Shortcuts

Trainer

Once you’ve organized your PyTorch code into a LightningModule, the Trainer automates everything else.


This abstraction achieves the following:

  1. You maintain control over all aspects via PyTorch code without an added abstraction.

  2. The trainer uses best practices embedded by contributors and users from top AI labs such as Facebook AI Research, NYU, MIT, Stanford, etc…

  3. The trainer allows overriding any key part that you don’t want automated.



Basic use

This is the basic use of the trainer:

model = MyLightningModule()

trainer = Trainer()
trainer.fit(model, train_dataloader, val_dataloader)

Under the hood

Under the hood, the Lightning Trainer handles the training loop details for you, some examples include:

  • Automatically enabling/disabling grads

  • Running the training, validation and test dataloaders

  • Calling the Callbacks at the appropriate times

  • Putting batches and computations on the correct devices

Here’s the pseudocode for what the trainer does under the hood (showing the train loop only)

# put model in train mode
model.train()
torch.set_grad_enabled(True)

losses = []
for batch in train_dataloader:
    # calls hooks like this one
    on_train_batch_start()

    # train step
    loss = training_step(batch)

    # clear gradients
    optimizer.zero_grad()

    # backward
    loss.backward()

    # update parameters
    optimizer.step()

    losses.append(loss)

Trainer in Python scripts

In Python scripts, it’s recommended you use a main function to call the Trainer.

from argparse import ArgumentParser


def main(hparams):
    model = LightningModule()
    trainer = Trainer(accelerator=hparams.accelerator, devices=hparams.devices)
    trainer.fit(model)


if __name__ == "__main__":
    parser = ArgumentParser()
    parser.add_argument("--accelerator", default=None)
    parser.add_argument("--devices", default=None)
    args = parser.parse_args()

    main(args)

So you can run it like so:

python main.py --accelerator 'gpu' --devices 2

Note

Pro-tip: You don’t need to define all flags manually. Lightning can add them automatically

from argparse import ArgumentParser


def main(args):
    model = LightningModule()
    trainer = Trainer.from_argparse_args(args)
    trainer.fit(model)


if __name__ == "__main__":
    parser = ArgumentParser()
    parser = Trainer.add_argparse_args(parser)
    args = parser.parse_args()

    main(args)

So you can run it like so:

python main.py --accelerator 'gpu' --devices 2 --max_steps 10 --limit_train_batches 10 --any_trainer_arg x

Note

If you want to stop a training run early, you can press “Ctrl + C” on your keyboard. The trainer will catch the KeyboardInterrupt and attempt a graceful shutdown, including running accelerator callback on_train_end to clean up memory. The trainer object will also set an attribute interrupted to True in such cases. If you have a callback which shuts down compute resources, for example, you can conditionally run the shutdown logic for only uninterrupted runs.


Validation

You can perform an evaluation epoch over the validation set, outside of the training loop, using validate(). This might be useful if you want to collect new metrics from a model right at its initialization or after it has already been trained.

trainer.validate(model=model, dataloaders=val_dataloaders)

Testing

Once you’re done training, feel free to run the test set! (Only right before publishing your paper or pushing to production)

trainer.test(dataloaders=test_dataloaders)

Reproducibility

To ensure full reproducibility from run to run you need to set seeds for pseudo-random generators, and set deterministic flag in Trainer.

Example:

from pytorch_lightning import Trainer, seed_everything

seed_everything(42, workers=True)
# sets seeds for numpy, torch and python.random.
model = Model()
trainer = Trainer(deterministic=True)

By setting workers=True in seed_everything(), Lightning derives unique seeds across all dataloader workers and processes for torch, numpy and stdlib random number generators. When turned on, it ensures that e.g. data augmentations are not repeated across workers.


Trainer flags

accelerator

Supports passing different accelerator types ("cpu", "gpu", "tpu", "ipu", "auto") as well as custom accelerator instances.

# CPU accelerator
trainer = Trainer(accelerator="cpu")

# Training with GPU Accelerator using 2 GPUs
trainer = Trainer(devices=2, accelerator="gpu")

# Training with TPU Accelerator using 8 tpu cores
trainer = Trainer(devices=8, accelerator="tpu")

# Training with GPU Accelerator using the DistributedDataParallel strategy
trainer = Trainer(devices=4, accelerator="gpu", strategy="ddp")

Note

The "auto" option recognizes the machine you are on, and selects the respective Accelerator.

# If your machine has GPUs, it will use the GPU Accelerator for training
trainer = Trainer(devices=2, accelerator="auto")

You can also modify hardware behavior by subclassing an existing accelerator to adjust for your needs.

Example:

class MyOwnAcc(CPUAccelerator):
    ...

Trainer(accelerator=MyOwnAcc())

Note

If the devices flag is not defined, it will assume devices to be "auto" and fetch the auto_device_count from the accelerator.

# This is part of the built-in `CUDAAccelerator`
class CUDAAccelerator(Accelerator):
    """Accelerator for GPU devices."""

    @staticmethod
    def auto_device_count() -> int:
        """Get the devices when set to auto."""
        return torch.cuda.device_count()


# Training with GPU Accelerator using total number of gpus available on the system
Trainer(accelerator="gpu")

Warning

Passing training strategies (e.g., "ddp") to accelerator has been deprecated in v1.5.0 and will be removed in v1.7.0. Please use the strategy argument instead.

accumulate_grad_batches


Accumulates grads every k batches or as set up in the dict. Trainer also calls optimizer.step() for the last indivisible step number.

# default used by the Trainer (no accumulation)
trainer = Trainer(accumulate_grad_batches=1)

Example:

# accumulate every 4 batches (effective batch size is batch*4)
trainer = Trainer(accumulate_grad_batches=4)

# no accumulation for epochs 1-4. accumulate 3 for epochs 5-10. accumulate 20 after that
trainer = Trainer(accumulate_grad_batches={5: 3, 10: 20})

amp_backend


Use PyTorch AMP (‘native’), or NVIDIA apex (‘apex’).

