Reza Rezvan

PhD student position in neuro-symbolic AI in my group. Cross-disciplinary project with applied mechanics solving problems with recycling aluminium, contributing to sustainability: www.chalmers.se/en/about-cha...

AIME team fika in our cozy lunch room.

AIME at MSc thesis project fair.

AIME at Burger Thursdays.

🚨 We are currently recruiting 2 PhD students to join our team at Chalmers, which you can learn more about on our team website: ailab.bio/join-us

Details in thread below! (1/5)

The official opening to this position was just posted: www.chalmers.se/en/about-cha...

We are looking for someone to join the group as a postdoc to help us with scaling implicit transfer operators. If you are interested in this, please reach out to me through email. Include CV, with publications and brief motivational statement. RTs appreciated!

"mistakes" due to new posteriors and malicious intentions can be hard to differentiate without the right context :) i would like to think that people assume the former but sometimes you wonder 😅

We propose Neurosymbolic Diffusion Models! We find diffusion is especially compelling for neurosymbolic approaches, combining powerful multimodal understanding with symbolic reasoning 🚀

Read more 👇

when i've been reading/hearing a language {فارسی , 官话, or even English sometimes} i start to subconsciously hear that language from crowd noise, sometimes it's just nonsense and the people are not even speaking that language

cool how the brain works

sioyek.info

unsure if you can hover, but you can jump to labels, citations etc., and jump back, but it's a bit more keyboard oriented (vim bindings)

My department at Chalmers is recruiting various lecturers in Computer Science and Engineering: www.chalmers.se/en/about-cha... Do apply if you are interested and spread the word! Deadline May 16th.

I’ll be at AISTATS presenting our work on designing efficient and scalable equivariant models for 3D molecular generation. You can find me at poster session 2, no 132. Also very happy to meet up during the conference!

In continuous generative diffusion, the conditional entropy rate is the constant term that separates the score matching and the denoising score matching loss

This can be directly interpreted as the information transfer (bit rate) from the state x_t and the final generation x_0.

Also, having the ability to search (prompt) back and forth is easier with an (L)LM I think.

However, I do miss old Google for simple easy searches :/

I agree, but I think when you are aware if you are "searching" with an LLM, your query is a bit different.

I.e., "I'm trying to find a pop song that goes like dun dun... [any additional info]" VS. "pop song that goes dun dun", the latter case is harder to use additional info in the query.

🌌🛰️🔭Wanna know which features are universal vs unique in your models and how to find them? Excited to share our preprint: "Universal Sparse Autoencoders: Interpretable Cross-Model Concept Alignment"!

arxiv.org/abs/2502.03714

(1/9)

Decisions during generative diffusion are analogous to phase transitions in physics. They can be identified as peaks in the conditional entropy rate curve!

GitHub - daskol/typst-templates: A list of paper templates in the area of machine learning. A list of paper templates in the area of machine learning. - daskol/typst-templates

github.com/daskol/typst...

Can LaTeX be uncrowned :o?

The AI Researcher's Guide to a Non-Boring Bluesky Feed | Naomi Saphra How to migrate to bsky without a boring feed.

I wrote something up for AI people who want to get into bluesky and either couldn't assemble an exciting feed or gave up doomscrolling when their Following feed switched to talking politics 24/7.

1 week left to apply for this Assistant Professorship at Chalmers!

2025 CHAIR Structured Learning Workshop Welcome to the 2025 Chalmers AI Research Center Workshop for Structured Learning. In this workshop we broadly cover topics related to Structured Learning targeting specifically the following questions...

Registration for this years CHAIR Structured Learning Workshop is open. Speakers include: Klaus Robert Müller, Jens Sjölund, @alextong.bsky.social ,
@janstuehmer.bsky.social, @arnauddoucet.bsky.social, @marcocuturi.bsky.social , Marta Betcke,
Elena Agliari, Beatriz Seoane, Alessandro Ingrosso

Excited to present our poster on Boltzmann priors for Implicit Transfer Operators tomorrow at @iclr-conf.bsky.social!
See you tomorrow at poster 13, 10-12:30.

Easter is for the rebirth of new things, redesigning my website and my other tools :)

Check out cool new work from our group in collaboration with Pfizer and AstraZeneca, lead by Julian Cremer and Ross Irwin on FLOWR, a flow-based ligand generation approach, and highly sanitized benchmark dataset, SPINDR, for the SBDD community!

My colleague Michaela Wenzel at Chalmers is recruiting 2 PhD students in AI-assisted antibiotic discovery. Very interesting projects with a great supervisor! You can learn more about the positions (and apply by May 5th) here: www.chalmers.se/en/about-cha...

support open science:)

t.co/Nvjs1NTZFr

New blog post!

I've taken some time to write a blog post about classical machine learning and how I would explain it (given some prerequisites).

**Note**: I thought I would entirely finish up but I did not 😅, but I want to publish this anyway, **all types of feedback are welcome**.

Link below

rezarezvan.com/posts/claude...

First post here!
---

I originally wrote this post this summer, but I rushed it and never made a public announcement.

But now, I'm happy to present my BSc. thesis.
ClaudesLens, I've spent this week to try to explain what we did, hopefully it will be a good holiday read for you all :)

I usually use:

`DEVICE = torch.device("cuda" if torch.cuda.is_available() else "cpu")`

In all my projects as a global across the project, very handy