An SGLang-native post-training framework for RL scaling.
engineering
A fully open reproduction of DeepSeek-R1.
Building effective agents using the smolagents library.
learning
Course to get into large language models with roadmaps and Colab notebooks.