John F. Wu

3.4K posts

John F. Wu banner
John F. Wu

John F. Wu

@jwuphysics

Associate Astronomer and Applied AI @SpaceTelescope. Secondary appointments in @JohnsHopkins Astro + CS. Opinions my own. He/him. 🦋 @jwuphysics.bsky.social

Baltimore, MD, USA شامل ہوئے Haziran 2020
1K فالونگ1.7K فالوورز
پن کیا گیا ٹویٹ
John F. Wu
John F. Wu@jwuphysics·
New blog post. In the era of LLMs, what does learning look like? What happens when are overconfident in our understanding due to AI sycophancy? Is it even useful to use LLMs for education without outsourcing our thinking? jwuphysics.github.io/blog/2025/12/l…
John F. Wu tweet media
English
8
78
741
51.5K
John F. Wu
John F. Wu@jwuphysics·
@TheZvi Missing child report or mayoral assault aside, seems like the curse is broken!
English
0
0
0
7
Zvi Mowshowitz
Zvi Mowshowitz@TheZvi·
@jwuphysics If I was them I would reverse the initial curse event. Have Mr. and Mrs. met slap Mamdani in the face.
English
1
0
4
138
John F. Wu
John F. Wu@jwuphysics·
@in_defensible @zetalyrae I yearn for the days where we got this instead of slack or ms teams. I mean it still would've sunset in 2023 after going blockbuster in 2020, as is googles imperative, but hey a dream's a dream
English
0
0
2
119
Fernando 🌺🌌
Fernando 🌺🌌@zetalyrae·
Claude is really good at doing corporate y2k boringpunk web design.
Fernando 🌺🌌 tweet media
English
131
239
5.6K
296.5K
Jess Sorrell
Jess Sorrell@JessSorrell·
I took ~7 years off during undergrad. Worked at Starbucks, the postal service, a diner. Wasn't until making friends with some CS PhD students at UW Madison, who suggested sitting in on Eric Bach's class on the physics of computation, that I decided to go back (and then get a PhD)
Vru@vrundasays_

I need more examples of people in academia who haven't had a linear path at all and missed years on the way to PhD and still did their PhD. I don't wanna feel all isolated here 🫩

English
16
232
4K
210.4K
John F. Wu
John F. Wu@jwuphysics·
@francoisfleuret Understand (verb): when ensembles of bags of salty water and organic molecules perform fancy statistics, and it's working on the basis having survived in an enormous amount of evolutionary pressures over time.
English
0
0
2
400
John F. Wu
John F. Wu@jwuphysics·
@sarahookr One of my best friends does make dedicated trips to Montreal just for Bota Bota. Saunas/spas are not my thing, but my wife also said it's an incredible experience!
English
1
0
1
165
Aksel
Aksel@akseljoonas·
Introducing ml-intern, the agent that just automated the post-training team @huggingface It's an open-source implementation of the real research loop that our ML researchers do every day. You give it a prompt, it researches papers, goes through citations, implements ideas in GPU sandboxes, iterates and builds deeply research-backed models for any use case. All built on the Hugging Face ecosystem. It can pull off crazy things: We made it train the best model for scientific reasoning. It went through citations from the official benchmark paper. Found OpenScience and NemoTron-CrossThink, added 7 difficulty-filtered dataset variants from ARC/SciQ/MMLU, and ran 12 SFT runs on Qwen3-1.7B. This pushed the score 10% → 32% on GPQA in under 10h. Claude Code's best: 22.99%. In healthcare settings it inspected available datasets, concluded they were too low quality, and wrote a script to generate 1100 synthetic data points from scratch for emergencies, hedging, multilingual etc. Then upsampled 50x for training. Beat Codex on HealthBench by 60%. For competitive mathematics, it wrote a full GRPO script, launched training with A100 GPUs on hf.co/spaces, watched rewards claim and then collapse, and ran ablations until it succeeded. All fully backed by papers, autonomously. How it works? ml-intern makes full use of the HF ecosystem: - finds papers on arxiv and hf.co/papers, reads them fully, walks citation graphs, pulls datasets referenced in methodology sections and on hf.co/datasets - browses the Hub, reads recent docs, inspects datasets and reformats them before training so it doesn't waste GPU hours on bad data - launches training jobs on HF Jobs if no local GPUs are available, monitors runs, reads its own eval outputs, diagnoses failures, retrains ml-intern deeply embodies how researchers work and think. It knows how data should look like and what good models feel like. Releasing it today as a CLI and a web app you can use from your phone/desktop. CLI: github.com/huggingface/ml… Web + mobile: huggingface.co/spaces/smolage… And the best part? We also provisioned 1k$ GPU resources and Anthropic credits for the quickest among you to use.
English
118
545
4.1K
785.2K
John F. Wu ری ٹویٹ کیا
Space Telescope Science Institute
Hubble marks its 36th anniversary with a shimmering close-up of star-formation in the Trifid Nebula! Tiny, actively forming stars are eating and spewing material all around. (One at top left in brown, and two fiery red jets.) Explore it all: news.stsci.edu/4cvi5jL
Space Telescope Science Institute tweet media
English
1
15
47
2.7K
John F. Wu
John F. Wu@jwuphysics·
Breathtaking.
Reid Wiseman@astro_reid

