util/deepspeed_configs/accelerate_config_machine_multi.yaml

compute_environment: LOCAL_MACHINE
distributed_type: DEEPSPEED
deepspeed_config:
 deepspeed_config_file: util/deepspeed_configs/zero_stage2_config.json
 deepspeed_hostfile: util/deepspeed_configs/hostfile.txt
fsdp_config: {}
machine_rank: 0
main_process_ip: 100.64.24.6
main_process_port: 12343
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