BusyBrain⌨'s Avatar

BusyBrain⌨

@w42.bsky.social

Interested in how machine can speak human language and automated theorem proving

426 Followers  |  4,660 Following  |  141 Posts  |  Joined: 28.10.2024  |  1.7453

Latest posts by w42.bsky.social on Bluesky

Post image

🤯 We use the term 'intelligence' a lot, but wth do we mean?

We got 303 survey responses from researchers. The most agreed-on criteria are generalization, adaptability & reasoning.

ACL Findings preprint: arxiv.org/abs/2505.20959
with @brtrm.bsky.social @terne.bsky.social @heinrichst.bsky.social /1

02.06.2025 09:27 — 👍 62    🔁 15    💬 4    📌 3
Post image

Happy Friday everyone! I just posted what I think is an important blog post on my website. It is a critique of meta-meta-analyses: meta-analyses of meta-analyses.

Link: matthewbjane.github.io/blog-posts/b...

#stats #metascience

23.05.2025 22:47 — 👍 85    🔁 26    💬 6    📌 4
Post image

1/🧵ICLR 2025 Spotlight Research on LM & Memorization!
Language models (LMs) often "memorize" data, leading to privacy risks. This paper explores ways to reduce that!
Paper: arxiv.org/pdf/2410.02159
Code: github.com/msakarvadia/...
Blog: mansisak.com/memorization/

04.03.2025 18:15 — 👍 4    🔁 2    💬 1    📌 1
Video thumbnail

Microtubule regulation drives an asymmetry in the regeneration of sensory neurons, with specific proteins controlling growth.
buff.ly/4ijksHC

04.03.2025 18:27 — 👍 11    🔁 2    💬 0    📌 0
Post image

1/13 New Paper!! We try to understand why some LMs self-improve their reasoning while others hit a wall. The key? Cognitive behaviors! Read our paper on how the right cognitive behaviors can make all the difference in a model's ability to improve with RL! 🧵

04.03.2025 18:15 — 👍 57    🔁 17    💬 2    📌 3
Video thumbnail

Reading and Writing Google Sheets in DuckDB duckdb.org/2025/02/26/g...

01.03.2025 10:29 — 👍 19    🔁 4    💬 0    📌 0
Preview
Brain-wide presynaptic networks of functionally distinct cortical neurons - Nature Behavioural-state-dependent pyramidal neurons have a distinct pattern of long-range glutamatergic inputs, with a larger proportion of thalamic versus motor cortex inputs compared with non-behavio...

www.nature.com/articles/s41... awesome new work out today! From the Lee lab in the intramural research program at NIMH!

27.02.2025 00:21 — 👍 82    🔁 34    💬 0    📌 1

Our online book on systems principles of LLM scaling is live at jax-ml.github.io/scaling-book/

We hope that it helps you make the most of your computing resources. Enjoy!

04.02.2025 18:59 — 👍 34    🔁 9    💬 3    📌 1
InfAlign: Inference-aware language model alignment
Ananth Balashankar, Ziteng Sun, Jonathan Berant, Jacob Eisenstein, Michael Collins, Adrian Hutter, Jong Lee, Chirag Nagpal, Flavien Prost, Aradhana Sinha, Ananda Theertha Suresh, Ahmad Beirami

InfAlign: Inference-aware language model alignment Ananth Balashankar, Ziteng Sun, Jonathan Berant, Jacob Eisenstein, Michael Collins, Adrian Hutter, Jong Lee, Chirag Nagpal, Flavien Prost, Aradhana Sinha, Ananda Theertha Suresh, Ahmad Beirami

Excited to share 𝐈𝐧𝐟𝐀𝐥𝐢𝐠𝐧!

Alignment optimization objective implicitly assumes 𝘴𝘢𝘮𝘱𝘭𝘪𝘯𝘨 from the resulting aligned model. But we are increasingly using different and sometimes sophisticated inference-time compute algorithms.

How to resolve this discrepancy?🧵

01.01.2025 19:59 — 👍 55    🔁 11    💬 2    📌 1

what kind news is this? lol

02.02.2025 07:44 — 👍 2    🔁 0    💬 0    📌 0
Post image

On monday in our reading group we discuss "Flow Matching with General Discrete Paths: A Kinetic-Optimal Perspective" arxiv.org/abs/2412.03487
With Neta Shaul.

Join on zoom on Monday at 9am PT / 12pm ET / 6pm CET: portal.valencelabs.com/logg

25.01.2025 21:08 — 👍 13    🔁 2    💬 0    📌 0

woah, wish you luck back there

01.02.2025 20:32 — 👍 2    🔁 0    💬 0    📌 0
Preview
4.5 Million (Suspected) Fake Stars in GitHub: A Growing Spiral of Popularity Contests, Scams, and Malware GitHub, the de-facto platform for open-source software development, provides a set of social-media-like features to signal high-quality repositories. Among them, the star count is the most widely used...

🚨 Researchers uncover 4.5M fake stars on GitHub 🌟, often boosting malware disguised as pirated software & crypto bots. Fake stars surge in 2024, posing major risks to open-source trust & security.