Only one chance in this lifetime… Like watching sunset at the beach from the most foreign seat in the cosmos, I couldn’t resist a cell phone video of Earthset. You can hear the shutter on the Nikon as @Astro_Christina is hammering away on 3-shot brackets and capturing those exceptional Earthset photos through the 400mm lens. @AstroVicGlover was in window 3 watching with @Astro_Jeremy next to him. I could barely see the Moon through the docking hatch window but the iPhone was the perfect size to catch the view…this is uncropped, uncut with 8x zoom which is quite comparable to the view of the human eye. Enjoy.

English
0
0
5
163
John F. Wu
John F. Wu@jwuphysics·
@Miles_Brundage Agreed, Claude does a good job here by actually prompting for specifications that I feel like I should've provided. Also there's usually only 3-5 of them, which I assume is intentional, but still a reasonable followup that I don't mind answering.
English
0
0
1
21
Miles Brundage
Miles Brundage@Miles_Brundage·
Think Deep Research from OAI is what started that kind of thing, though the GUI version is nice - don't like having to type follow-up Qs because then I either feel rude due to terseness or feel like I'm wasting time writing a long answer
English
3
0
9
1.2K
Miles Brundage
Miles Brundage@Miles_Brundage·
I like the little multiple choice follow-up question things in the Claude app (maybe has happened in ChatGPT / Codex sometimes also, I forget)
English
8
0
60
12K
John F. Wu
John F. Wu@jwuphysics·
@justin_fenton Agree, though I prefer roller coasters that aren't like this
GIF
English
0
0
1
43
Justin Fenton
Justin Fenton@justin_fenton·
Damn. That ninth inning was a roller coaster ride. #orioles
English
2
0
14
2.3K
John F. Wu
John F. Wu@jwuphysics·
So cool that Annual Reviews sent me a physical volume of ARA&A 63! A great memento from when I served as a guest reviewer.
John F. Wu tweet media
English
2
0
12
446
John F. Wu
John F. Wu@jwuphysics·
Excited to deliver the University of Kentucky astro seminar! First time in Lexington too -- I did not realize how many horse farms there are here! as.uky.edu/physics-astron…
English
0
0
1
116
John F. Wu
John F. Wu@jwuphysics·
Feels like an astronomer wrote this! The covariance shows up everywhere -- from the CMB angular power spectrum (we literally ignore the mean) to radio interferometry (which I worked on for 3 months in South Africa during my PhD). Clever to see this used in embeddings!
Goodfire@GoodfireAI

New research: we propose *covariance pooling* as a better replacement for mean pooling that improves probing for sequence-level properties. E.g., genomic model embeddings are often mean-pooled to understand genes - but that throws away all info about feature co-occurrence! (1/3)

English
0
0
9
630
John F. Wu
John F. Wu@jwuphysics·
Just taught a bunch of 5-year-old kids about astronomy. They sure had a lot of questions!
English
0
0
5
113
John F. Wu ری ٹویٹ کیا
Kanjun 🐙
Kanjun 🐙@kanjun·
Twitter’s algorithm is optimized for addiction, not for us. We deserve better. We’re releasing Bouncer today so you can take back control of your feed. Describe what you don't want, and Bouncer removes it. It’s free, doesn’t collect your data, and will be open source soon.
English
212
294
3.1K
579.9K