Ahmad Mustafa Anis

Happy to share my first 1st author paper In this paper, we explore that Vision-Language Models (Clip/Siglip) fail in understanding basic Image Transforms. [1/4]

Craft is truly caring about what you make. Software or sushi, either way it is about absolute dedication to creating something people want. The Lesson of Craft youtu.be/B93mNF2hxF0

Multiplicity is an interesting impact metric. If you are an academic or researcher, how many mentees choose to pursue research or academia? If you are a startup founder, how many team members grew enough to become founders themselves? Most important: how many surpassed you!

Today, we at @OpenAI achieved a milestone that many considered years away: gold medal-level performance on the 2025 IMO with a general reasoning LLM—under the same time limits as humans, without tools. As remarkable as that sounds, it’s even more significant than the headline 🧵

Be sure to join our Regional Asia group next week on July 23rd as they sit down with @fengyao1909 for an insightful session on "DenseMixer: Improving MoE Post-Training with Precise Router Gradient" Thanks to @KanwalMehreen2 and @AhmadMustafaAn1 for organizing this session! 👏

What makes some researchers truly exceptional versus merely good? This came up in our recent @Cohere_Labs summer school session. I actually think is one of the most important questions you can ask.

Every few years I think you should jump into a new deep-end by doing something that pushes you out of your comfort zone. However, people rarely do because it is easier to stick with what is known. I think this is true in research, life partners and big life decisions.

So much research is being done about LLMs that it's hard to stay on top of the literature. To help with this, I've made a list of all the most important papers from the past 8 years: rtmccoy.com/pubs/ I hope you enjoy!

This is my lecture from 2 months ago at @Cornell “How do I increase my output?” One natural answer is "I will just work a few more hours." Working longer can help, but eventually you hit a physical limit. A better question is, “How do I increase my output without increasing…

Thinking Machines Lab exists to empower humanity through advancing collaborative general intelligence. We're building multimodal AI that works with how you naturally interact with the world - through conversation, through sight, through the messy way we collaborate. We're…

~400 people have joined us for the Research Mentorship Session by @sarahookr at @Cohere_Labs Machine Learning Summer School. So much great advice for early career and aspiring researchers 🔥🔥❤️

We will be hosting @sarahookr on the last day (Monday) of our Machine Learning Summer School at @Cohere_Labs for an interactive session on Research Mentorship. Add the event to your calendar: sites.google.com/cohere.com/coh…

Join us in ~6 hours to learn about Distributed Training in Machine Learning. Add event to your calendar: sites.google.com/cohere.com/coh…

clickbait title

such a stacked author list. such a great paper

Had a lot of fun at this session. As usual, I’m behind on time and could only cover 50% of material. 😅 Maybe I should do a short YT course on this? 🤔

Video understanding isn't just recognizing —it demands reasoning across thousands of frames. Meet Long-RL🚀 Highlights: 🧠 Dataset: LongVideo-Reason — 52K QAs with reasoning. ⚡ System: MR-SP - 2.1× faster RL for long videos. 📈 Scalability: Hour-long videos (3,600 frames) RL…

🚀New Paper & Dataset out! ImplicitQA: Going Beyond Frames Toward Implicit Video Reasoning 🎬 We challenge VideoQA models to reason about what’s not visible - like implied actions, hidden causality, and off-screen context. Paper: arxiv.org/abs/2506.21742 🧵👇 #VideoQA #AI #LLM