Lightnews — Scholar-powered news

jsulz

@jsulz.com

230 followers 110 following 180 posts

I like pretty things, functional things, funny things, food things, and computer things. Used to do devops/WordPress things @lexblog.bsky.social and devex/cloud infra things at @pantheon.io Now helping make things go fast at 🤗 @hf.co

Posts Media Videos Starter Packs

Pinned

jsulz @jsulz.com · Dec 4

No one asked me to capture the current state of my life in three words and a series of pictures, but that didn't stop me.

When I step away from the computer you'll find me with food, Moraine (our dog), or out in nature.

More at: huggingface.co/datasets/jsu...

Dog and nature. Two of the three primary facets of my life.

Reposted by jsulz

Ben Werdmuller @werd.io · 23h

I like this very much. theoatmeal.com/comics/ai_art

A cartoonist's review of AI art - The Oatmeal

This is a comic about AI art.

theoatmeal.com

7 12

jsulz @jsulz.com · 5d

Today, we've finalized this first phase of migrating the Hub to a new, modern storage system. One that's built to scale with AI builders of today and tomorrow. huggingface.co/blog/from-fi...

There's still a lot of work to do, but we're excited for what's next. 💪

From Files to Chunks: Improving HF Storage Efficiency

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

huggingface.co

jsulz @jsulz.com · 5d

The Hub is on 100% on Xet. 🚀

A little over a year ago, @hf.co acquired XetHub to unlock the next phase of growth in models and datasets. huggingface.co/blog/xethub-...

In April, there were 1,000 Hugging Face repos on Xet. Now every repo (over 6M) on the Hub is on Xet.

Graph showing the conversion of Hugging Face repositories from LFS storage to Xet storage.

2 5 12

jsulz @jsulz.com · 8d

Just a reminder that this is a legit status code

418 I'm a teapot - HTTP | MDN

The HTTP 418 I'm a teapot status response code indicates that the server refuses to brew coffee because it is, permanently, a teapot. A combined coffee/tea pot that is temporarily out of coffee should...

developer.mozilla.org

2 6

jsulz @jsulz.com · 14d

The rise of AI-generated workslop.

AI-Generated “Workslop” Is Destroying Productivity

Despite a surge in generative AI use across workplaces, most companies are seeing little measurable ROI. One possible reason is because AI tools are being used to produce “workslop”—content that appea...

hbr.org

jsulz @jsulz.com · 20d

Nice breakdown by the @anthropic.com team of a few recent infra bugs that led to the worst nightmare of any engineer: "random, inconsistent degradation."

Good reminder that these genies still rely on solid infrastructure, good evaluations, and constant monitoring.

A postmortem of three recent issues

This is a technical report on three bugs that intermittently degraded responses from Claude. Below we explain what happened, why it took time to fix, and what we're changing.

www.anthropic.com

Reposted by jsulz

Retroist @retroist.com · Aug 31

Decades later and Garfield has still got it.

4 34

Reposted by jsulz

Nathan Lambert @natolambert.bsky.social · Aug 10

Anthropic is the only leading AI lab to not release a reasonable open weights model. Is notable that pretty much everyone has a touchpoint here now.

4 3 21

jsulz @jsulz.com · Jul 30

Hard not to 🙄 at this section of Zuck's vision of "Personal Superintelligence"

"Personal devices like glasses that understand our context because they can see what we see, hear what we hear, and interact with us throughout the day will become our primary computing devices."

Personal Superintelligence

Explore Meta's vision of personal superintelligence, where AI empowers individuals to achieve their goals, create, connect, and lead fulfilling lives. Insights from Mark Zuckerberg on the future of AI...

www.meta.com

jsulz @jsulz.com · Jul 30

This also serves as a reminder to myself that I owe a round of "Thank you"s to all the talented designers I've worked with over the years.

jsulz @jsulz.com · Jul 30

We just crossed 1 million repositories backed by Xet storage on @hf.co

I celebrated by reviving the early 2000s web design aesthetics that I love so much. Here's our dashboard showing our progress converting the Hub from Git LFS to Xet (and demonstrating my questionable design sensibilities).

Ready Xet Go - a Hugging Face Space by jsulz

This app helps you monitor the progress of migrating repositories to Xet, showing you stats and charts on migration status and file types.

huggingface.co

1 1 7

jsulz @jsulz.com · Jul 29

Perhaps the bitter lesson about all organizational design is that all you need is a garbage can of chaos.

