Triton-Puzzles and Tensor-Puzzles

These are complementary learning resources—Triton-Puzzles teaches the Triton compiler for GPU kernels while Tensor-Puzzles teaches PyTorch fundamentals, so learners typically progress from PyTorch understanding to optimized Triton kernel implementation.

Triton-Puzzles
65
Established
Tensor-Puzzles
46
Emerging
Maintenance 20/25
Adoption 10/25
Maturity 16/25
Community 19/25
Maintenance 0/25
Adoption 10/25
Maturity 16/25
Community 20/25
Stars: 2,338
Forks: 207
Downloads: —
Commits (30d): 11
Language: Jupyter Notebook
License: Apache-2.0
Stars: 3,976
Forks: 360
Downloads: —
Commits (30d): 0
Language: Jupyter Notebook
License: MIT
No Package No Dependents
Stale 6m No Package No Dependents

About Triton-Puzzles

gpu-mode/Triton-Puzzles

Puzzles for learning Triton

About Tensor-Puzzles

srush/Tensor-Puzzles

Solve puzzles. Improve your pytorch.

Comprised of 21 self-contained puzzles that isolate specific tensor operations, each solvable in a single line using only broadcasting, indexing, and arithmetic—no standard library functions allowed. Emphasizes mastery of NumPy/PyTorch's broadcasting semantics through constraint-based learning, with visual diagram feedback and property-based testing via Hypothesis. Targets PyTorch and NumPy users seeking deeper understanding of low-level tensor manipulation fundamentals.

Scores updated daily from GitHub, PyPI, and npm data. How scores work