Papers I’ve read, with brief notes on what I found useful.

Why keep a reading list?

In the age of AI, I’ve noticed that I have less patience for reading papers closely. But the devil is in the details. I keep this list to get back into the habit of reading carefully and to develop my research taste.

I look for papers that have won awards or are well regarded in their fields, papers relevant to my research, and recent work that catches my interest.

Blog

The flavor of the bitter lesson for computer vision - Vincent Sitzmann

Scalable training truly matters! Recently, GPT-6 Astra has shown strong potential for solving embodied intelligence problems. Maybe treating 3D as a training objective to learn physics directly from video makes much more sense.

Recent

[2609.37725] Context Language Models

target question: context manage

AI manage context and online RL, I think it’s native and reasonable. But has no one done it before?

Classical