# using PyTorch built-in AMP, default used by the Trainer
trainer = Trainer(amp_backend="native")

# using NVIDIA Apex
trainer = Trainer(amp_backend="apex")

auto_scale_batch_size


Automatically tries to find the largest batch size that fits into memory, before any training.

# default used by the Trainer (no scaling of batch size)
trainer = Trainer(auto_scale_batch_size=None)

# run batch size scaling, result overrides hparams.batch_size
trainer = Trainer(auto_scale_batch_size="binsearch")

# call tune to find the batch size
trainer.tune(model)

auto_select_gpus


If enabled and devices is an integer, pick available GPUs automatically. This is especially useful when GPUs are configured to be in “exclusive mode”, such that only one process at a time can access them.

Example:

# no auto selection (picks first 2 GPUs on system, may fail if other process is occupying)
trainer = Trainer(accelerator="gpu", devices=2, auto_select_gpus=False)

# enable auto selection (will find two available GPUs on system)
trainer = Trainer(accelerator="gpu", devices=2, auto_select_gpus=True)

# specifies all GPUs regardless of its availability
Trainer(accelerator="gpu", devices=-1, auto_select_gpus=False)

# specifies all available GPUs (if only one GPU is not occupied, uses one gpu)
Trainer(accelerator="gpu", devices=-1, auto_select_gpus=True)

auto_lr_find


Runs a learning rate finder algorithm (see this paper) when calling trainer.tune(), to find optimal initial learning rate.

# default used by the Trainer (no learning rate finder)
trainer = Trainer(auto_lr_find=False)

Example:

# run learning rate finder, results override hparams.learning_rate
trainer = Trainer(auto_lr_find=True)

# call tune to find the lr
trainer.tune(model)

Example:

# run learning rate finder, results override hparams.my_lr_arg
trainer = Trainer(auto_lr_find='my_lr_arg')

# call tune to find the lr
trainer.tune(model)

benchmark


The value (True or False) to set torch.backends.cudnn.benchmark to. The value for torch.backends.cudnn.benchmark set in the current session will be used (False if not manually set). If deterministic is set to True, this will default to False. You can read more about the interaction of torch.backends.cudnn.benchmark and torch.backends.cudnn.deterministic here

Setting this flag to True can increase the speed of your system if your input sizes don’t change. However, if they do, then it might make your system slower. The CUDNN auto-tuner will try to find the best algorithm for the hardware when a new input size is encountered. This might also increase the memory usage. Read more about it here.

Example:

# Will use whatever the current value for torch.backends.cudnn.benchmark, normally False
trainer = Trainer(benchmark=None)  # default

# you can overwrite the value
trainer = Trainer(benchmark=True)

deterministic


This flag sets the torch.backends.cudnn.deterministic flag. Might make your system slower, but ensures reproducibility. Also sets $HOROVOD_FUSION_THRESHOLD=0.

For more info check PyTorch docs.

Example:

# default used by the Trainer
trainer = Trainer(deterministic=False)

callbacks


Add a list of Callback. Callbacks run sequentially in the order defined here with the exception of ModelCheckpoint callbacks which run after all others to ensure all states are saved to the checkpoints.

# a list of callbacks
callbacks = [PrintCallback()]
trainer = Trainer(callbacks=callbacks)

Example:

from pytorch_lightning.callbacks import Callback

class PrintCallback(Callback):
    def on_train_start(self, trainer, pl_module):
        print("Training is started!")
    def on_train_end(self, trainer, pl_module):
        print("Training is done.")

Model-specific callbacks can also be added inside the LightningModule through configure_callbacks(). Callbacks returned in this hook will extend the list initially given to the Trainer argument, and replace the trainer callbacks should there be two or more of the same type. ModelCheckpoint callbacks always run last.

check_val_every_n_epoch


Check val every n train epochs.

Example:

# default used by the Trainer
trainer = Trainer(check_val_every_n_epoch=1)

# run val loop every 10 training epochs
trainer = Trainer(check_val_every_n_epoch=10)

default_root_dir


Default path for logs and weights when no logger or pytorch_lightning.callbacks.ModelCheckpoint callback passed. On certain clusters you might want to separate where logs and checkpoints are stored. If you don’t then use this argument for convenience. Paths can be local paths or remote paths such as s3://bucket/path or ‘hdfs://path/’. Credentials will need to be set up to use remote filepaths.

# default used by the Trainer
trainer = Trainer(default_root_dir=os.getcwd())

devices

Number of devices to train on (int), which devices to train on (list or str), or "auto". It will be mapped to either gpus, tpu_cores, num_processes or ipus, based on the accelerator type ("cpu", "gpu", "tpu", "ipu", "auto").

# Training with CPU Accelerator using 2 processes
trainer = Trainer(devices=2, accelerator="cpu")

# Training with GPU Accelerator using GPUs 1 and 3
trainer = Trainer(devices=[1, 3], accelerator="gpu")

# Training with TPU Accelerator using 8 tpu cores
trainer = Trainer(devices=8, accelerator="tpu")

Tip

The "auto" option recognizes the devices to train on, depending on the Accelerator being used.

# If your machine has GPUs, it will use all the available GPUs for training
trainer = Trainer(devices="auto", accelerator="auto")

# Training with CPU Accelerator using 1 process
trainer = Trainer(devices="auto", accelerator="cpu")

# Training with TPU Accelerator using 8 tpu cores
trainer = Trainer(devices="auto", accelerator="tpu")

# Training with IPU Accelerator using 4 ipus
trainer = Trainer(devices="auto", accelerator="ipu")

Note

If the devices flag is not defined, it will assume devices to be "auto" and fetch the auto_device_count from the accelerator.

# This is part of the built-in `CUDAAccelerator`
class CUDAAccelerator(Accelerator):
    """Accelerator for GPU devices."""

