iiinova — where your digital self keeps evolving

Reposted by KS

mr. TIM @timkellogg.me · 01/10/2026

Context Language Models New agent architecture where the LLM can edit its own context it seems to have emergent capabilities, creates its own memory management & organization algorithms, and coordinates multi agents github.com/facebookrese...

Diagram titled "How a Context Language Model edits its context: A simple step-by-step view" outlining an 8-step process:
* Start of turn: Current editable context exists in memory with old messages.
* LLM reads the context: The LLM evaluates the context and decides to run a bash command to edit it.
* Harness mirrors context: The harness mirrors the old editable context into a file at /tmp/.live_ctx/LIVE_CTX_MAIN.txt.
* Bash command runs: The bash command executes and may edit that file.
* Harness parses file: If the file changed, the harness parses it back into a new edited context.
* Tool call appended: The current assistant tool call is appended to the edited context.
* Tool result appended: The tool result is appended below the tool call.
* Next turn starts: The next turn begins with the edited old context, previous tool call, and previous tool result.
Key idea box: "The model edits the prior context first. The tool call and tool result from the current turn are appended afterward, so they can only be compacted on the next turn."
Footer summary:
* Ordinary LM: Context mostly grows by appending.
* CLM: The model can rewrite the editable part of context between turns.

1620515

Reposted by KS

Dave Willner @dwillner.bsky.social · 27/05/2026

@samidh.bsky.social and I are releasing CoPE-B today, the next version of our policy-adaptive content classifier. It delivers at-or-better-than-frontier classification while being self-hostable, faster, and cheaper to run. Full writeup with benchmarks at blog.zentropi.ai/meet-cope-b-.... 🧵 1/9

blog.zentropi.ai

Meet CoPE-B: Frontier-Quality Content Classification You Can Self-Host

TL;DR: * Today we're releasing CoPE-B, our next-gen small language model for policy-adaptive content classification * CoPE-B-A4B (text-only) is open weights under Apache 2.0 and free to use * CoPE...

25617

Reposted by KS

American Meteorological Society @ametsoc.org · 13/05/2026

Scientists have developed a new method called GOFLOW (Geostationary Ocean Flow) using AI to measure ocean surface currents over large areas in greater detail than ever before. Discover how scientists use deep learning in their research: bit.ly/4sAJHtM

bit.ly

New AI Approach Reveals Ocean Currents in Unprecedented Detail

Scientists have developed a new method to measure ocean surface currents over large areas in greater detail than ever before.

0113

Reposted by KS

Unsloth AI @unsloth.ai · 05/05/2026

We made a guide on how to run open LLMs in Claude Code, Codex and OpenClaw. Use Gemma 4 and Qwen3.6 GGUFs for local agentic coding on 24GB RAM Run with self-healing tool calls, code execution, web search via the Unsloth API endpoint and llama.cpp Guide: unsloth.ai/docs/basics/...

2364

Reposted by KS

Sophia Sirko-Galouchenko 🇺🇦 @ssirko.bsky.social · 17/04/2026

1/n New paper - V-GIFT 🎁 Self-supervised tasks like rotation prediction or colorization were big in 2018. Do they still matter? Yes. We turn them into visual instruction tuning data for MLLMs. Result: models rely more on the image and perform better on vision tasks 👀

1248

Reposted by KS

Ai2 @ai2.bsky.social · 21/01/2026

Introducing HiRO-ACE: an AI framework that makes highly detailed climate simulations dramatically more accessible. It generates decades of high-resolution precipitation data for any region in a day on a single GPU—no supercomputing cluster required. 🧵

1348

Reposted by KS

Simon Willison @simonwillison.net · 25/12/2025

I built a new Python CLI tool called claude-code-transcripts that can create nice readable HTML versions of your Claude Code sessions, both local and pulled from Claude Code for web, and makes it easy to publish them online too simonwillison.net/2025/Dec/25/...

simonwillison.net

A new way to extract detailed transcripts from Claude Code

I’ve released claude-code-transcripts, a new Python CLI tool for converting Claude Code transcripts to detailed HTML pages that provide a better interface for understanding what Claude Code has done t...

1014912

Reposted by KS

Daphne Cornelisse @daphne-cornelisse.bsky.social · 13/10/2025

Rapid RL experimentation is great. But how do you catch silent errors before they slip by? In this post, I share tools and habits that help me move quickly from idea to result without sacrificing reliability.

open.substack.com

How to catch subtle RL bugs before they catch you

Tools and habits for reliable, fast RL experimentation and development

0425

Reposted by KS

N. Thuerey's research group at TUM @thuereygroup.bsky.social · 10/12/2024

If you're training models with more than one loss term, I can again strongly recommend our ConFIG optimizer: tum-pbs.github.io/ConFIG/ , simply swap out Adam&Co. for ConFIG, and you can potentially see substantial reductions in your training loss 😁 We'd also be curious to hear how it works for you

0204

Reposted by KS

Prof Michael Meredith 🌊🧪🥼❄️ @oceanandice.bsky.social · 04/01/2025

The densest water in the world ocean forms around Antarctica, and then spreads out to flood the global abyss. We’ve seen the production of this water dwindle over the past few decades, a change that is now impacting the North Atlantic. 🌊🧪❄️ www.nature.com/articles/s41...

nature.com

Weakening of the Atlantic Meridional Overturning Circulation abyssal limb in the North Atlantic - Nature Geoscience

Mooring observations and hydrographic data suggest the Atlantic Meridional Overturning Circulation abyssal limb has weakened over the past two decades in the North Atlantic, most likely due to reduced...

