The RedMonk Programming Language Rankings: January 2025 (redmonk.com)

I've been let off the waitlist. So far, I'm impressed with the Instant Edits. It's crazy fast. I can provide a big HTML file and prompt it to change a color theme and it makes careful edits to just the relevant parts. It seems to be able to parallelize the same instruction to multiple parts of the input. This is incredible for refactoring.

I copied a shader toy example, asked it to rename all the variables to be more descriptive and it edited just the variable names. I was able to compile and run in shader toy.

adt · 30d ago

Good to see some more diffusion models:

https://lifearchitect.ai/models-table/

gs17 · 31d ago

It's ludicrously fast, but it's not ludicrously intelligent, so trying their examples simply led to it failing 100x faster than normal Gemini. Still impressed though. It made a nice tic tac toe ish game, except the computer player became a human player after a few moves and couldn't fix it.

minimaxir · 31d ago

> Diffusion models work differently. Instead of predicting text directly, they learn to generate outputs by refining noise, step-by-step. This means they can iterate on a solution very quickly and error correct during the generation process. This helps them excel at tasks like editing, including in the context of math and code.

This is deliberately unhelpful as it begs the question "why hasn't anyone else made a good text diffusion model in the years since the technology has been available?"

The answer to that question is that unlike latent diffusion for images which can be fuzzy and imprecise before generating the result image, text has discrete outputs and therefore must be more precise, so Google is utilizing some secret sauce to work around that limitation and is keeping it annoyingly close to the chest.

smallerize · 31d ago

Mercury Coder https://news.ycombinator.com/item?id=43187518 has a paper and published code. https://huggingface.co/papers/2503.07197

heliophobicdude · 31d ago

Thank you for sharing this. I'm amazed! Are there any known emergent abilities of it? I ran my evals and seems to struggle in very similar ways to smaller transformer based LLMs

erinaceousjones · 31d ago

I heard it's been a difficult project to justify spending the research/computer time on at scale, because the models use an equivalent amount of compute for training and inference, but more parallelizable. So 5 times more compute units can be required and they get the work done 5 times faster. On a Google scale, that meant the hard internal sell of justifying burning through $25 million worth of compute units over 1 day instead of $5 million each day for 5 days. Something like that.

No comments yet

Brazz Shader (mcpedl.com)

The RedMonk Programming Language Rankings: January 2025 (redmonk.com)

Life expectancy of men in Canadian provinces (lemire.me)

The Problem is Safari for iPad (2024) (spyglass.org)

C++ implementation of the Python NumPy library (github.com)

Show HN: Rancho – An AI tutor that teaches with games and videos

WhatsApp ultimatum: Share data with Facebook or stop using the app (2021) (arstechnica.com)

Show HN: Atlarch – Organize and manage all your interviews from a single place (atlarch.com)

If Day – A Simulated Nazi German Invasion and Occupation of Winnipeg (Canada) (en.wikipedia.org)

Apple Adds Energy and Battery Labels to iPhone and iPad Pages in EU (macrumors.com)

Huawei Unveils HarmonyOS 6 Beta, AI Agents, and Supernode Cloud (gazeon.site)

Meta Messaging Interoperability (developers.facebook.com)

Does RL Incentivize Reasoning Capacity in LLMs Beyond the Base Model? (lesswrong.com)

Social success not about who you know – it's about knowing who knows whom (science.org)

K Isn't Lisp (nsl.com)

Show HN: Stillflo: a minimalist productivity suite to help you focus (stillflo.com)

Delta Chat is a decentralized and secure messenger app (delta.chat)

Dispelling Myths and Misinformation (matrix.org)

Sega mistakenly reveals sales numbers of popular games (gematsu.com)

Is Iran close to building a nuke? (abc.net.au)

Weizmann Institute scrambles to save work after Iranian missile hits labs (thehindu.com)

Show HN: ClipBin; the Simplest, Open Source and Secure Way of Sharing Text/Code (github.com)

People instantly decide whether to trust a product based on design (andrewcoyle.com)

Apple Executives Have Held Internal Talks About Buying AI Startup Perplexity (bloomberg.com)

Vibe coding a data room app (djpardis.medium.com)

Israel-Iran conflict unleashes wave of AI disinformation (bbc.com)

A Hedge Fund Director's Guide to Arbitraging the Tech Job Market (twitter.com)

Why Can't We Break Cryptography? (blog.computationalcomplexity.org)

Practical Uses for Bitwise Operations (beyondit.blog)

I Tried to Warn Them, but We've Lost Control Geoffrey Hinton [video] (youtube.com)

Signal – An Ethical Replacement for WhatsApp (greenstarsproject.org)

Ask HN: How do you programmatically track changes in SEC filings?

Liora Journal – Digital Journaling Assistant (liorajournal.com)

Practical Type Inference with Levels (PLDI 2025 – Distinguished Paper Award) (pldi25.sigplan.org)

Integer Partitions Detect the Primes (pnas.org)

Romance Accelerator (A Dating Show for Founders and Senior Tech Folk) (romanceaccelerator.com)

MonomaOS: A Python-based command-line OS-like environment (v0.0.1) (github.com)

Glass bottles found to contain more microplastics than plastic bottles (phys.org)

AI Expert for Kortix Suna AI Project (suna.so)

The Ethical Compiler: Addressing the Is-Ought Gap in Compilation (PEPM 2025) [video] (youtube.com)

The AGI economy is coming faster than you think (freethink.com)

It's time for California home prices to fall (ocregister.com)

Of Course ML Has Monads (2011) (existentialtype.wordpress.com)

BYOK – Bring Your Own Keyboard (byok.io)

Have a heavy awkward metal sled delivered to you (sled.meteoscientific.com)

A Free Clone of Civilization Game – C-Evo (c-evo.org)

Computer Science via Hip Hop [video] (youtube.com)

Show HN: AI based app building with browser containers (bunnies.build)

The Cult of Personality and Its Consequences (en.wikipedia.org)

Ask HN: How did you meet your co-founder?

Gemini Diffusion

Comments (7)