    @staticmethod
    def auto_device_count() -> int:
        """Get the devices when set to auto."""
        return torch.cuda.device_count()


# Training with GPU Accelerator using total number of gpus available on the system
Trainer(accelerator="gpu")

enable_checkpointing


By default Lightning saves a checkpoint for you in your current working directory, with the state of your last training epoch, Checkpoints capture the exact value of all parameters used by a model. To disable automatic checkpointing, set this to False.

# default used by Trainer, saves the most recent model to a single checkpoint after each epoch
trainer = Trainer(enable_checkpointing=True)

# turn off automatic checkpointing
trainer = Trainer(enable_checkpointing=False)

You can override the default behavior by initializing the ModelCheckpoint callback, and adding it to the callbacks list. See Saving and Loading Checkpoints for how to customize checkpointing.

from pytorch_lightning.callbacks import ModelCheckpoint

# Init ModelCheckpoint callback, monitoring 'val_loss'
checkpoint_callback = ModelCheckpoint(monitor="val_loss")

# Add your callback to the callbacks list
trainer = Trainer(callbacks=[checkpoint_callback])

fast_dev_run


Runs n if set to n (int) else 1 if set to True batch(es) to ensure your code will execute without errors. This applies to fitting, validating, testing, and predicting. This flag is only recommended for debugging purposes and should not be used to limit the number of batches to run.

# default used by the Trainer
trainer = Trainer(fast_dev_run=False)

# runs only 1 training and 1 validation batch and the program ends
trainer = Trainer(fast_dev_run=True)
trainer.fit(...)

# runs 7 predict batches and program ends
trainer = Trainer(fast_dev_run=7)
trainer.predict(...)

This argument is different from limit_{train,val,test,predict}_batches because side effects are avoided to reduce the impact to subsequent runs. These are the changes enabled:

  • Sets Trainer(max_epochs=1).

  • Sets Trainer(max_steps=...) to 1 or the number passed.

  • Sets Trainer(num_sanity_val_steps=0).

  • Sets Trainer(val_check_interval=1.0).

  • Sets Trainer(check_every_n_epoch=1).

  • Disables all loggers.

  • Disables passing logged metrics to loggers.

  • The ModelCheckpoint callbacks will not trigger.

  • The EarlyStopping callbacks will not trigger.

  • Sets limit_{train,val,test,predict}_batches to 1 or the number passed.

  • Disables the Tuner.

  • If using the CLI, the configuration file is not saved.

gpus

Warning

gpus=x has been deprecated in v1.7 and will be removed in v2.0. Please use accelerator='gpu' and devices=x instead.


  • Number of GPUs to train on (int)

  • or which GPUs to train on (list)

  • can handle strings

# default used by the Trainer (ie: train on CPU)
trainer = Trainer(gpus=None)

# equivalent
trainer = Trainer(gpus=0)

Example:

# int: train on 2 gpus
trainer = Trainer(gpus=2)

# list: train on GPUs 1, 4 (by bus ordering)
trainer = Trainer(gpus=[1, 4])
trainer = Trainer(gpus='1, 4') # equivalent

# -1: train on all gpus
trainer = Trainer(gpus=-1)
trainer = Trainer(gpus='-1') # equivalent

# combine with num_nodes to train on multiple GPUs across nodes
# uses 8 gpus in total
trainer = Trainer(gpus=2, num_nodes=4)

# train only on GPUs 1 and 4 across nodes
trainer = Trainer(gpus=[1, 4], num_nodes=4)
See Also:

gradient_clip_val


Gradient clipping value

  • 0 means don’t clip.

# default used by the Trainer
trainer = Trainer(gradient_clip_val=0.0)

limit_train_batches


How much of training dataset to check. Useful when debugging or testing something that happens at the end of an epoch.

# default used by the Trainer
trainer = Trainer(limit_train_batches=1.0)

Example:

# default used by the Trainer
trainer = Trainer(limit_train_batches=1.0)

# run through only 25% of the training set each epoch
trainer = Trainer(limit_train_batches=0.25)

# run through only 10 batches of the training set each epoch
trainer = Trainer(limit_train_batches=10)

limit_test_batches


How much of test dataset to check.

# default used by the Trainer
trainer = Trainer(limit_test_batches=1.0)

# run through only 25% of the test set each epoch
trainer = Trainer(limit_test_batches=0.25)

# run for only 10 batches
trainer = Trainer(limit_test_batches=10)

In the case of multiple test dataloaders, the limit applies to each dataloader individually.

limit_val_batches


How much of validation dataset to check. Useful when debugging or testing something that happens at the end of an epoch.

# default used by the Trainer
trainer = Trainer(limit_val_batches=1.0)

# run through only 25% of the validation set each epoch
trainer = Trainer(limit_val_batches=0.25)

# run for only 10 batches
trainer = Trainer(limit_val_batches=10)

# disable validation
trainer = Trainer(limit_val_batches=0)

In the case of multiple validation dataloaders, the limit applies to each dataloader individually.

log_every_n_steps


How often to add logging rows (does not write to disk)

# default used by the Trainer
trainer = Trainer(log_every_n_steps=50)
See Also:

logger


Logger (or iterable collection of loggers) for experiment tracking. A True value uses the default TensorBoardLogger shown below. False will disable logging.

from pytorch_lightning.loggers import TensorBoardLogger

# default logger used by trainer
logger = TensorBoardLogger(save_dir=os.getcwd(), version=1, name="lightning_logs")
Trainer(logger=logger)

max_epochs


Stop training once this number of epochs is reached

# default used by the Trainer
trainer = Trainer(max_epochs=1000)

If both max_epochs and max_steps aren’t specified, max_epochs will default to 1000. To enable infinite training, set max_epochs = -1.

min_epochs


Force training for at least these many epochs

# default used by the Trainer
trainer = Trainer(min_epochs=1)

max_steps


Stop training after this number of global steps. Training will stop if max_steps or max_epochs have reached (earliest).

# Default (disabled)
trainer = Trainer(max_steps=-1)

# Stop after 100 steps
trainer = Trainer(max_steps=100)

If max_steps is not specified, max_epochs will be used instead (and max_epochs defaults to 1000 if max_epochs is not specified). To disable this default, set max_steps = -1.

min_steps


Force training for at least this number of global steps. Trainer will train model for at least min_steps or min_epochs (latest).

