← OnlineNOTES / ARCHIVE
NOTES / 2026
Keep tuning.
Newest first · scroll down to look back
2026-07 / OBSERVATION
A readable introduction to reinforcement learning
From ChatGPT and AlphaGo to DeepSeek-R1: how reward and trial and error complement large training datasets.
Read Chinese original ↗Wheel / swipe / arrow keys