// the find
MorvanZhou/tutorials
机器学习相关教程
A large, Chinese-language collection of short, single-purpose scripts teaching ML and Python basics — numpy/pandas, matplotlib, threading/multiprocessing, tkinter, and neural nets in TensorFlow, Theano, Keras, PyTorch, and sklearn. It's the companion code to Zhou Mofan's (莫烦Python) video tutorial series, aimed squarely at Chinese-speaking beginners going from zero to a first neural net.
Each concept gets its own minimal runnable file (e.g. tensorflow10_def_add_layer.py) rather than one sprawling notebook, so you can diff two examples and see exactly what changed. Coverage is unusually broad for a beginner repo — it doesn't stop at ML, it also covers threading, multiprocessing, git, and Linux basics that most ML tutorials skip. Code pairs with a matching video series, which is a real asset for people who learn better watching someone run it than reading a README.
Dead since December 2020 — the TensorFlow examples use tf.Session and feed_dict (tensorflow6_session.py, tensorflow8_feeds.py), which is TF1-only and won't run against any TF2 install without a rewrite; Theano itself is unmaintained upstream. No requirements.txt or environment file anywhere, so pinning compatible versions of numpy/pandas/sklearn/TF is left entirely to the reader. Everything is Chinese-only, comments included, which is a hard wall for non-Chinese-speaking developers. The repo checks in binary junk (MNIST .gz files, tfevents logs) instead of gitignoring them, and there are no tests or CI, so nothing catches when an example silently rots.