# Default (disabled)
trainer = Trainer(min_steps=None)

# Run at least for 100 steps (disable min_epochs)
trainer = Trainer(min_steps=100, min_epochs=0)

max_time

Set the maximum amount of time for training. Training will get interrupted mid-epoch. For customizable options use the Timer callback.

# Default (disabled)
trainer = Trainer(max_time=None)

# Stop after 12 hours of training or when reaching 10 epochs (string)
trainer = Trainer(max_time="00:12:00:00", max_epochs=10)

# Stop after 1 day and 5 hours (dict)
trainer = Trainer(max_time={"days": 1, "hours": 5})

In case max_time is used together with min_steps or min_epochs, the min_* requirement always has precedence.

num_nodes


Number of GPU nodes for distributed training.

# default used by the Trainer
trainer = Trainer(num_nodes=1)

# to train on 8 nodes
trainer = Trainer(num_nodes=8)

num_processes

Warning

num_processes=x has been deprecated in v1.7 and will be removed in v2.0. Please use accelerator='cpu' and devices=x instead.


Number of processes to train with. Automatically set to the number of GPUs when using strategy="ddp". Set to a number greater than 1 when using accelerator="cpu" and strategy="ddp" to mimic distributed training on a machine without GPUs. This is useful for debugging, but will not provide any speedup, since single-process Torch already makes efficient use of multiple CPUs. While it would typically spawns subprocesses for training, setting num_nodes > 1 and keeping num_processes = 1 runs training in the main process.

# Simulate DDP for debugging on your GPU-less laptop
trainer = Trainer(accelerator="cpu", strategy="ddp", num_processes=2)

num_sanity_val_steps


Sanity check runs n batches of val before starting the training routine. This catches any bugs in your validation without having to wait for the first validation check. The Trainer uses 2 steps by default. Turn it off or modify it here.

# default used by the Trainer
trainer = Trainer(num_sanity_val_steps=2)

# turn it off
trainer = Trainer(num_sanity_val_steps=0)

# check all validation data
trainer = Trainer(num_sanity_val_steps=-1)

This option will reset the validation dataloader unless num_sanity_val_steps=0.

overfit_batches


Uses this much data of the training & validation set. If the training & validation dataloaders have shuffle=True, Lightning will automatically disable it.

Useful for quickly debugging or trying to overfit on purpose.

# default used by the Trainer
trainer = Trainer(overfit_batches=0.0)

# use only 1% of the train & val set
trainer = Trainer(overfit_batches=0.01)

# overfit on 10 of the same batches
trainer = Trainer(overfit_batches=10)

plugins


Plugins allow you to connect arbitrary backends, precision libraries, clusters etc. For example:

To define your own behavior, subclass the relevant class and pass it in. Here’s an example linking up your own ClusterEnvironment.

from pytorch_lightning.plugins.environments import ClusterEnvironment


class MyCluster(ClusterEnvironment):
    def main_address(self):
        return your_main_address

    def main_port(self):
        return your_main_port

    def world_size(self):
        return the_world_size


trainer = Trainer(plugins=[MyCluster()], ...)

precision


Lightning supports either double (64), float (32), bfloat16 (bf16), or half (16) precision training.

Half precision, or mixed precision, is the combined use of 32 and 16 bit floating points to reduce memory footprint during model training. This can result in improved performance, achieving +3X speedups on modern GPUs.

# default used by the Trainer
trainer = Trainer(precision=32)

# 16-bit precision
trainer = Trainer(precision=16, accelerator="gpu", devices=1)  # works only on CUDA

# bfloat16 precision
trainer = Trainer(precision="bf16")

# 64-bit precision
trainer = Trainer(precision=64)

Note

When running on TPUs, torch.bfloat16 will be used but tensor printing will still show torch.float32.

profiler


To profile individual steps during training and assist in identifying bottlenecks.

See the profiler documentation. for more details.

from pytorch_lightning.profilers import SimpleProfiler, AdvancedProfiler

# default used by the Trainer
trainer = Trainer(profiler=None)

# to profile standard training events, equivalent to `profiler=SimpleProfiler()`
trainer = Trainer(profiler="simple")

# advanced profiler for function-level stats, equivalent to `profiler=AdvancedProfiler()`
trainer = Trainer(profiler="advanced")

enable_progress_bar

Whether to enable or disable the progress bar. Defaults to True.

# default used by the Trainer
trainer = Trainer(enable_progress_bar=True)

# disable progress bar
trainer = Trainer(enable_progress_bar=False)

reload_dataloaders_every_n_epochs


Set to a positive integer to reload dataloaders every n epochs from your currently used data source. DataSource can be a LightningModule or a LightningDataModule.

# if 0 (default)
train_loader = model.train_dataloader()
# or if using data module: datamodule.train_dataloader()
for epoch in epochs:
    for batch in train_loader:
        ...

# if a positive integer
for epoch in epochs:
    if not epoch % reload_dataloaders_every_n_epochs:
        train_loader = model.train_dataloader()
        # or if using data module: datamodule.train_dataloader()
    for batch in train_loader:
        ...

The pseudocode applies also to the val_dataloader.

replace_sampler_ddp


Enables auto adding of DistributedSampler. In PyTorch, you must use it in distributed settings such as TPUs or multi-node. The sampler makes sure each GPU sees the appropriate part of your data. By default it will add shuffle=True for train sampler and shuffle=False for val/test sampler. If you already use a custom sampler, Lightning will wrap it in a way that it samples from your sampler in a distributed manner. If you want to customize it, you can set replace_sampler_ddp=False and add your own distributed sampler. If replace_sampler_ddp=True and a distributed sampler was already added, Lightning will not replace the existing one.

