cards/machine-learning/practice/cuda-memory.md

Data parallelism in pytorch