{
  "gradient_accumulation_steps": "auto",
  "train_micro_batch_size_per_gpu": "auto",
  "prescale_gradients": false,
  "zero_allow_untested_optimizer": true,
  "optimizer": {
    "type": "AdamW",
    "params": {
      "lr": "auto",
      "eps": "auto",
      "betas": "auto",
      "weight_decay": "auto"
    }
  },
  "tensorboard": {
    "enabled": true,
    "output_path": "logs/",
    "job_name": "openbuddy_llama-7b-pt"
  },
  "zero_optimization": {
    "stage": 3,
    "contiguous_gradients": false,
    "allgather_bucket_size": 3e8,
    "reduce_bucket_size": 3e8,
    "overlap_comm": true,
    "reduce_scatter": true
  },
  "steps_per_print": 16,
  "gradient_clipping": 1.0,
  "wall_clock_breakdown": true,
  "bf16": {
    "enabled": true
  }
}