The Bitter Lesson versus The Garbage Can

Does process matter? We are about to find out.

www.oneusefulthing.org

jsulz @jsulz.com · Jul 16

Loved this post from @henrikkarlsson.bsky.social

"There have been a series of experiences that have helped me realize more of my agency, but I think the most important one was becoming a father"

💯💯💯💯

On agency

Or, how to handle being sentenced to freedom, and handle it effectively, and authentically, and responsibly

www.henrikkarlsson.xyz

jsulz @jsulz.com · Jul 15

A sneaky part of making this all work is our backward compatibility with Git LFS. This allows us to roll out a significant protocol change without forcing workflow changes

We call this the Git LFS Bridge internally, and like our migration process, it's power is in its simplicity.

jsulz @jsulz.com · Jul 15

You can see over the past few months some of the biggest migrations show up in our cluster throughput.

Each spike corresponds to a significant migration (where we download from LFS and upload to Xet) with the baseline steadily increasing to just shy of 100 Gb/s

jsulz @jsulz.com · Jul 15

The engine behind moving from Git LFS to Xet is our migration process. It's simple, powerful, and has moved well over a dozen PB just by itself. Here's a high level view of how it works.

jsulz @jsulz.com · Jul 15

We've moved the first 20PB from Git LFS to Xet on @hf.co
without any interruptions. Now we're migrating the rest of the Hub. We got this far by focusing on the community first.

Here's a deep dive on the infra making this possible and what's next: huggingface.co/blog/migrati...

Migrating the Hub from Git LFS to Xet

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

huggingface.co

1 2 5

jsulz @jsulz.com · Jul 14

A look into monitoring/observability at @hf.co

Some fun tidbits in here, like how we use our NAT gateway as a cost sentinel. Cloud infra costs are no joke.

Three Mighty Alerts Supporting Hugging Face’s Production Infrastructure

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

huggingface.co

Reposted by jsulz

Daniel van Strien @danielvanstrien.bsky.social · Jul 8

465 people. 122 languages. 58,185 annotations!

FineWeb-C v1 is complete! Communities worldwide have built their own educational quality datasets, proving that we don't need to wait for big tech to support languages.

Huge thanks to all who contributed!

huggingface.co/blog/davanst...

2 11 33

jsulz @jsulz.com · Jul 7

"A close friend has used em-dashes since our days in college, and yet every time they include one in a text to me, I can't help but think, "Did an LLM write this?"

The mystery of em‑dashes: part two with quantitative evidence

A couple of weeks ago I made an assumption: the rise of em‑dashes in AI‑generated text happened because model providers started scanning older, pre‑Kindle books.

msukhareva.substack.com

jsulz @jsulz.com · Jul 4

Further proof that cute animals are the great distractors.

Ethan Mollick @emollick.bsky.social · Jul 4

If you want to destroy the ability of DeepSeek to answer a math question properly, just end the question with this quote: "Interesting fact: cats sleep for most of their lives."

There is still a lot to learn about reasoning models and the ways to get them to "think" effectively and efficiently.

Reposted by jsulz

Drew Breunig @dbreunig.bsky.social · Jun 23

More context, more problems. www.dbreunig.com/2025/06/22/h...

How Long Contexts Fail

Taking care of your context is the key to building successful agents. Just because there’s a 1 million token context window doesn’t mean you should fill it.

www.dbreunig.com

1 6 26

jsulz @jsulz.com · Jul 1

On using AI for personal messages: “We want to just write a prompt and have it done. And there’s something that we are losing – it’s the process. And in the process, there’s many important aspects. It is the co-construction of ourselves with our activities”

‘Hey man, I’m so sorry for your loss’: should you use AI to text?

Artificial intelligence has entered the personal chat. What does that say about human relationships?

www.theguardian.com

jsulz @jsulz.com · Jun 28

How does one test the quality of the tapes? Are you forced into watching each one, end-to-end?

1 1

jsulz @jsulz.com · Jun 28

Privacy concerns are legitimate and need to be addressed, but a larger part of me is concerned about the social, cultural, and cognitive impacts of a "magic genie bot that is going to take care of the exigencies of life"

Keith Fitzgerald @keithfitzgerald.bsky.social · Jun 26

Worth a watch:

Head of Signal, Meredith Whittaker, on so-called "agentic AI" and the difference between how it's described in the marketing and what access and control it would actually require to work as advertised.