scripts/accelerate_configs/multi_node_example.yaml

compute_environment: LOCAL_MACHINE
distributed_type: DEEPSPEED
deepspeed_config:
 deepspeed_config_file: scripts/accelerate_configs/zero2.json
 deepspeed_hostfile: /remote-home1/yeyang/Open-Sora-Plan/scripts/accelerate_configs/hostfile
fsdp_config: {}
machine_rank: 0
main_process_ip: 10.10.10.55
main_process_port: 29501
main_training_function: main
num_machines: 2
num_processes: 16
rdzv_backend: static
same_network: true
tpu_env: []
tpu_use_cluster: false
tpu_use_sudo: false
use_cpu: false