compute_environment: LOCAL_MACHINE
distributed_type: DEEPSPEED
num_processes: 4
machine_rank: 0
num_machines: 1
gpu_ids: all
mixed_precision: fp16
deepspeed_config:
  gradient_accumulation_steps: 4
  gradient_clipping: 1.0
  offload_optimizer_device: cpu  # 优化器状态卸载到CPU
  offload_param_device: cpu      # 参数卸载到CPU
  zero3_init_flag: true
  zero_stage: 3  # ZeRO-3