DecryptPrompt

Why we included this project

DecryptPrompt is a Chinese-language study index for anyone trying to keep up with prompt engineering and LLM research. It gathers open models and benchmark leaderboards, training and inference frameworks, SFT, RLHF, and pretraining datasets, and AIGC applications into one place, so you can find what actually exists before picking a stack. The real draw is the long-running 'DecryptPrompt' blog series, which reads papers in order, from tuning-free prompting and instruction tuning through chain-of-thought, RLHF alignment, and agent designs like ReAct and Toolformer, explaining how each technique builds on the last. Researchers, students, and engineers who want the theory behind common practices will find it a useful guided path. It is a living collection of links, summaries, and commentary rather than runnable software, so treat it as a map and check details against the papers it points to.