Beginning today, I will try to tweet things that I am thinking about. I am fine with no comments or likes or engagement. In a way, I just want this to be my public " idea journal".

Personally I think more debates are needed on issues to expose their core merits, flaws, alternatives etc. We need more platforms that encourage civil debates kialo.com.

*⃣ Resource alert for people applying to CS PhD programs this cycle *⃣ cs-sop.org contains >60 example statements of purpose! It's made possible by the many generous submissions from new applicants, and new ones are always welcome! 😊

We recently released AircraftVerse, which is a dataset of over 27,000 diverse aircraft designs. Lots of opportunities for multimodal ML applications applied to cyber-physical systems! Paper: arxiv.org/abs/2306.05562 Dataset: doi.org/10.5281/zenodo… Github: github.com/SRI-CSL/Aircra…

People won't be so easily taken by "LLM ingenuity" if they stop thinking that they have a clue of what really is on the web.. 😛

Before trying to solve info retrieval using LLMs on the internet level, can we do it for the records from academic journals and conferences and arxiv. After all, they should be the least noisiest (???). #ICLR2024 #Neurips2023 #Arxiv

Whats up with the increase in bot activity lately ? Everyday I get 3-5 bots following or sending requests .. could we look into it Elon Musk

Glad that someone found our benchmark to be useful. Evaluations with strong counter factual examples should be the future. In TallyQA, we use human annotators to make sure of that.

Assuming a 10% down payment, Austinites now need an income of $149,000+ to “comfortably” afford a home. Up from $84,000 pre-Covid. About a 77% jump in just 4 years. As always Lance Lambert puts out a great news letter with amazing info.

Crowdsourcing decision-making/recommendation using AI doesnot always give high quality results, the end goal is a moving target. We are replacing "Slow"/"good" with "fast"/"OK". We need to either change how we live or change the design of AI that we depend on.

This seems kinda...radical? ASU makes its courses available to anyone for $25/course. After you take the class, if you want the grade you got added to an official transcript with a credit you can use, +$400. These are real college credits. 8 year olds are getting college credits!

Meet the game-changer: SO-100 🦾 Crafted by Rob Knight and Hugging Face 🤗 At 1/3 the cost and 2x the capabilities of our previous arms, it's the most accessible, high-performance robotic arm for $115. Easiest DIY at home! 1/🧵 Link and details in thread 👇

USCIS just denied my US green card application. claiming my work lacks impact "beyond that of Apple, Inc." -- even though they acknowledge that I am the Apple CTO. WTF? I have NEVER worked at Apple! I got my cs PhD, cofounded an AI startup, and raised $20M. Yet, after waiting

We made 5 challenges and if you score 47 points we'll offer you $500K/year + equity to join us at 🦥Unsloth AI! No experience or PhD needed. $400K - $500K/yr: Founding Engineer (47 points) $250K - $300K/yr: ML Engineer (32 points) Challenges: 1. Convert nf4 / BnB 4bit to

I just published Rethinking LLMs: A Personal Take medium.com/p/rethinking-l…

aha .. next will be a predetermined chain of buses .. aka "trains"

with this amount of ML papers we need better tools to do lit survey ( + avoid repeating papers) basically high fine grained retrieval performance ..... what tools/techniques do people use ? AFAIK google search / arxiv uses keywords or maybe some shallow text embeddings

🚨New preprint 🚨 Turning Down the Heat: A Critical Analysis of Min-p Sampling in Language Models We examine min-p sampling (ICLR 2025 oral) & find significant problems in all 4 lines of evidence: human eval, NLP evals, LLM-as-judge evals, community adoption claims 1/8