remi 🍉

did u know that everyone in America is female youtu.be/DR8fHT31aPA?...

im an alien

Ten simple rules for developing good reading habits during graduate school and beyond To me, the most important are: Read often, read broadly (incl. older papers and outside your field), and learn to read some papers in detail and others more superficially (and quickly)

What's your favorite anime and why? #anime #otaku #japan #animesky

Wanna bit of writing advice? I try to avoid using ‘this’ in the sense of, for example, “one of the features of this case” unless it’s very close to the referent. It puts cognitive load on the reader to dredge up what it refers to. Better to say “one of the features of the case of the red dog”.

But before I get to the reasoning model space... if you are looking to do some focused offline reading this weekend, I just re-compiled my take on the "noteworthy AI research papers of 2024" into one PDF-export-friendly 47-page mega-post with TOC and all: sebastianraschka.com/blog/2025/ll...

Behind the scenes with what its like to build language models and pursue (hopefully) cutting edge AI research Interviewing OLMo 2 leads: Open secrets of training language models What we have learned and are going to do next. YouTube: https://buff.ly/40IlSFF Podcast / notes:

Dear algorithm, please connect me with: ✅book lovers ✅science lovers ✅data science and AI lovers ✅anime lovers

Moonshot AI, a startup team from China, has today released another multi-modal model that reaches the level of OpenAI's o1: Kimi k1.5. Currently, only the Chinese version is available, but the English version website will be launched soon. Project: www.moonshot.cn Key Technologies of Kimi k1.5

Ace your NLP Interview by neuralnetwork trite-song-d6a.notion.site/Ace-your-NLP...

Understanding LSTM Networks Long Short Term Memory networks – called “LSTMs” – are a special kind of RNN, capable of learning long-term dependencies. They were introduced by Hochreiter & Schmidhuber, and were refined and popularized by many people in following work. colah.github.io/posts/2015-0...

my friend committed suicide on Monday. if only I told him how much he meant to me..... he wouldn't have done it maybe.. I failed to be a good friend. I hate this.

Playing around with AI agents, and I reckon Gradio spaces on the hub make the perfect tools. - super easy to connect your agents to a bunch of useful tools and apps. - find a Space you like on Hugging Face Hub or make your own with Gradio. - link it up with smolagents. 🧵

This looks very cool: neuroevolutionbook.com

This feels interesting. I have not tried it yet. OmniThink YOU can create a long-form article that is not just longer but also richer, more insightful, and packed with real-world value. 📄 Paper: arxiv.org/abs/2501.09751 🎥 Demo: www.modelscope.cn/studios/iic/... 💻 Code: github.com/zjunlp/OmniT...

Excited to share that today our paper recommender platform www.scholar-inbox.com has reached 20k users! We hope to reach 100k by the end of the year.. Lots of new features are being worked on currently and rolled out soon.

how to cope with the death of a very close friend, by suicide?

@dabeaz.com's courses ("the kind of courses you wish you could take in graduate school, but without all of the exams and psychological trauma") look like fun. If I had the luxury of a sabbatical/free time I would sign up! www.dabeaz.com/courses.html

🚀 Skyrocketing！ 🚀 (200+ new stars) 📦 srbhr / Resume-Matcher ⭐ 7,210 (+486) 🗒 Python Resume Matcher is an open source, free tool to improve your resume. It works by using AI, Reader LLMs, to compare and rank resumes with job descriptions.

I think there’s a misconception out there that good writing means phrasing things in a way nobody ever has or would—that if a sentence feels familiar in the mind it fails. The opposite is true. Good writing is phrasing things in a way everyone has always wanted to and recognizes in their hearts.

Good pointer to a paper I might have otherwise overlooked in my current reading rotation but for the clear eval implications Also reminder you should probably be following Ehud on here for a scoop on eval and nlg stuff

If you want to understand that no, Mark Zuckerman isn't some harmless money-grubbing schmuck, go read the truly shocking "Mindf*ck: Cambridge Analytica And The Plot To Break America" by whistleblower Christopher Wylie. You will have trouble sleeping after reading this book.

Recent DSLC.io club meetings: 🟢 Practical Deep Learning for Coders: Karras et al youtu.be/K6LDsL4T5Bk Visit dslc.video for hours of new #DataScience videos every week! #dataBS #PyData #DeepLearning #AI

Adobe's TransPixar - it generates video with transparent background.

How to write: Stop not writing. Keep your butt in the chair. Write really bad small sections of the whole—passages, moments, episodes, memories—til you have an incredibly shitty 1st draft. Then take out the boring parts, the lies and pretensions. Then write a better 2nd draft.

Before diving into the technical aspects of my #DeepLearning course, our next lecture will explore Ethics, Security, Privacy in ML, and the Future of Work. Course slides: www.eecs.yorku.ca/~kosta/Cours...

Carpenter and Grossberg paper that changed things for me:

"To do the useful thing, to say the courageous thing, to contemplate the beautiful thing: that is enough for one man's life." -- T.S. Eliot, who died #OTD 1965

Questionable practices in machine learning arxiv.org/pdf/2407.12220 #machinelearning #artificialintelligence #datascience

People make time for what they want. Read that again and again and again.

John Berger, most beloved, passed away on January 2, 2017. "What reconciles me to my own death more than anything else is the image of a place: a place where your bones and mine are buried...With you I can imagine a place where to be phosphate of calcium is enough.” #everynightapoem #ofsorts

Excited to finally share this work w/ @suryaganguli.bsky.social Tl;dr: we find the first closed-form analytical theory that replicates the outputs of the very simplest diffusion models, with median pixel wise r^2 values of 90%+. arxiv.org/abs/2412.20292

Remember when we used to go and look at all the different folks personal and blog sites and get inspiration for features and easter eggs for our own projects? Back in the day before we all made the same site over and over? Miss those days.

Facts

🚨 New paper (with @eitanwagner.bsky.social @joebarnby.bsky.social and Omri Abend) on how we (should) evaluate Theory of Mind in Large Language Models. While recent work claims LLMs have ToM capabilities, we're missing crucial aspects from cognitive science. Here's why this matters 🧵

[23/30] 76 Likes, 2 Comments, 2 Posts 2412.09624, cs․CV | cs․RO, 12 Dec 2024 🆕GenEx: Generating an Explorable World Taiming Lu, Tianmin Shu, Junfei Xiao, Luoxin Ye, Jiahao Wang, Cheng Peng, Chen Wei, Daniel Khashabi, Rama Chellappa, Alan Yuille, Jieneng Chen

For anyone interested in fine-tuning or aligning LLMs, I’m running this free and open course called smol course. It’s not a big deal, it’s just smol. 🧵>>

🧵 Today with @polymathicai.bsky.social and others we're releasing two massive datasets that span dozens of fields - from bacterial growth to supernova! We want this to enable multi-disciplinary foundation model research.

Check out this link for my 6100+ (and counting!) #ComputerVision slides 😅 www.eecs.yorku.ca/~kosta/Cours...

New starter packs are narrowing down their topics to the point to which I have no idea what they are talking about (but they sure seem to). This is great 💪!

I think spreading hate for a mistake is really gross and out of touch with human nature. Do these people not make mistakes from time to time? Are they born perfect? I pray for the times where we don't cancel someone for a mistake, let alone a whole company...