Paper Pro. Any ideas/takers?
42.
New Ways to Use NotebookLM(blog.google)
4 points by diwank 1y ago|0 comments
43.
Veo 3 and Veo 3 Fast – new pricing, new configurations and better resolution(developers.googleblog.com)
2 points by diwank 1y ago|0 comments
44.
Using AI to perceive the universe in greater depth(deepmind.google)
55 points by diwank 1y ago|23 comments
45.
The Himalayan village where you can see the cosmos(bbc.com)
3 points by diwank 1y ago|0 comments
46.
by diwank 1y ago
Google's response: "Read our statement on today’s decision in the case involving Google Search." https://blog.google/outreach-initiatives/public-policy/doj-s...
47.
Beyond the terminal: Gemini CLI comes to Zed(developers.googleblog.com)
3 points by diwank 1y ago|0 comments
48.
Principal–Agent Problem(en.wikipedia.org)
2 points by diwank 1y ago|0 comments
49.
by diwank 1y ago
Agreed. The fact that it has any structure at all is fascinating (and super pretty). Could signal at interesting internal structures. I would love to see a version for Qwen-3 and Mistral too! I wonder if being trained on significant amounts
50.
by diwank 1y ago
also ettin is a new favorite and a solid alternative: https://huggingface.co/jhu-clsp/ettin-encoder-1b I'd encourage you to give setfit a try, along with aggressively deduplicating your training set, finding top ~
51.
by diwank 1y ago
It's coming soon! I think this experiment has really taught me a lot about the limits of agentic code assistants, stuff that they're good at, they're insanely good at, and stuff that they're horrible at and cannot seem t
52.
by diwank 1y ago
yup. I started a fully autonomous, 100% vibe coded side project called steadytext, mostly expecting it to hit a wall, with LLMs eventually struggling to maintain or fix any non-trivial bug in it. turns out I was wrong, not only has claude o
53.
by diwank 1y ago
> Future robots may learn in their dreams... So prescient. I definitely think this will be a thing in the near future ~12-18 months time horizon
54.
Codestral 25.08 and the Mistral Coding Stack(mistral.ai)
1 points by diwank 1y ago|0 comments
55.
Phi-Ground: Advancing Perception in GUI Grounding(zhangmiaosen2000.github.io)
2 points by diwank 1y ago|0 comments
56.
Voyage-context-3: focused chunk-level details with document context(blog.voyageai.com)
2 points by diwank 1y ago|0 comments
57.
The last human choice: surrendering agency to AI [video](youtube.com)
2 points by diwank 1y ago|0 comments
58.
by diwank 1y ago
Same! Guan’s work on sample packing during finetuning has become a staple. His openchat code is also super simple and easy to understand.
59.
by diwank 1y ago
I think that’s too harsh a position solely for not being peer reviewed yet. Neither of yhe original mamba1 and mamba2 papers were peer reviewed. That said, strong claims warrant strong proofs, and I’m also trying to reproduce the results lo
60.
Physics of Language Models: The Magic of Canon Layers(papers.ssrn.com)
3 points by diwank 1y ago|0 comments

More ›