# -*- coding: utf-8 -*-
import importlib
import multiprocessing
import os
import platform
import sys
import time
import traceback
from argparse import Namespace
from dataclasses import dataclass, field
from inspect import signature
from typing import Callable, Dict, List, Tuple, Union
import torch
import torch.distributed as dist
from coqpit import Coqpit
from torch import nn
from torch.nn.parallel import DistributedDataParallel as DDP_th
from torch.utils.data import DataLoader
from TTS.utils.callbacks import TrainerCallback
from TTS.utils.distribute import init_distributed
from TTS.utils.generic_utils import (
KeepAverage,
count_parameters,
get_experiment_folder_path,
get_git_branch,
remove_experiment_folder,
set_init_dict,
to_cuda,
)
from TTS.utils.io import copy_model_files, load_fsspec, save_best_model, save_checkpoint
from TTS.utils.logging import ConsoleLogger, TensorboardLogger, WandbLogger, init_dashboard_logger
from TTS.utils.trainer_utils import (
get_last_checkpoint,
get_optimizer,
get_scheduler,
is_apex_available,
setup_torch_training_env,
)
multiprocessing.set_start_method("fork")
if platform.system() != "Windows":
# https://github.com/pytorch/pytorch/issues/973
import resource
rlimit = resource.getrlimit(resource.RLIMIT_NOFILE)
resource.setrlimit(resource.RLIMIT_NOFILE, (4096, rlimit[1]))
if is_apex_available():
from apex import amp
@dataclass
class TrainingArgs(Coqpit):
"""Trainer arguments to be defined externally. It helps integrating the `Trainer` with the higher level APIs and
set the values for distributed training."""
continue_path: str = field(
default="",
metadata={
"help": "Path to a training folder to continue training. Restore the model from the last checkpoint and continue training under the same folder."
},
)
restore_path: str = field(
default="",
metadata={
"help": "Path to a model checkpoit. Restore the model with the given checkpoint and start a new training."
},
)
best_path: str = field(
default="",
metadata={
"help": "Best model file to be used for extracting the best loss. If not specified, the latest best model in continue path is used"
},
)
skip_train_epoch: bool = field(
default=False, metadata={"help": "Run only evaluation iteration. Useful for debugging."}
)
config_path: str = field(default="", metadata={"help": "Path to the configuration file."})
rank: int = field(default=0, metadata={"help": "Process rank in distributed training."})
group_id: str = field(default="", metadata={"help": "Process group id in distributed training."})
use_ddp: bool = field(
default=False,
metadata={"help": "Use DDP in distributed training. It is to set in `distribute.py`. Do not set manually."},
)
class Trainer:
def __init__( # pylint: disable=dangerous-default-value
self,
args: Union[Coqpit, Namespace],
config: Coqpit,
output_path: str,
c_logger: ConsoleLogger = None,
dashboard_logger: Union[TensorboardLogger, WandbLogger] = None,
model: nn.Module = None,
get_model: Callable = None,
get_data_samples: Callable = None,
train_samples: List = None,
eval_samples: List = None,
cudnn_benchmark: bool = False,
training_assets: Dict = {},
parse_command_line_args: bool = True,
) -> None:
"""Simple yet powerful 🐸💬 TTS trainer for PyTorch. It can train all the available `tts` and `vocoder` models
or easily be customized.
Notes:
Supports Automatic Mixed Precision training. If `Apex` is availabe, it automatically picks that, else
it uses PyTorch's native `amp` module. `Apex` may provide more stable training in some cases.
Args:
args (Union[Coqpit, Namespace]): Training arguments parsed either from console by `argparse` or `TrainingArgs`
config object.
config (Coqpit): Model config object. It includes all the values necessary for initializing, training, evaluating
and testing the model.
output_path (str): Path to the output training folder. All the files are saved under thi path.
c_logger (ConsoleLogger, optional): Console logger for printing training status. If not provided, the default
console logger is used. Defaults to None.
dashboard_logger Union[TensorboardLogger, WandbLogger]: Dashboard logger. If not provided, the tensorboard logger is used.
Defaults to None.
model (nn.Module, optional): Initialized and ready-to-train model. If it is not defined, `Trainer`
initializes a model from the provided config. Defaults to None.
get_model (Callable):
A function that returns a model. It is used to initialize the model when `model` is not provided.
It either takes the config as the only argument or does not take any argument.
Defaults to None
get_data_samples (Callable):
A function that returns a list of training and evaluation samples. Used if `train_samples` and
`eval_samples` are None. Defaults to None.
train_samples (List):
A list of training samples used by the model's `get_data_loader` to init the `dataset` and the
`data_loader`. Defaults to None.
eval_samples (List):
A list of evaluation samples used by the model's `get_data_loader` to init the `dataset` and the
`data_loader`. Defaults to None.
cudnn_benchmark (bool): enable/disable PyTorch cudnn benchmarking. It is better to disable if the model input
length is changing batch to batch along the training.
training_assets (Dict):
A dictionary of assets to be used at training and passed to the model's ```train_log(), eval_log(), get_data_loader()```
during training. It can include `AudioProcessor` or/and `Tokenizer`. Defaults to {}.
parse_command_line_args (bool):
If true, parse command-line arguments and update `TrainingArgs` and model `config` values. Set it
to false if you parse the arguments yourself. Defaults to True.
Examples:
Running trainer with HifiGAN model.
>>> args = TrainingArgs(...)
>>> config = HifiganConfig(...)
>>> model = GANModel(config)
>>> ap = AudioProcessor(**config.audio)
>>> assets = {"audio_processor": ap}
>>> trainer = Trainer(args, config, output_path, model=model, training_assets=assets)
>>> trainer.fit()
TODO:
- Wrap model for not calling .module in DDP.
- Accumulate gradients b/w batches.
- Deepspeed integration
- Profiler integration.
- Overfitting to a batch.
- TPU training
- NOTE: Consider moving `training_assets` to the model implementation.
"""
if parse_command_line_args:
# parse command-line arguments for TrainerArgs()
args, coqpit_overrides = self.parse_argv(args)
# get ready for training and parse command-line arguments for the model config
config = self.init_training(args, coqpit_overrides, config)
# set the output path
if args.continue_path:
# use the same path as the continuing run
output_path = args.continue_path
else:
# override the output path if it is provided
output_path = config.output_path if output_path is None else output_path
# create a new output folder name
output_path = get_experiment_folder_path(config.output_path, config.run_name)
os.makedirs(output_path, exist_ok=True)
# copy training assets to the output folder
copy_model_files(config, output_path)
# init class members
self.args = args
self.config = config
self.output_path = output_path
self.config.output_log_path = output_path
self.training_assets = training_assets
# setup logging
log_file = os.path.join(self.output_path, f"trainer_{args.rank}_log.txt")
self._setup_logger_config(log_file)
time.sleep(1.0) # wait for the logger to be ready
# set and initialize Pytorch runtime
self.use_cuda, self.num_gpus = setup_torch_training_env(True, cudnn_benchmark, args.use_ddp)
# init loggers
self.c_logger = ConsoleLogger() if c_logger is None else c_logger
self.dashboard_logger = dashboard_logger
# only allow dashboard logging for the main process in DDP mode
if self.dashboard_logger is None and args.rank == 0:
self.dashboard_logger = init_dashboard_logger(config)
if not self.config.log_model_step:
self.config.log_model_step = self.config.save_step
self.total_steps_done = 0
self.epochs_done = 0
self.restore_step = 0
self.best_loss = float("inf")
self.train_loader = None
self.eval_loader = None
self.keep_avg_train = None
self.keep_avg_eval = None
self.use_apex = self._is_apex_available()
self.use_amp_scaler = self.config.mixed_precision and self.use_cuda
# load data samples
if train_samples is None and get_data_samples is None:
raise ValueError("[!] `train_samples` and `get_data_samples` cannot both be None.")
if train_samples is not None:
self.train_samples = train_samples
self.eval_samples = eval_samples
else:
self.train_samples, self.eval_samples = self.run_get_data_samples(config, get_data_samples)
# init TTS model
if model is None and get_model is None:
raise ValueError("[!] `model` and `get_model` cannot both be None.")
if model is not None:
self.model = model
else:
self.run_get_model(self.config, get_model)
# setup criterion
self.criterion = self.get_criterion(self.model)
# DISTRUBUTED
if self.num_gpus > 1:
init_distributed(
args.rank,
self.num_gpus,
args.group_id,
self.config.distributed_backend,
self.config.distributed_url,
)
if self.use_cuda:
self.model.cuda()
if isinstance(self.criterion, list):
self.criterion = [x.cuda() for x in self.criterion]
else:
self.criterion.cuda()
# setup optimizer
self.optimizer = self.get_optimizer(self.model, self.config)
# CALLBACK
self.callbacks = TrainerCallback()
self.callbacks.on_init_start(self)
# init AMP
if self.use_amp_scaler:
if self.use_apex:
self.scaler = None
self.model, self.optimizer = amp.initialize(self.model, self.optimizer, opt_level="O1")
# if isinstance(self.optimizer, list):
# self.scaler = [torch.cuda.amp.GradScaler()] * len(self.optimizer)
# else:
self.scaler = torch.cuda.amp.GradScaler()
else:
self.scaler = None
if self.args.restore_path:
self.model, self.optimizer, self.scaler, self.restore_step = self.restore_model(
self.config, args.restore_path, self.model, self.optimizer, self.scaler
)
# setup scheduler
self.scheduler = self.get_scheduler(self.model, self.config, self.optimizer)
if self.scheduler is not None:
if self.args.continue_path:
if isinstance(self.scheduler, list):
for scheduler in self.scheduler:
if scheduler is not None:
scheduler.last_epoch = self.restore_step
else:
self.scheduler.last_epoch = self.restore_step
# DISTRIBUTED
if self.num_gpus > 1:
self.model = DDP_th(self.model, device_ids=[args.rank], output_device=args.rank)
# count model size
num_params = count_parameters(self.model)
print("\n > Model has {} parameters".format(num_params))
self.callbacks.on_init_end(self)
@staticmethod
def parse_argv(args: Union[Coqpit, List]):
"""Parse command line arguments to init or override `TrainingArgs()`."""
if isinstance(args, Coqpit):
parser = args.init_argparse(arg_prefix="")
else:
train_config = TrainingArgs()
parser = train_config.init_argparse(arg_prefix="")
training_args, coqpit_overrides = parser.parse_known_args()
args.parse_args(training_args)
return args, coqpit_overrides
def init_training(
self, args: TrainingArgs, coqpit_overrides: Dict, config: Coqpit = None
): # pylint: disable=no-self-use
"""Initialize training and update model configs from command line arguments.
Args:
args (argparse.Namespace or dict like): Parsed input arguments.
config_overrides (argparse.Namespace or dict like): Parsed config overriding arguments.
config (Coqpit): Model config. If none, it is generated from `args`. Defaults to None.
Returns:
c (TTS.utils.io.AttrDict): Config paramaters.
"""
# set arguments for continuing training
if args.continue_path:
experiment_path = args.continue_path
args.config_path = os.path.join(args.continue_path, "config.json")
args.restore_path, best_model = get_last_checkpoint(args.continue_path)
if not args.best_path:
args.best_path = best_model
# override config values from command-line args
# TODO: Maybe it is better to do it outside
if len(coqpit_overrides) > 0:
config.parse_known_args(coqpit_overrides, arg_prefix="coqpit", relaxed_parser=True)
experiment_path = args.continue_path
# update the config.json fields and copy it to the output folder
if args.rank == 0:
new_fields = {}
if args.restore_path:
new_fields["restore_path"] = args.restore_path
new_fields["github_branch"] = get_git_branch()
copy_model_files(config, experiment_path, new_fields)
return config
@staticmethod
def run_get_model(config: Coqpit, get_model: Callable) -> nn.Module:
"""Run the `get_model` function and return the model.
Args:
config (Coqpit): Model config.
Returns:
nn.Module: initialized model.
"""
if len(signature(get_model).sig.parameters) == 1:
model = get_model(config)
else:
model = get_model()
return model
@staticmethod
def run_get_data_samples(config: Coqpit, get_data_samples: Callable) -> nn.Module:
if callable(get_data_samples):
if len(signature(get_data_samples).sig.parameters) == 1:
train_samples, eval_samples = get_data_samples(config)
else:
train_samples, eval_samples = get_data_samples()
return train_samples, eval_samples
return None, None
def restore_model(
self,
config: Coqpit,
restore_path: str,
model: nn.Module,
optimizer: torch.optim.Optimizer,
scaler: torch.cuda.amp.GradScaler = None,
) -> Tuple[nn.Module, torch.optim.Optimizer, torch.cuda.amp.GradScaler, int]:
"""Restore training from an old run. It restores model, optimizer, AMP scaler and training stats.
Args:
config (Coqpit): Model config.
restore_path (str): Path to the restored training run.
model (nn.Module): Model to restored.
optimizer (torch.optim.Optimizer): Optimizer to restore.
scaler (torch.cuda.amp.GradScaler, optional): AMP scaler to restore. Defaults to None.
Returns:
Tuple[nn.Module, torch.optim.Optimizer, torch.cuda.amp.GradScaler, int]: [description]
"""
def _restore_list_objs(states, obj):
if isinstance(obj, list):
for idx, state in enumerate(states):
obj[idx].load_state_dict(state)
else:
obj.load_state_dict(states)
return obj
print(" > Restoring from %s ..." % os.path.basename(restore_path))
checkpoint = load_fsspec(restore_path, map_location="cpu")
try:
print(" > Restoring Model...")
model.load_state_dict(checkpoint["model"])
print(" > Restoring Optimizer...")
optimizer = _restore_list_objs(checkpoint["optimizer"], optimizer)
if "scaler" in checkpoint and self.use_amp_scaler and checkpoint["scaler"]:
print(" > Restoring Scaler...")
scaler = _restore_list_objs(checkpoint["scaler"], scaler)
except (KeyError, RuntimeError, ValueError):
print(" > Partial model initialization...")
model_dict = model.state_dict()
model_dict = set_init_dict(model_dict, checkpoint["model"], config)
model.load_state_dict(model_dict)
del model_dict
if isinstance(self.optimizer, list):
for idx, optim in enumerate(optimizer):
for group in optim.param_groups:
group["lr"] = self.get_lr(model, config)[idx]
else:
for group in optimizer.param_groups:
group["lr"] = self.get_lr(model, config)
print(
" > Model restored from step %d" % checkpoint["step"],
)
restore_step = checkpoint["step"]
torch.cuda.empty_cache()
return model, optimizer, scaler, restore_step
#########################
# DATA LOADING FUNCTIONS
#########################
def _get_loader(
self,
model: nn.Module,
config: Coqpit,
assets: Dict,
is_eval: bool,
data_items: List,
verbose: bool,
num_gpus: int,
) -> DataLoader:
if num_gpus > 1:
if hasattr(model.module, "get_data_loader"):
loader = model.module.get_data_loader(
config, assets, is_eval, data_items, verbose, num_gpus, self.args.rank
)
else:
if hasattr(model, "get_data_loader"):
loader = model.get_data_loader(config, assets, is_eval, data_items, verbose, num_gpus)
return loader
def get_train_dataloader(self, training_assets: Dict, data_items: List, verbose: bool) -> DataLoader:
"""Initialize and return a training data loader.
Args:
ap (AudioProcessor): Audio processor.
data_items (List): Data samples used for training.
verbose (bool): enable/disable printing loader stats at initialization.
Returns:
DataLoader: Initialized training data loader.
"""
return self._get_loader(self.model, self.config, training_assets, False, data_items, verbose, self.num_gpus)
def get_eval_dataloader(self, training_assets: Dict, data_items: List, verbose: bool) -> DataLoader:
return self._get_loader(self.model, self.config, training_assets, True, data_items, verbose, self.num_gpus)
def format_batch(self, batch: List) -> Dict:
"""Format the dataloader output and return a batch.
Args:
batch (List): Batch returned by the dataloader.
Returns:
Dict: Formatted batch.
"""
if self.num_gpus > 1:
batch = self.model.module.format_batch(batch)
else:
batch = self.model.format_batch(batch)
if self.use_cuda:
for k, v in batch.items():
batch[k] = to_cuda(v)
return batch
######################
# TRAIN FUNCTIONS
######################
@staticmethod
def master_params(optimizer: torch.optim.Optimizer):
"""Generator over parameters owned by the optimizer.
Used to select parameters used by the optimizer for gradient clipping.
Args:
optimizer: Target optimizer.
"""
for group in optimizer.param_groups:
for p in group["params"]:
yield p
@staticmethod
def _model_train_step(
batch: Dict, model: nn.Module, criterion: nn.Module, optimizer_idx: int = None
) -> Tuple[Dict, Dict]:
"""
Perform a trainig forward step. Compute model outputs and losses.
Args:
batch (Dict): [description]
model (nn.Module): [description]
criterion (nn.Module): [description]
optimizer_idx (int, optional): [description]. Defaults to None.
Returns:
Tuple[Dict, Dict]: [description]
"""
input_args = [batch, criterion]
if optimizer_idx is not None:
input_args.append(optimizer_idx)
# unwrap model in DDP training
if hasattr(model, "module"):
return model.module.train_step(*input_args)
return model.train_step(*input_args)
def _optimize(
self,
batch: Dict,
model: nn.Module,
optimizer: Union[torch.optim.Optimizer, List],
scaler: "AMPScaler",
criterion: nn.Module,
scheduler: Union[torch.optim.lr_scheduler._LRScheduler, List], # pylint: disable=protected-access
config: Coqpit,
optimizer_idx: int = None,
) -> Tuple[Dict, Dict, int]:
"""Perform a forward - backward pass and run the optimizer.
Args:
batch (Dict): Input batch. If
model (nn.Module): Model for training. Defaults to None.
optimizer (Union[nn.optim.Optimizer, List]): Model's optimizer. If it is a list then, `optimizer_idx` must be defined to indicate the optimizer in use.
scaler (AMPScaler): AMP scaler.
criterion (nn.Module): Model's criterion.
scheduler (torch.optim.lr_scheduler._LRScheduler): LR scheduler used by the optimizer.
config (Coqpit): Model config.
optimizer_idx (int, optional): Target optimizer being used. Defaults to None.
Raises:
RuntimeError: When the loss is NaN.
Returns:
Tuple[Dict, Dict, int, torch.Tensor]: model outputs, losses, step time and gradient norm.
"""
step_start_time = time.time()
# zero-out optimizer
optimizer.zero_grad()
# forward pass and loss computation
with torch.cuda.amp.autocast(enabled=config.mixed_precision):
if optimizer_idx is not None:
outputs, loss_dict = self._model_train_step(batch, model, criterion, optimizer_idx=optimizer_idx)
else:
outputs, loss_dict = self._model_train_step(batch, model, criterion)
# skip the rest
if outputs is None:
step_time = time.time() - step_start_time
return None, {}, step_time
# # check nan loss
# if torch.isnan(loss_dict["loss"]).any():
# raise RuntimeError(f" > NaN loss detected - {loss_dict}")
# set gradient clipping threshold
if "grad_clip" in config and config.grad_clip is not None:
if optimizer_idx is not None:
grad_clip = config.grad_clip[optimizer_idx]
else:
grad_clip = config.grad_clip
else:
grad_clip = 0.0 # meaning no gradient clipping
# optimizer step
grad_norm = 0
update_lr_scheduler = True
if self.use_amp_scaler:
if self.use_apex:
# TODO: verify AMP use for GAN training in TTS
# https://nvidia.github.io/apex/advanced.html?highlight=accumulate#backward-passes-with-multiple-optimizers
with amp.scale_loss(loss_dict["loss"], optimizer) as scaled_loss:
scaled_loss.backward()
grad_norm = torch.nn.utils.clip_grad_norm_(amp.master_params(optimizer), grad_clip)
else:
# model optimizer step in mixed precision mode
scaler.scale(loss_dict["loss"]).backward()
if grad_clip > 0:
scaler.unscale_(optimizer)
grad_norm = torch.nn.utils.clip_grad_norm_(self.master_params(optimizer), grad_clip)
scale_prev = scaler.get_scale()
scaler.step(optimizer)
scaler.update()
update_lr_scheduler = scale_prev <= scaler.get_scale()
loss_dict["amp_scaler"] = scaler.get_scale() # for logging
else:
# main model optimizer step
loss_dict["loss"].backward()
if grad_clip > 0:
grad_norm = torch.nn.utils.clip_grad_norm_(model.parameters(), grad_clip)
optimizer.step()
# pytorch skips the step when the norm is 0. So ignore the norm value when it is NaN
if isinstance(grad_norm, torch.Tensor) and (torch.isnan(grad_norm) or torch.isinf(grad_norm)):
grad_norm = 0
step_time = time.time() - step_start_time
# setup lr
if scheduler is not None and update_lr_scheduler and not self.config.scheduler_after_epoch:
scheduler.step()
# detach losses
loss_dict = self._detach_loss_dict(loss_dict)
if optimizer_idx is not None:
loss_dict[f"loss_{optimizer_idx}"] = loss_dict.pop("loss")
loss_dict[f"grad_norm_{optimizer_idx}"] = grad_norm
else:
loss_dict["grad_norm"] = grad_norm
return outputs, loss_dict, step_time
def train_step(self, batch: Dict, batch_n_steps: int, step: int, loader_start_time: float) -> Tuple[Dict, Dict]:
"""Perform a training step on a batch of inputs and log the process.
Args:
batch (Dict): Input batch.
batch_n_steps (int): Number of steps needed to complete an epoch. Needed for logging.
step (int): Current step number in this epoch.
loader_start_time (float): The time when the data loading is started. Needed for logging.
Returns:
Tuple[Dict, Dict]: Model outputs and losses.
"""
self.callbacks.on_train_step_start(self)
# format data
batch = self.format_batch(batch)
loader_time = time.time() - loader_start_time
# conteainers to hold model outputs and losses for each optimizer.
outputs_per_optimizer = None
loss_dict = {}
if not isinstance(self.optimizer, list):
# training with a single optimizer
outputs, loss_dict_new, step_time = self._optimize(
batch, self.model, self.optimizer, self.scaler, self.criterion, self.scheduler, self.config
)
loss_dict.update(loss_dict_new)
else:
# training with multiple optimizers (e.g. GAN)
outputs_per_optimizer = [None] * len(self.optimizer)
total_step_time = 0
for idx, optimizer in enumerate(self.optimizer):
criterion = self.criterion
# scaler = self.scaler[idx] if self.use_amp_scaler else None
scaler = self.scaler
scheduler = self.scheduler[idx]
outputs, loss_dict_new, step_time = self._optimize(
batch, self.model, optimizer, scaler, criterion, scheduler, self.config, idx
)
# skip the rest if the model returns None
total_step_time += step_time
outputs_per_optimizer[idx] = outputs
# merge loss_dicts from each optimizer
# rename duplicates with the optimizer idx
# if None, model skipped this optimizer
if loss_dict_new is not None:
for k, v in loss_dict_new.items():
if k in loss_dict:
loss_dict[f"{k}-{idx}"] = v
else:
loss_dict[k] = v
step_time = total_step_time
outputs = outputs_per_optimizer
# update avg runtime stats
keep_avg_update = {}
keep_avg_update["avg_loader_time"] = loader_time
keep_avg_update["avg_step_time"] = step_time
self.keep_avg_train.update_values(keep_avg_update)
# update avg loss stats
update_eval_values = {}
for key, value in loss_dict.items():
update_eval_values["avg_" + key] = value
self.keep_avg_train.update_values(update_eval_values)
# print training progress
if self.total_steps_done % self.config.print_step == 0:
# log learning rates
lrs = {}
if isinstance(self.optimizer, list):
for idx, optimizer in enumerate(self.optimizer):
current_lr = self.optimizer[idx].param_groups[0]["lr"]
lrs.update({f"current_lr_{idx}": current_lr})
else:
current_lr = self.optimizer.param_groups[0]["lr"]
lrs = {"current_lr": current_lr}
# log run-time stats
loss_dict.update(lrs)
loss_dict.update(
{
"step_time": round(step_time, 4),
"loader_time": round(loader_time, 4),
}
)
self.c_logger.print_train_step(
batch_n_steps, step, self.total_steps_done, loss_dict, self.keep_avg_train.avg_values
)
if self.args.rank == 0:
# Plot Training Iter Stats
# reduce TB load and don't log every step
if self.total_steps_done % self.config.plot_step == 0:
self.dashboard_logger.train_step_stats(self.total_steps_done, loss_dict)
if self.total_steps_done % self.config.save_step == 0 and self.total_steps_done != 0:
if self.config.checkpoint:
# checkpoint the model
target_avg_loss = self._pick_target_avg_loss(self.keep_avg_train)
save_checkpoint(
self.config,
self.model,
self.optimizer,
self.scaler if self.use_amp_scaler else None,
self.total_steps_done,
self.epochs_done,
self.output_path,
model_loss=target_avg_loss,
)
if self.total_steps_done % self.config.log_model_step == 0:
# log checkpoint as artifact
aliases = [f"epoch-{self.epochs_done}", f"step-{self.total_steps_done}"]
self.dashboard_logger.log_artifact(self.output_path, "checkpoint", "model", aliases)
# training visualizations
if hasattr(self.model, "module") and hasattr(self.model.module, "train_log"):
self.model.module.train_log(
batch, outputs, self.dashboard_logger, self.training_assets, self.total_steps_done
)
elif hasattr(self.model, "train_log"):
self.model.train_log(
batch, outputs, self.dashboard_logger, self.training_assets, self.total_steps_done
)
self.dashboard_logger.flush()
self.total_steps_done += 1
self.callbacks.on_train_step_end(self)
return outputs, loss_dict
def train_epoch(self) -> None:
"""Main entry point for the training loop. Run training on the all training samples."""
# initialize the data loader
self.train_loader = self.get_train_dataloader(
self.training_assets,
self.train_samples,
verbose=True,
)
# set model to training mode
if self.num_gpus > 1:
self.model.module.train()
else:
self.model.train()
epoch_start_time = time.time()
if self.use_cuda:
batch_num_steps = int(len(self.train_loader.dataset) / (self.config.batch_size * self.num_gpus))
else:
batch_num_steps = int(len(self.train_loader.dataset) / self.config.batch_size)
self.c_logger.print_train_start()
loader_start_time = time.time()
# iterate over the training samples
for cur_step, batch in enumerate(self.train_loader):
_, _ = self.train_step(batch, batch_num_steps, cur_step, loader_start_time)
loader_start_time = time.time()
epoch_time = time.time() - epoch_start_time
# plot self.epochs_done Stats
if self.args.rank == 0:
epoch_stats = {"epoch_time": epoch_time}
epoch_stats.update(self.keep_avg_train.avg_values)
self.dashboard_logger.train_epoch_stats(self.total_steps_done, epoch_stats)
if self.config.model_param_stats:
self.logger.model_weights(self.model, self.total_steps_done)
# scheduler step after the epoch
if self.scheduler is not None and self.config.scheduler_after_epoch:
if isinstance(self.scheduler, list):
for scheduler in self.scheduler:
if scheduler is not None:
scheduler.step()
else:
self.scheduler.step()
#######################
# EVAL FUNCTIONS
#######################
@staticmethod
def _model_eval_step(
batch: Dict, model: nn.Module, criterion: nn.Module, optimizer_idx: int = None
) -> Tuple[Dict, Dict]:
"""
Perform a evaluation forward pass. Compute model outputs and losses with no gradients.
Args:
batch (Dict): IBatch of inputs.
model (nn.Module): Model to call evaluation.
criterion (nn.Module): Model criterion.
optimizer_idx (int, optional): Optimizer ID to define the closure in multi-optimizer training. Defaults to None.
Returns:
Tuple[Dict, Dict]: model outputs and losses.
"""
input_args = [batch, criterion]
if optimizer_idx is not None:
input_args.append(optimizer_idx)
if hasattr(model, "module"):
return model.module.eval_step(*input_args)
return model.eval_step(*input_args)
def eval_step(self, batch: Dict, step: int) -> Tuple[Dict, Dict]:
"""Perform a evaluation step on a batch of inputs and log the process.
Args:
batch (Dict): Input batch.
step (int): Current step number in this epoch.
Returns:
Tuple[Dict, Dict]: Model outputs and losses.
"""
with torch.no_grad():
outputs = []
loss_dict = {}
if not isinstance(self.optimizer, list):
outputs, loss_dict = self._model_eval_step(batch, self.model, self.criterion)
else:
outputs = [None] * len(self.optimizer)
for idx, _ in enumerate(self.optimizer):
criterion = self.criterion
outputs_, loss_dict_new = self._model_eval_step(batch, self.model, criterion, idx)
outputs[idx] = outputs_
if loss_dict_new is not None:
loss_dict_new[f"loss_{idx}"] = loss_dict_new.pop("loss")
loss_dict.update(loss_dict_new)
loss_dict = self._detach_loss_dict(loss_dict)
# update avg stats
update_eval_values = {}
for key, value in loss_dict.items():
update_eval_values["avg_" + key] = value
self.keep_avg_eval.update_values(update_eval_values)
if self.config.print_eval:
self.c_logger.print_eval_step(step, loss_dict, self.keep_avg_eval.avg_values)
return outputs, loss_dict
def eval_epoch(self) -> None:
"""Main entry point for the evaluation loop. Run evaluation on the all validation samples."""
self.eval_loader = (
self.get_eval_dataloader(
self.training_assets,
self.eval_samples,
verbose=True,
)
if self.config.run_eval
else None
)
self.model.eval()
self.c_logger.print_eval_start()
loader_start_time = time.time()
batch = None
for cur_step, batch in enumerate(self.eval_loader):
# format data
batch = self.format_batch(batch)
loader_time = time.time() - loader_start_time
self.keep_avg_eval.update_values({"avg_loader_time": loader_time})
outputs, _ = self.eval_step(batch, cur_step)
loader_start_time = time.time()
# plot epoch stats, artifacts and figures
if self.args.rank == 0:
if hasattr(self.model, "module") and hasattr(self.model.module, "eval_log"):
self.model.module.eval_log(
batch, outputs, self.dashboard_logger, self.training_assets, self.total_steps_done
)
elif hasattr(self.model, "eval_log"):
self.model.eval_log(batch, outputs, self.dashboard_logger, self.training_assets, self.total_steps_done)
self.dashboard_logger.eval_stats(self.total_steps_done, self.keep_avg_eval.avg_values)
def test_run(self) -> None:
"""Run test and log the results. Test run must be defined by the model.
Model must return figures and audios to be logged by the Tensorboard."""
if hasattr(self.model, "test_run") or (self.num_gpus > 1 and hasattr(self.model.module, "test_run")):
if self.eval_loader is None:
self.eval_loader = self.get_eval_dataloader(
self.training_assets,
self.eval_samples,
verbose=True,
)
if hasattr(self.eval_loader.dataset, "load_test_samples"):
samples = self.eval_loader.dataset.load_test_samples(1)
if self.num_gpus > 1:
figures, audios = self.model.module.test_run(self.training_assets, samples, None)
else:
figures, audios = self.model.test_run(self.training_assets, samples, None)
else:
if self.num_gpus > 1:
figures, audios = self.model.module.test_run(self.training_assets)
else:
figures, audios = self.model.test_run(self.training_assets)
self.dashboard_logger.test_audios(self.total_steps_done, audios, self.config.audio["sample_rate"])
self.dashboard_logger.test_figures(self.total_steps_done, figures)
def _restore_best_loss(self):
"""Restore the best loss from the args.best_path if provided else
from the model (`args.restore_path` or `args.continue_path`) used for resuming the training"""
if self.restore_step != 0 or self.args.best_path:
print(f" > Restoring best loss from {os.path.basename(self.args.best_path)} ...")
ch = load_fsspec(self.args.restore_path, map_location="cpu")
if "model_loss" in ch:
self.best_loss = ch["model_loss"]
print(f" > Starting with loaded last best loss {self.best_loss}.")
###################################
# FIT FUNCTIONS
###################################
def _fit(self) -> None:
"""🏃 train -> evaluate -> test for the number of epochs."""
self._restore_best_loss()
self.total_steps_done = self.restore_step
for epoch in range(0, self.config.epochs):
if self.num_gpus > 1:
# let all processes sync up before starting with a new epoch of training
dist.barrier()
self.callbacks.on_epoch_start(self)
self.keep_avg_train = KeepAverage()
self.keep_avg_eval = KeepAverage() if self.config.run_eval else None
self.epochs_done = epoch
self.c_logger.print_epoch_start(epoch, self.config.epochs, self.output_path)
if not self.args.skip_train_epoch:
self.train_epoch()
if self.config.run_eval:
self.eval_epoch()
if epoch >= self.config.test_delay_epochs and self.args.rank <= 0:
self.test_run()
self.c_logger.print_epoch_end(
epoch, self.keep_avg_eval.avg_values if self.config.run_eval else self.keep_avg_train.avg_values
)
if self.args.rank in [None, 0]:
self.save_best_model()
self.callbacks.on_epoch_end(self)
def fit(self) -> None:
"""Where the ✨️magic✨️ happens..."""
try:
self._fit()
if self.args.rank == 0:
self.dashboard_logger.finish()
except KeyboardInterrupt:
self.callbacks.on_keyboard_interrupt(self)
# if the output folder is empty remove the run.
remove_experiment_folder(self.output_path)
# clear the DDP processes
if self.num_gpus > 1:
dist.destroy_process_group()
# finish the wandb run and sync data
if self.args.rank == 0:
self.dashboard_logger.finish()
# stop without error signal
try:
sys.exit(0)
except SystemExit:
os._exit(0) # pylint: disable=protected-access
except BaseException: # pylint: disable=broad-except
remove_experiment_folder(self.output_path)
traceback.print_exc()
sys.exit(1)
def save_best_model(self) -> None:
"""Save the best model. It only saves if the current target loss is smaller then the previous."""
# set the target loss to choose the best model
target_loss_dict = self._pick_target_avg_loss(self.keep_avg_eval if self.keep_avg_eval else self.keep_avg_train)
# save the model and update the best_loss
self.best_loss = save_best_model(
target_loss_dict,
self.best_loss,
self.config,
self.model,
self.optimizer,
self.scaler if self.use_amp_scaler else None,
self.total_steps_done,
self.epochs_done,
self.output_path,
keep_all_best=self.config.keep_all_best,
keep_after=self.config.keep_after,
)
#####################
# GET FUNCTIONS
#####################
@staticmethod
def get_optimizer(model: nn.Module, config: Coqpit) -> Union[torch.optim.Optimizer, List]:
"""Receive the optimizer from the model if model implements `get_optimizer()` else
check the optimizer parameters in the config and try initiating the optimizer.
Args:
model (nn.Module): Training model.
config (Coqpit): Training configuration.
Returns:
Union[torch.optim.Optimizer, List]: A optimizer or a list of optimizers. GAN models define a list.
"""
if hasattr(model, "get_optimizer"):
optimizer = model.get_optimizer()
if optimizer is None:
optimizer_name = config.optimizer
optimizer_params = config.optimizer_params
return get_optimizer(optimizer_name, optimizer_params, config.lr, model)
return optimizer
@staticmethod
def get_lr(model: nn.Module, config: Coqpit) -> Union[float, List[float]]:
"""Set the initial learning rate by the model if model implements `get_lr()` else try setting the learning rate
fromthe config.
Args:
model (nn.Module): Training model.
config (Coqpit): Training configuration.
Returns:
Union[float, List[float]]: A single learning rate or a list of learning rates, one for each optimzier.
"""
lr = None
if hasattr(model, "get_lr"):
lr = model.get_lr()
if lr is None:
lr = config.lr
return lr
@staticmethod
def get_scheduler(
model: nn.Module, config: Coqpit, optimizer: Union[torch.optim.Optimizer, List]
) -> Union[torch.optim.lr_scheduler._LRScheduler, List]: # pylint: disable=protected-access
"""Receive the scheduler from the model if model implements `get_scheduler()` else
check the config and try initiating the scheduler.
Args:
model (nn.Module): Training model.
config (Coqpit): Training configuration.
Returns:
Union[torch.optim.Optimizer, List]: A scheduler or a list of schedulers, one for each optimizer.
"""
scheduler = None
if hasattr(model, "get_scheduler"):
scheduler = model.get_scheduler(optimizer)
if scheduler is None:
lr_scheduler = config.lr_scheduler
lr_scheduler_params = config.lr_scheduler_params
return get_scheduler(lr_scheduler, lr_scheduler_params, optimizer)
return scheduler
@staticmethod
def get_criterion(model: nn.Module) -> nn.Module:
"""Receive the criterion from the model. Model must implement `get_criterion()`.
Args:
model (nn.Module): Training model.
Returns:
nn.Module: Criterion layer.
"""
criterion = None
criterion = model.get_criterion()
return criterion
####################
# HELPER FUNCTIONS
####################
@staticmethod
def _detach_loss_dict(loss_dict: Dict) -> Dict:
"""Detach loss values from autograp.
Args:
loss_dict (Dict): losses.
Returns:
Dict: losses detached from autograph.
"""
loss_dict_detached = {}
for key, value in loss_dict.items():
if isinstance(value, (int, float)):
loss_dict_detached[key] = value
else:
loss_dict_detached[key] = value.detach().item()
return loss_dict_detached
def _pick_target_avg_loss(self, keep_avg_target: KeepAverage) -> Dict:
"""Pick the target loss to compare models"""
target_avg_loss = None
# return if target loss defined in the model config
if "target_loss" in self.config and self.config.target_loss:
return keep_avg_target[f"avg_{self.config.target_loss}"]
# take the average of loss_{optimizer_idx} as the target loss when there are multiple optimizers
if isinstance(self.optimizer, list):
target_avg_loss = 0
for idx in range(len(self.optimizer)):
target_avg_loss += keep_avg_target[f"avg_loss_{idx}"]
target_avg_loss /= len(self.optimizer)
else:
target_avg_loss = keep_avg_target["avg_loss"]
return target_avg_loss
def _setup_logger_config(self, log_file: str) -> None:
"""Write log strings to a file and print logs to the terminal.
TODO: Causes formatting issues in pdb debugging."""
class Logger(object):
def __init__(self, print_to_terminal=True):
self.print_to_terminal = print_to_terminal
self.terminal = sys.stdout
self.log_file = log_file
def write(self, message):
if self.print_to_terminal:
self.terminal.write(message)
with open(self.log_file, "a", encoding="utf-8") as f:
f.write(message)
def flush(self):
# this flush method is needed for python 3 compatibility.
# this handles the flush command by doing nothing.
# you might want to specify some extra behavior here.
pass
# don't let processes rank > 0 write to the terminal
sys.stdout = Logger(self.args.rank == 0)
@staticmethod
def _is_apex_available() -> bool:
"""Check if Nvidia's APEX is available."""
return importlib.util.find_spec("apex") is not None