Free Documentation

Compute and compilers

cuda-guide, amd-hip, triton-tutorials, and apache-tvm establish execution, memory, kernel, and compiler concepts.

Distributed execution

pytorch-distributed and nccl-docs define APIs and collective semantics. See Module 07.

Serving and evaluation

vllm-docs, triton-inference-server, helm, and lm-eval-harness support deployment and measurement. pytorch-tutorials provides framework-level experiments across modules.

Documentation versions move. Canonical URLs and maintenance metadata live in data; source notes should record what was evaluated and when.