# default used by the Trainer
trainer = Trainer(replace_sampler_ddp=True)

By setting to False, you have to add your own distributed sampler:

# in your LightningModule or LightningDataModule
def train_dataloader(self):
    # default used by the Trainer
    sampler = torch.utils.data.distributed.DistributedSampler(dataset, shuffle=True)
    dataloader = DataLoader(dataset, batch_size=32, sampler=sampler)
    return dataloader

Note

For iterable datasets, we don’t do this automatically.

resume_from_checkpoint

Warning

resume_from_checkpoint is deprecated in v1.5 and will be removed in v2.0. Please pass trainer.fit(ckpt_path="some/path/to/my_checkpoint.ckpt") instead.


To resume training from a specific checkpoint pass in the path here. If resuming from a mid-epoch checkpoint, training will start from the beginning of the next epoch.

# default used by the Trainer
trainer = Trainer(resume_from_checkpoint=None)

# resume from a specific checkpoint
trainer = Trainer(resume_from_checkpoint="some/path/to/my_checkpoint.ckpt")

strategy

Supports passing different training strategies with aliases (ddp, ddp_spawn, etc) as well as custom strategies.

# Training with the DistributedDataParallel strategy on 4 GPUs
trainer = Trainer(strategy="ddp", accelerator="gpu", devices=4)

# Training with the DDP Spawn strategy using 4 cpu processes
trainer = Trainer(strategy="ddp_spawn", accelerator="cpu", devices=4)

Note

Additionally, you can pass your custom strategy to the strategy argument.

from pytorch_lightning.strategies import DDPStrategy


class CustomDDPStrategy(DDPStrategy):
    def configure_ddp(self):
        self._model = MyCustomDistributedDataParallel(
            self.model,
            device_ids=...,
        )


trainer = Trainer(strategy=CustomDDPStrategy(), accelerator="gpu", devices=2)
See Also:

sync_batchnorm


Enable synchronization between batchnorm layers across all GPUs.

trainer = Trainer(sync_batchnorm=True)

track_grad_norm


  • no tracking (-1)

  • Otherwise tracks that norm (2 for 2-norm)

# default used by the Trainer
trainer = Trainer(track_grad_norm=-1)

# track the 2-norm
trainer = Trainer(track_grad_norm=2)

tpu_cores

Warning

tpu_cores=x has been deprecated in v1.7 and will be removed in v2.0. Please use accelerator='tpu' and devices=x instead.


  • How many TPU cores to train on (1 or 8).

  • Which TPU core to train on [1-8]

A single TPU v2 or v3 has 8 cores. A TPU pod has up to 2048 cores. A slice of a POD means you get as many cores as you request.

Your effective batch size is batch_size * total tpu cores.

This parameter can be either 1 or 8.

Example:

# your_trainer_file.py

# default used by the Trainer (ie: train on CPU)
trainer = Trainer(tpu_cores=None)

# int: train on a single core
trainer = Trainer(tpu_cores=1)

# list: train on a single selected core
trainer = Trainer(tpu_cores=[2])

# int: train on all cores few cores
trainer = Trainer(tpu_cores=8)

# for 8+ cores must submit via xla script with
# a max of 8 cores specified. The XLA script
# will duplicate script onto each TPU in the POD
trainer = Trainer(tpu_cores=8)

To train on more than 8 cores (ie: a POD), submit this script using the xla_dist script.

Example:

python -m torch_xla.distributed.xla_dist
--tpu=$TPU_POD_NAME
--conda-env=torch-xla-nightly
--env=XLA_USE_BF16=1
-- python your_trainer_file.py

val_check_interval


How often within one training epoch to check the validation set. Can specify as float or int.

  • pass a float in the range [0.0, 1.0] to check after a fraction of the training epoch.

  • pass an int to check after a fixed number of training batches. An int value can only be higher than the number of training batches when check_val_every_n_epoch=None, which validates after every N training batches across epochs or iteration-based training.

# default used by the Trainer
trainer = Trainer(val_check_interval=1.0)

# check validation set 4 times during a training epoch
trainer = Trainer(val_check_interval=0.25)

# check validation set every 1000 training batches in the current epoch
trainer = Trainer(val_check_interval=1000)

# check validation set every 1000 training batches across complete epochs or during iteration-based training
# use this when using iterableDataset and your dataset has no length
# (ie: production cases with streaming data)
trainer = Trainer(val_check_interval=1000, check_val_every_n_epoch=None)
# Here is the computation to estimate the total number of batches seen within an epoch.

# Find the total number of train batches
total_train_batches = total_train_samples // (train_batch_size * world_size)

# Compute how many times we will call validation during the training loop
val_check_batch = max(1, int(total_train_batches * val_check_interval))
val_checks_per_epoch = total_train_batches / val_check_batch

# Find the total number of validation batches
total_val_batches = total_val_samples // (val_batch_size * world_size)

# Total number of batches run
total_fit_batches = total_train_batches + total_val_batches

enable_model_summary

Whether to enable or disable the model summarization. Defaults to True.

# default used by the Trainer
trainer = Trainer(enable_model_summary=True)

# disable summarization
trainer = Trainer(enable_model_summary=False)

# enable custom summarization
from pytorch_lightning.callbacks import ModelSummary

trainer = Trainer(enable_model_summary=True, callbacks=[ModelSummary(max_depth=-1)])

inference_mode

Whether to use torch.inference_mode() or torch.no_grad() mode during evaluation (validate/test/predict)

# default used by the Trainer
trainer = Trainer(inference_mode=True)

# Use `torch.no_grad` instead
trainer = Trainer(inference_mode=False)

With torch.inference_mode() disabled, you can enable the grad of your model layers if required.

class LitModel(LightningModule):
    def validation_step(self, batch, batch_idx):
        preds = self.layer1(batch)
        with torch.enable_grad():
            grad_preds = preds.requires_grad_()
            preds2 = self.layer2(grad_preds)


model = LitModel()
trainer = Trainer(inference_mode=False)
trainer.validate(model)

