Model Parallelism

Tensor Slicing

A fundamental operation in tensor parallelism involving dividing a tensor along a specific dimension (e.g., row, column) to distribute it across multiple devices.

← Quay lại