compute_environment: LOCAL_MACHINE distributed_type: MULTI_GPU mixed_precision: bf16 num_processes: 8 num_machines: 1 machine_rank: 0 gpu_ids: all rdzv_backend: static same_network: true tpu_use_cluster: false tpu_use_sudo: false use_cpu: false debug: false enable_cpu_affinity: false main_training_function: main downcast_bf16: 'no'