Enter to send·Shift+Enter new line
tensor_parallel_size: Optional[int] = 1
The number of GPUs to use for distributed execution with tensor parallelism.