Trainer class API

Methods

init

Trainer.__init__(logger=True, enable_checkpointing=True, callbacks=None, default_root_dir=None, gradient_clip_val=None, gradient_clip_algorithm=None, num_nodes=1, num_processes=None, devices=None, gpus=None, auto_select_gpus=False, tpu_cores=None, ipus=None, enable_progress_bar=True, overfit_batches=0.0, track_grad_norm=- 1, check_val_every_n_epoch=1, fast_dev_run=False, accumulate_grad_batches=None, max_epochs=None, min_epochs=None, max_steps=- 1, min_steps=None, max_time=None, limit_train_batches=None, limit_val_batches=None, limit_test_batches=None, limit_predict_batches=None, val_check_interval=None, log_every_n_steps=50, accelerator=None, strategy=None, sync_batchnorm=False, precision=32, enable_model_summary=True, num_sanity_val_steps=2, resume_from_checkpoint=None, profiler=None, benchmark=None, deterministic=None, reload_dataloaders_every_n_epochs=0, auto_lr_find=False, replace_sampler_ddp=True, detect_anomaly=False, auto_scale_batch_size=False, plugins=None, amp_backend='native', amp_level=None, move_metrics_to_cpu=False, multiple_trainloader_mode='max_size_cycle', inference_mode=True)[source]

Customize every aspect of training via flags.

Parameters:
  • accelerator (Union[str, Accelerator, None]) –

    Supports passing different accelerator types (“cpu”, “gpu”, “tpu”, “ipu”, “hpu”, “mps, “auto”) as well as custom accelerator instances.

    Deprecated since version v1.5: Passing training strategies (e.g., ‘ddp’) to accelerator has been deprecated in v1.5.0 and will be removed in v1.7.0. Please use the strategy argument instead.

  • accumulate_grad_batches (Union[int, Dict[int, int], None]) – Accumulates grads every k batches or as set up in the dict. Default: None.

  • amp_backend (str) – The mixed precision backend to use (“native” or “apex”). Default: 'native''.

  • amp_level (Optional[str]) –

    The optimization level to use (O1, O2, etc…). By default it will be set to “O2” if amp_backend is set to “apex”.

    Deprecated since version v1.8: Setting amp_level inside the Trainer is deprecated in v1.8.0 and will be removed in v1.10.0. Please set it inside the specific precision plugin and pass it to the Trainer.

  • auto_lr_find (Union[bool, str]) – If set to True, will make trainer.tune() run a learning rate finder, trying to optimize initial learning for faster convergence. trainer.tune() method will set the suggested learning rate in self.lr or self.learning_rate in the LightningModule. To use a different key set a string instead of True with the key name. Default: False.

  • auto_scale_batch_size (Union[str, bool]) – If set to True, will initially run a batch size finder trying to find the largest batch size that fits into memory. The result will be stored in self.batch_size in the LightningModule or LightningDataModule depending on your setup. Additionally, can be set to either power that estimates the batch size through a power search or binsearch that estimates the batch size through a binary search. Default: False.

  • auto_select_gpus (bool) – If enabled and gpus or devices is an integer, pick available gpus automatically. This is especially useful when GPUs are configured to be in “exclusive mode”, such that only one process at a time can access them. Default: False.

  • benchmark (Optional[bool]) – The value (True or False) to set torch.backends.cudnn.benchmark to. The value for torch.backends.cudnn.benchmark set in the current session will be used (False if not manually set). If deterministic is set to True, this will default to False. Override to manually set a different value. Default: None.

  • callbacks (Union[List[Callback], Callback, None]) – Add a callback or list of callbacks. Default: None.

  • enable_checkpointing (bool) – If True, enable checkpointing. It will configure a default ModelCheckpoint callback if there is no user-defined ModelCheckpoint in callbacks. Default: True.

  • check_val_every_n_epoch (Optional[int]) – Perform a validation loop every after every N training epochs. If None, validation will be done solely based on the number of training batches, requiring val_check_interval to be an integer value. Default: 1.

  • default_root_dir (Union[str, Path, None]) – Default path for logs and weights when no logger/ckpt_callback passed. Default: os.getcwd(). Can be remote file paths such as s3://mybucket/path or ‘hdfs://path/’

  • detect_anomaly (bool) – Enable anomaly detection for the autograd engine. Default: False.

  • deterministic (Union[bool, Literal['warn'], None]) – If True, sets whether PyTorch operations must use deterministic algorithms. Set to "warn" to use deterministic algorithms whenever possible, throwing warnings on operations that don’t support deterministic mode (requires PyTorch 1.11+). If not set, defaults to False. Default: None.

  • devices (Union[List[int], str, int, None]) – Will be mapped to either gpus, tpu_cores, num_processes or ipus, based on the accelerator type.

  • fast_dev_run (Union[int, bool]) – Runs n if set to n (int) else 1 if set to True batch(es) of train, val and test to find any bugs (ie: a sort of unit test). Default: False.

  • gpus (Union[List[int], str, int, None]) –

    Number of GPUs to train on (int) or which GPUs to train on (list or str) applied per node Default: None.

    Deprecated since version v1.7: gpus has been deprecated in v1.7 and will be removed in v2.0. Please use accelerator='gpu' and devices=x instead.

  • gradient_clip_val (Union[int, float, None]) – The value at which to clip gradients. Passing gradient_clip_val=None disables gradient clipping. If using Automatic Mixed Precision (AMP), the gradients will be unscaled before. Default: None.

  • gradient_clip_algorithm (Optional[str]) – The gradient clipping algorithm to use. Pass gradient_clip_algorithm="value" to clip by value, and gradient_clip_algorithm="norm" to clip by norm. By default it will be set to "norm".

  • limit_train_batches (Union[int, float, None]) – How much of training dataset to check (float = fraction, int = num_batches). Default: 1.0.

  • limit_val_batches (Union[int, float, None]) – How much of validation dataset to check (float = fraction, int = num_batches). Default: 1.0.

  • limit_test_batches (Union[int, float, None]) – How much of test dataset to check (float = fraction, int = num_batches). Default: 1.0.

  • limit_predict_batches (Union[int, float, None]) – How much of prediction dataset to check (float = fraction, int = num_batches). Default: 1.0.