115574

Reposted by KS

Ketan Joshi @ketanjoshi.co · 26/11/2024

Okay, I have to do a thread on this pretty amazing article because it is just seriously chockers with eye-opening stats about 'direct air capture' - ie, sucking up air and trying to remove carbon dioxide from it news.mit.edu/2024/reality...

news.mit.edu

Reality check on technologies to remove carbon dioxide from the air

An MIT Energy Initiative study finds many climate-stabilization plans are based on questionable assumptions about the future cost and deployment of “direct air capture” and therefore may not bring abo...

49828349

Reposted by KS

Hamed Shirzad @hamedshirzad.bsky.social · 05/12/2024

Graph Transformers (GTs) can handle long-range dependencies and resolve information bottlenecks, but they’re computationally expensive. Our new model, Spexphormer, helps scale them to much larger graphs – check it out at NeurIPS next week, or the preview here! [1/13] #NeurIPS2024

16514

Reposted by KS

ruiqigao.bsky.social @ruiqigao.bsky.social · 02/12/2024

A common question nowadays: Which is better, diffusion or flow matching? 🤔 Our answer: They’re two sides of the same coin. We wrote a blog post to show how diffusion models and Gaussian flow matching are equivalent. That’s great: It means you can use them interchangeably.

625459

Reposted by KS

Miles Cranmer @milescranmer.bsky.social · 02/12/2024

🧵 Today with @polymathicai.bsky.social and others we're releasing two massive datasets that span dozens of fields - from bacterial growth to supernova! We want this to enable multi-disciplinary foundation model research.

38825

Reposted by KS

Kai Kornhuber @kornhuber.bsky.social · 26/11/2024

'Global emergence of regional heatwave hotspots outpaces climate model simulations' our new paper in @pnas.org with S. Bartusek, R. Seager. J. Schellnhuber and M. Ting investigating the tail behaviour of extreme heatwave trends. @iiasa.ac.at @columbiaclimate.bsky.social @lamontearth.bsky.social

15356147

Reposted by KS

N. Thuerey's research group at TUM @thuereygroup.bsky.social · 27/11/2024

I'm excited to share our APEBench paper arxiv.org/abs/2411.00180 and code github.com/tum-pbs/apeb..., to be presented at #NeurIPS. Congratulations Felix and Simon 😀 👍 At its core, APEBench features a lightning-fast ⚡️ fully differentiable spectral solver with a huge range of different PDEs.

0213

Reposted by KS

Luca Ambrogioni @lucamb.bsky.social · 22/11/2024

I am happy to share here our paper: "Spontaneous symmetry breaking in generative diffusion models", published at Neurips 2023. We found that the generative capabilities of diffusion models are the result of a phase transition! Preprint: arxiv.org/abs/2305.19693 Code: github.com/gabrielraya/...

1365

Reposted by KS

Mufan Li @mufan-li.bsky.social · 22/11/2024

Since I'm moving over, I would like to reshare my most recent blog post on why you should learn Optimal Transport. Part 2 is coming, most likely during the winter break :) mufan-li.github.io/OT1/

mufan-li.github.io

Everyone Should Learn Optimal Transport, Part 1

In my opinion, optimal transport (OT) is a seriously underrated topic. I think part of the reason is the way OT is often introduced: as an optimization problem or a metric on probability distributions...

1356

Reposted by KS

Christoph Molnar @christophmolnar.bsky.social · 20/11/2024

Even as an interpretable ML researcher, I wasn't sure what to make of Mechanistic Interpretability, which seemed to come out of nowhere not too long ago. But then I found the paper "Mechanistic?" by @nsaphra.bsky.social and @sarah-nlp.bsky.social, which clarified things.

723026

Reposted by KS

Gabriele Corso @gcorso.bsky.social · 17/11/2024

Thrilled to announce Boltz-1, the first open-source and commercially available model to achieve AlphaFold3-level accuracy on biomolecular structure prediction! An exciting collaboration with Jeremy, Saro, and an amazing team at MIT and Genesis Therapeutics. A thread!

18610204

Reposted by KS

David Picard @davidpicard.eurosky.social · 20/11/2024

🍏 New preprint alert! 🍏 PoM: Efficient Image and Video Generation with the Polynomial Mixer arxiv.org/abs/2411.12663 This is my latest "summer project" and it was so big I had to call in reinforcements (Thanks @nicolasdufour.bsky.social) TL;DR Transformers are for boomers, welcome to the future 🧵👇

arxiv.org

PoM: Efficient Image and Video Generation with the Polynomial Mixer

Diffusion models based on Multi-Head Attention (MHA) have become ubiquitous to generate high quality images and videos. However, encoding an image or a video as a sequence of patches results in costly...

19323

Reposted by KS

Christoph Molnar @christophmolnar.bsky.social · 15/11/2024

Interested in machine learning in science? Timo and I recently published a book, and even if you are not a scientist, you'll find useful overviews of topics like causality and robustness. The best part is that you can read it for free: ml-science-book.com

712929