trainer

Example

For a complete list of all parameters and their detailed descriptions, please refer to the official PyTorch Lightning Trainer documentation: https://lightning.ai/docs/pytorch/stable/common/trainer.html

Note

Below is an example of usage. If a parameter comes from the internal implementation of TACE, it may not be the most up-to-date. For the latest parameters, please refer to the corresponding configuration files on GitHub. A complete list of parameters along with detailed explanations is provided.

trainer:
  # As long as your system has NVIDIA GPUs, the following configuration will automatically detect and use all available GPUs for training.
  # If you wish to control which GPUs are visible, you can do so by setting the environment variable like export CUDA_VISIBLE_DEVICES=0,2.
  _target_: lightning.Trainer
  # limit_train_batches: 0.1
  num_nodes: 1
  accelerator: auto
  devices: auto        # auto or list
  max_time: '90:00:00:00' # 90 days
  max_epochs: 1000
  min_epochs: 1        # null or int > 1
  precision: 64       # 64 > 32 > bf16-mixed > 16-mixed > 16-true
  # strategy: auto
  strategy:
    _target_: tace.utils.strategy.SimpleDDPStrategy  # Recommended for multi-GPU training.
                                                     # For single-GPU training, simply use `auto`.
  gradient_clip_val: 1.0
  enable_progress_bar: true # If you don't want to see the progress bar, you can turn it off
  log_every_n_steps: 10000

  # Generally, no modification is required
  enable_model_summary: true
  enable_checkpointing: true
  check_val_every_n_epoch: 1
  detect_anomaly: false
  inference_mode: false
  deterministic: false
  # benchmark: true