  • logger (Union[Logger, Iterable[Logger], bool]) – Logger (or iterable collection of loggers) for experiment tracking. A True value uses the default TensorBoardLogger. False will disable logging. If multiple loggers are provided, local files (checkpoints, profiler traces, etc.) are saved in the log_dir of the first logger. Default: True.

  • log_every_n_steps (int) – How often to log within steps. Default: 50.

  • enable_progress_bar (bool) – Whether to enable to progress bar by default. Default: True.

  • profiler (Union[Profiler, str, None]) – To profile individual steps during training and assist in identifying bottlenecks. Default: None.

  • overfit_batches (Union[int, float]) – Overfit a fraction of training/validation data (float) or a set number of batches (int). Default: 0.0.

  • plugins (Union[PrecisionPlugin, ClusterEnvironment, CheckpointIO, LayerSync, str, List[Union[PrecisionPlugin, ClusterEnvironment, CheckpointIO, LayerSync, str]], None]) – Plugins allow modification of core behavior like ddp and amp, and enable custom lightning plugins. Default: None.

  • precision (Union[int, str]) – Double precision (64), full precision (32), half precision (16) or bfloat16 precision (bf16). Can be used on CPU, GPU, TPUs, HPUs or IPUs. Default: 32.

  • max_epochs (Optional[int]) – Stop training once this number of epochs is reached. Disabled by default (None). If both max_epochs and max_steps are not specified, defaults to max_epochs = 1000. To enable infinite training, set max_epochs = -1.

  • min_epochs (Optional[int]) – Force training for at least these many epochs. Disabled by default (None).

  • max_steps (int) – Stop training after this number of steps. Disabled by default (-1). If max_steps = -1 and max_epochs = None, will default to max_epochs = 1000. To enable infinite training, set max_epochs to -1.

  • min_steps (Optional[int]) – Force training for at least these number of steps. Disabled by default (None).

  • max_time (Union[str, timedelta, Dict[str, int], None]) – Stop training after this amount of time has passed. Disabled by default (None). The time duration can be specified in the format DD:HH:MM:SS (days, hours, minutes seconds), as a datetime.timedelta, or a dictionary with keys that will be passed to datetime.timedelta.

  • num_nodes (int) – Number of GPU nodes for distributed training. Default: 1.

  • num_processes (Optional[int]) –

    Number of processes for distributed training with accelerator="cpu". Default: 1.

    Deprecated since version v1.7: num_processes has been deprecated in v1.7 and will be removed in v2.0. Please use accelerator='cpu' and devices=x instead.

  • num_sanity_val_steps (int) – Sanity check runs n validation batches before starting the training routine. Set it to -1 to run all batches in all validation dataloaders. Default: 2.

  • reload_dataloaders_every_n_epochs (int) – Set to a non-negative integer to reload dataloaders every n epochs. Default: 0.

  • replace_sampler_ddp (bool) – Explicitly enables or disables sampler replacement. If not specified this will toggled automatically when DDP is used. By default it will add shuffle=True for train sampler and shuffle=False for val/test sampler. If you want to customize it, you can set replace_sampler_ddp=False and add your own distributed sampler.

  • resume_from_checkpoint (Union[str, Path, None]) –

    Path/URL of the checkpoint from which training is resumed. If there is no checkpoint file at the path, an exception is raised. If resuming from mid-epoch checkpoint, training will start from the beginning of the next epoch.

    Deprecated since version v1.5: resume_from_checkpoint is deprecated in v1.5 and will be removed in v2.0. Please pass the path to Trainer.fit(..., ckpt_path=...) instead.

  • strategy (Union[str, Strategy, None]) – Supports different training strategies with aliases as well custom strategies. Default: None.

  • sync_batchnorm (bool) – Synchronize batch norm layers between process groups/whole world. Default: False.

  • tpu_cores (Union[List[int], str, int, None]) –

    How many TPU cores to train on (1 or 8) / Single TPU to train on (1) Default: None.

    Deprecated since version v1.7: tpu_cores has been deprecated in v1.7 and will be removed in v2.0. Please use accelerator='tpu' and devices=x instead.

  • ipus (Optional[int]) –

    How many IPUs to train on. Default: None.

    Deprecated since version v1.7: ipus has been deprecated in v1.7 and will be removed in v2.0. Please use accelerator='ipu' and devices=x instead.

  • track_grad_norm (Union[int, float, str]) – -1 no tracking. Otherwise tracks that p-norm. May be set to ‘inf’ infinity-norm. If using Automatic Mixed Precision (AMP), the gradients will be unscaled before logging them. Default: -1.

  • val_check_interval (Union[int, float, None]) – How often to check the validation set. Pass a float in the range [0.0, 1.0] to check after a fraction of the training epoch. Pass an int to check after a fixed number of training batches. An int value can only be higher than the number of training batches when check_val_every_n_epoch=None, which validates after every N training batches across epochs or during iteration-based training. Default: 1.0.

  • enable_model_summary (bool) – Whether to enable model summarization by default. Default: True.

  • move_metrics_to_cpu (bool) – Whether to force internal logged metrics to be moved to cpu. This can save some gpu memory, but can make training slower. Use with attention. Default: False.

  • multiple_trainloader_mode (str) – How to loop over the datasets when there are multiple train loaders. In ‘max_size_cycle’ mode, the trainer ends one epoch when the largest dataset is traversed, and smaller datasets reload when running out of their data. In ‘min_size’ mode, all the datasets reload when reaching the minimum length of datasets. Default: "max_size_cycle".

  • inference_mode (bool) – Whether to use torch.inference_mode() or torch.no_grad() during evaluation (validate/test/predict).

fit

Trainer.fit(model, train_dataloaders=None, val_dataloaders=None, datamodule=None, ckpt_path=None)[source]

Runs the full optimization routine.

Parameters:
Return type:

None

validate

Trainer.validate(model=None, dataloaders=None, ckpt_path=None, verbose=True, datamodule=None)[source]

Perform one evaluation epoch over the validation set.