#CyberSecurity #GitHub #OpenSource #SupplyChainSecurity

arxiv.org/abs/2412.13459

20.12.2024 20:58 — 👍 5    🔁 2    💬 0    📌 1

another feeling that so magical is trying to code the structure you see in your mind on computer and deconstruct it like kids playing LEGO

09.12.2024 09:31 — 👍 1    🔁 0    💬 0    📌 0

That exhilarating feeling that *everything is possible* when you open an editor to code, it hopefully never goes away.

09.12.2024 07:25 — 👍 65    🔁 6    💬 4    📌 0
Post image

Binary code is pervasive, and binary analysis is a key task in reverse engineering, malware classification, and vulnerability discovery. So, they created Assemblage - the dataset of source-to-binary projects compiled from GitHub.

Assemblage - A dataset of binary executable corpuses

08.12.2024 02:58 — 👍 19    🔁 5    💬 1    📌 0
This image shows how a divergent chemical mixture can 'evolve' into a mixture of species, or be dominated by just one. Credit: Otto Lab / University of Groningen

This image shows how a divergent chemical mixture can 'evolve' into a mixture of species, or be dominated by just one. Credit: Otto Lab / University of Groningen

What came first, life or evolution?
Does evolution act on non-living materials?

Competitive Exclusion among Self-Replicating Molecules Curtails the Tendency of Chemistry to Diversify 🧪
www.nature.com/articles/s41...

Self-replicating molecules demonstrate basic principles of Darwinian evolution

05.12.2024 13:30 — 👍 47    🔁 8    💬 3    📌 1

this morning walk, an ideas stuck me: can you play chess on Rubik's Cube (does not have to be 3x3 one)? not just chess with 6 sides, but normal chess board abstracted away to Rubik's Cube representation and operation

02.12.2024 18:27 — 👍 3    🔁 0    💬 0    📌 0

Half of Twitter right now is people getting mad at some random lady that got a literature PhD. Seems a bit crazy to get so mad about, but I do agree woke academia has become silly and we need to go back to when it was about real solid research, like measuring skull sizes to determine personalities

02.12.2024 12:02 — 👍 817    🔁 48    💬 20    📌 6
Video thumbnail

The amazing, new Qwen2.5-Coder 32B model can now write SQL for any @hf.co dataset ✨

02.12.2024 12:48 — 👍 19    🔁 4    💬 1    📌 0
Post image Post image Post image

Good news everyone! A new version of graph-tool is just out! @graph-tool.skewed.de

graph-tool.skewed.de

Graph-tool is a comprehensive and efficient Python library to work with networks, including structural, dynamical, and statistical algorithms, as well as visualization. 1/N

#networkscience

02.12.2024 12:55 — 👍 346    🔁 98    💬 8    📌 5

An aspect of flow matching which I find a bit interesting is that it is covariant under affine changes of coordinate (c.f. optimal transport, which need not be). This allows for a few nice WLOGs, which I imagine have more applications than I realise.

02.12.2024 13:21 — 👍 22    🔁 3    💬 1    📌 0
Preview
Claude Talk with Claude, an AI assistant from Anthropic

you should talk to claude.ai (I'm not paid), feel so hard not to assign persona to the other side. maybe that's our bias toward animalism

01.12.2024 22:17 — 👍 0    🔁 0    💬 0    📌 0
Post image

If you think the out (site) group isn't enjoying thinking like your ingroup, I've lost respect for you. Sorry.

01.12.2024 22:10 — 👍 1    🔁 0    💬 0    📌 0
01.12.2024 21:55 — 👍 3    🔁 1    💬 0    📌 0

do i need to pay my homie this time?

01.12.2024 21:59 — 👍 1    🔁 0    💬 0    📌 0
Preview
How we prevent conflicts in authoritative DNS configuration using formal verification We describe how Cloudflare uses a custom Lisp-like programming language and formal verifier (written in Racket and Rosette) to prevent logical contradictions in our authoritative DNS nameserver’s beha...

More formal verification, this time from the engineers at Cloudflare using a lesser-known verification stack:

Cloudflare uses racket & rosette, a solver-aided programming system to, ensure the correctness of their DNS query engine configuration

blog.cloudflare.com/topaz-policy...

21.11.2024 11:50 — 👍 12    🔁 1    💬 0    📌 2
Preview
The KDE Plasma desktop, in an atomic fashion

Fedora Kinoite is atomic distro, that mean you can mess the underlying system up. you use Flatpak heavily. I'm not a fan of that approach bc limit my tinkering hobby

fedoraproject.org/atomic-deskt...

01.12.2024 17:17 — 👍 1    🔁 0    💬 1    📌 0
Post image

Some recent discussions made me write up a short read on how I think about doing computer vision research when there's clear potential for abuse.

Alternative title: why I decided to stop working on tracking.

Curious about other's thoughts on this.

lb.eyer.be/s/cv-ethics....

29.11.2024 14:51 — 👍 175    🔁 20    💬 19    📌 7

I don't know if this was known or not, but if you open your Google search page, type 'Chicxulub' and press enter, something interesting happens.

Easter egg? But a funny one!

30.11.2024 18:59 — 👍 14    🔁 6    💬 2    📌 1

@w42 is following 20 prominent accounts