This is your work, valued

Deep learning educator and systems builder; created D2L textbook adopted by 500+ universities, now exploring multimodal AI at Boson.

Interactive deep learning textbooks & pedagogyMultimodal foundation models & audio-textComputer vision & face detection systemsMXNet framework & distributed deep learningVideo processing & media toolsML benchmarking & performance optimization

inferred from the public footprint

1
Wed
Thu
Fri
Sat
Sun
Mon
Yesterday
Active 1d ago
Palo Alto, CA

Mu Li

Top 14%
@mli

At Boson AI, Li creates structured learning materials and hands-on tools for the ML community, from mli/paper-reading for dissecting foundational research to mli/transformers-benchmarks measuring real-world GPU performance across hardware.

Paper Reading. I recorded detailed video explanations of deep learning research papers.

34k

Dive into Deep Learning. I wrote a free deep learning textbook with runnable code for everyone.

79k

AutoCut. I made a tool that cuts videos by editing text captions.

7.8k

transformers-benchmarks. real Transformer TeraFLOPS on various GPUs

912

Dive into Deep Learning Compiler. I wrote a free course teaching how AI models get optimized for speed.

649

Mintlify Starter Kit. I made a documentation starter template you can deploy and customize instantly.

27

Nyström. I implemented a faster version of Nyström approximation for large datasets.

16

D2L One-Day. I made a one-day deep learning bootcamp with a hackathon.

154

Powerball. I wrote MATLAB code for the powerball statistical method.

3

D2L 1Day. I made notebooks that teach deep learning fundamentals in one day.

60

Parsa. I built a tool that splits large graphs across multiple computers.

10

15721 TA. I wrote grading scripts for a CMU database course.

12

1Day Notebooks. I built notebooks that teach deep learning fundamentals in one day.

75

AutoGluon Docs. I'm rebuilding documentation for AutoGluon's machine learning toolkit.

14

MXNet Benchmark. I wrote benchmark scripts for testing MXNet performance.

9

Deps. I made a tool that shows you what your project depends on.

2

D2L TVM Colab. I made interactive notebooks teaching machine learning compiler optimization.

9

Lanyon Blog. I made a personal blog using Jekyll and the Lanyon theme.

5

Gluon. I made tutorials that teach AI model building with working code examples.

327

CTR Data. I collected a sample click-through rate dataset for model testing.

6

ResNet-v2. I reproduced ResNet-v2 image recognition models with better accuracy.

573

DLMark. I built an open benchmark for measuring deep learning system performance.

18

MXNet Docs. I'm rebuilding the documentation website for MXNet.

13

DeepNumPy Docs. I wrote documentation for MXNet's NumPy-compatible branch.

23