Nolan Koblischke

@astro_nolan

Astrophysics and language models. PhD student @UofT. Formerly @UBC, @EPFL, @PolymathicAI

nolank.ca

Joined July 2015

Replies Reposts Media

Pinned
Nolan Koblischke
@astro_nolan
Dec 16, 2025
We built a semantic search engine for millions of galaxy images by having LLMs write the captions. These images are completely unlabeled, but our method enables astronomers to search for rare phenomena via text. Try our app! 🔭👇
00:00
2
Nolan Koblischke
@astro_nolan
Sep 9
Submissions are still open for our Machine Learning for Physical Sciences workshop! Papers (4 pages) are due Sept 12, 11:59pm AoE. The ML4PS workshop will be on Dec 11 at Georgia Tech in Atlanta, co-located with NeurIPS and independently organized. ml4physicalsciences.github.io
1
Nolan Koblischke
@astro_nolan
Aug 6
It won't be long before a similar spike in papers happens in physics. So we ran a hackathon @LightconeR. We worked on solutions to ensure research remains observable and reproducible even with a dramatic increase in papers, mostly around our ASTRA spec. A lot of work lies ahead.
00:00
Pablo Groisman 🎲
@pgroisma
Aug 3
Math papers uploaded to @arxiv per month. Jan 1992 to Jul 2026. Source: arxiv.org/year/math/
1
Nolan Koblischke
@astro_nolan
Jul 24
Claude finally aces my plot reading task of reading 10 points off a curve. I was getting a bit worried when Fable failed it. 😅
Nolan Koblischke
@astro_nolan
Dec 11, 2025
GPT-5.2 seems to have fantastic plot reading capabilities, even better than Gemini 3 Pro. I introduced this challenge almost exactly one year ago and it looks like it's basically solved! 🪩 Of course, I will have slightly harder plot reading tasks to share :)
1
Nolan Koblischke
@astro_nolan
Jul 3
Today, our research on making 100M+ galaxy images searchable with AI was published in The Astrophysical Journal! Since the initial release, AION-Search was featured by OpenAI and we were able to discover 36 new galaxies with streams, useful for studying dark matter. [1/2]
Nolan Koblischke
@astro_nolan
Dec 16, 2025
We built a semantic search engine for millions of galaxy images by having LLMs write the captions. These images are completely unlabeled, but our method enables astronomers to search for rare phenomena via text. Try our app! 🔭👇
00:00
2