Parameters:
Return type:

List[Dict[str, float]]

Returns:

List of dictionaries with metrics logged during the validation phase, e.g., in model- or callback hooks like validation_step(), validation_epoch_end(), etc. The length of the list corresponds to the number of validation dataloaders used.

test

Trainer.test(model=None, dataloaders=None, ckpt_path=None, verbose=True, datamodule=None)[source]

Perform one evaluation epoch over the test set. It’s separated from fit to make sure you never run on your test set until you want to.

Parameters:
Return type:

List[Dict[str, float]]

Returns:

List of dictionaries with metrics logged during the test phase, e.g., in model- or callback hooks like test_step(), test_epoch_end(), etc. The length of the list corresponds to the number of test dataloaders used.

predict

Trainer.predict(model=None, dataloaders=None, datamodule=None, return_predictions=None, ckpt_path=None)[source]

Run inference on your data. This will call the model forward function to compute predictions. Useful to perform distributed and batched predictions. Logging is disabled in the predict hooks.

Parameters:
Return type:

Union[List[Any], List[List[Any]], None]

Returns:

Returns a list of dictionaries, one for each provided dataloader containing their respective predictions.

See Lightning inference section for more.

tune

Trainer.tune(model, train_dataloaders=None, val_dataloaders=None, dataloaders=None, datamodule=None, scale_batch_size_kwargs=None, lr_find_kwargs=None, method='fit')[source]

Runs routines to tune hyperparameters before training.

Parameters:
Return type:

_TunerResult

Properties

callback_metrics

The metrics available to callbacks. These are automatically set when you log via self.log

def training_step(self, batch, batch_idx):
    self.log("a_val", 2)


callback_metrics = trainer.callback_metrics
assert callback_metrics["a_val"] == 2

current_epoch

The number of epochs run.

if trainer.current_epoch >= 10:
    ...

datamodule

The current datamodule, which is used by the trainer.

used_datamodule = trainer.datamodule

is_last_batch

Whether trainer is executing last batch in the current epoch.

if trainer.is_last_batch:
    ...

global_step

The number of optimizer steps taken (does not reset each epoch). This includes multiple optimizers and TBPTT steps (if enabled).

if trainer.global_step >= 100:
    ...

logger

The current logger being used. Here’s an example using tensorboard

logger = trainer.logger
tensorboard = logger.experiment

loggers

The list of loggers currently being used by the Trainer.

# List of Logger objects
loggers = trainer.loggers
for logger in loggers:
    logger.log_metrics({"foo": 1.0})

logged_metrics

The metrics sent to the logger (visualizer).

def training_step(self, batch, batch_idx):
    self.log("a_val", 2, logger=True)


logged_metrics = trainer.logged_metrics
assert logged_metrics["a_val"] == 2

log_dir

The directory for the current experiment. Use this to save images to, etc…

def training_step(self, batch, batch_idx):
    img = ...
    save_img(img, self.trainer.log_dir)

is_global_zero

Whether this process is the global zero in multi-node training

def training_step(self, batch, batch_idx):
    if self.trainer.is_global_zero:
        print("in node 0, accelerator 0")

progress_bar_metrics

The metrics sent to the progress bar.

def training_step(self, batch, batch_idx):
    self.log("a_val", 2, prog_bar=True)


progress_bar_metrics = trainer.progress_bar_metrics
assert progress_bar_metrics["a_val"] == 2

predict_dataloaders

The current predict dataloaders of the trainer. Note that property returns a list of predict dataloaders.

used_predict_dataloaders = trainer.predict_dataloaders

estimated_stepping_batches

Check out estimated_stepping_batches().

state

The current state of the Trainer, including the current function that is running, the stage of execution within that function, and the status of the Trainer.

# fn in ("fit", "validate", "test", "predict", "tune")
trainer.state.fn
# status in ("initializing", "running", "finished", "interrupted")
trainer.state.status
# stage in ("train", "sanity_check", "validate", "test", "predict", "tune")
trainer.state.stage

should_stop

If you want to terminate the training during .fit, you can set trainer.should_stop=True to terminate the training as soon as possible. Note that, it will respect the arguments min_steps and min_epochs to check whether to stop. If these arguments are set and the current_epoch or global_step don’t meet these minimum conditions, training will continue until both conditions are met. If any of these arguments is not set, it won’t be considered for the final decision.

# setting `trainer.should_stop` at any point of training will terminate it
class LitModel(LightningModule):
    def training_step(self, *args, **kwargs):
        self.trainer.should_stop = True


trainer = Trainer()
model = LitModel()
trainer.fit(model)
# setting `trainer.should_stop` will stop training only after at least 5 epochs have run
class LitModel(LightningModule):
    def training_step(self, *args, **kwargs):
        if self.current_epoch == 2:
            self.trainer.should_stop = True


trainer = Trainer(min_epochs=5, max_epochs=100)
model = LitModel()
trainer.fit(model)
# setting `trainer.should_stop` will stop training only after at least 5 steps have run
class LitModel(LightningModule):
    def training_step(self, *args, **kwargs):
        if self.global_step == 2:
            self.trainer.should_stop = True


trainer = Trainer(min_steps=5, max_epochs=100)
model = LitModel()
trainer.fit(model)
# setting `trainer.should_stop` at any until both min_steps and min_epochs are satisfied
class LitModel(LightningModule):
    def training_step(self, *args, **kwargs):
        if self.global_step == 7:
            self.trainer.should_stop = True


trainer = Trainer(min_steps=5, min_epochs=5, max_epochs=100)
model = LitModel()
trainer.fit(model)

train_dataloader

The current train dataloader of the trainer.

used_train_dataloader = trainer.train_dataloader

test_dataloaders

The current test dataloaders of the trainer. Note that property returns a list of test dataloaders.

used_test_dataloaders = trainer.test_dataloaders

val_dataloaders

The current val dataloaders of the trainer. Note that property returns a list of val dataloaders.

used_val_dataloaders = trainer.val_dataloaders