🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

It’s so exciting to see BioCLIP 2 demonstrates a biologically meaningful embedding space while only trained to distinguish species. Can’t wait to see more applications of BioCLIP 2 in solving real world problems. I’m attending #CVPR2025 in Nashville. Happy to chat about it!

Quizzing BioClip about an animal/plant has been another fun activity we do at a zoo/garden. Most of the time, it does get things right! Now check out BioClip 2, with much stronger performance and nice properties!

📢 Introducing AutoSDT, a fully automatic pipeline that collects data-driven scientific coding tasks at scale! We use AutoSDT to collect AutoSDT-5K, enabling open co-scientist models that rival GPT-4o on ScienceAgentBench! Thread below ⬇️ (1/n)

we need out-of-the-box thinking for agentic web

Are you at #CVPR2025? RoboSpatial Oral is today! 📅 June 14 (Sat) | 🕐 1:00 PM | 📍Oral Session 4B @ ExHall A2

This is kinda sad. Putting the final stamp on the death notice of pre-training kind of sad.

Here is my best-effort account of what may be happening: - Fine-tuning as a term peaked in the BERT times. In LLMs, it's mostly used in the SFT context. - Fine-tuning for task specialization has been fading away, partly because most frontier LLMs are closed-source and

Wrapped up Stanford CS336 (Language Models from Scratch), taught with an amazing team Tatsunori Hashimoto Marcel Rød Neil Band Rohith Kuditipudi. Researchers are becoming detached from the technical details of how LMs work. In CS336, we try to fix that by having students build everything: