graphwar.training¶

`Trainer`	A simple trainer to train graph neural network models conveniently.
`get_trainer`	Get the default trainer using str or a model in `graphwar.nn.models`
`RobustGCNTrainer`	Custom trainer for `graphwar.nn.models.RobustGCN`
`SimPGCNTrainer`	Custom trainer for `graphwar.nn.models.SimPGCN`
`SATTrainer`	Custom trainer for `graphwar.nn.models.SAT`

class Trainer(model: Module, device: Union[str, device] = 'cpu', **cfg)[source]¶

A simple trainer to train graph neural network models conveniently.

Parameters

model (nn.Module) – the model used for training
device (Union[str, torch.device], optional) – the device used for training, by default ‘cpu’
cfg (other keyword arguments, such as lr and weight_decay.) –

Example

>>> from graphwar.training import Trainer
>>> model = ... # your model
>>> trainer = Trainer(model, device='cuda')

>>> data # PyG-like data, e.g., Cora
Data(x=[2485, 1433], edge_index=[2, 10138], y=[2485])

>>> # simple training
>>> trainer.fit({'data': data, 'mask': train_mask})

>>> # train with model picking
>>> from graphwar.training import ModelCheckpoint
>>> cb = ModelCheckpoint('my_ckpt', monitor='val_acc')
>>> trainer.fit({'data': data, 'mask': train_mask},
{'data': data, 'mask': val_mask}, callbacks=[cb])

>>> # get training logs
>>> history = trainer.model.history

>>> trainer.evaluate({'data': data, 'mask': data.test_mask}) # evaluation

>>> predict = trainer.predict({'data': data, 'mask': your_mask}) # prediction

fit(train_inputs, val_inputs: Optional[dict] = None, callbacks: Optional[Callback] = None, verbose: Optional[int] = 1, epochs: int = 100) → Trainer[source]¶

Simple training method designed for :attr:model

Parameters

train_inputs (dict like or custom inputs) – training data. It is used for train_step.
val_inputs (Optional[dict]) – used for validation.
callbacks (Optional[Callback], optional) – callbacks used for training, see graphwar.training.callbacks, by default None
verbose (Optional[int], optional) – verbosity during training, can be: None, 1, 2, 3, 4, by default 1
epochs (int, optional) – training epochs, by default 100

Example

>>> # simple training
>>> trainer.fit({'data': data, 'mask': train_mask})

>>> # train with model picking
>>> from graphwar.training import ModelCheckpoint
>>> cb = ModelCheckpoint('my_ckpt', monitor='val_acc')
>>> trainer.fit({'data': data, 'mask': train_mask},
{'data': data, 'mask': val_mask}, callbacks=[cb])

train_step(inputs: dict) → dict[source]¶

One-step training on the inputs.

Parameters: inputs (dict like or custom inputs) – the training data.
Returns: the output logs, including loss and acc, etc.
Return type: dict

evaluate(inputs: dict, verbose: Optional[int] = 1) → BunchDict[source]¶

Simple evaluation step for :attr:model

Parameters

inputs (dict like or custom inputs) – test data, it is used for test_step.
verbose (Optional[int], optional) – verbosity during evaluation, by default 1

Returns

the dict-like output logs

Return type

BunchDict

Example

>>> trainer.evaluate({'data': data, 'mask': data.test_mask}) # evaluation

test_step(inputs: dict) → dict[source]¶

One-step evaluation on the inputs.

Parameters: inputs (dict like or custom inputs) – the testing data.
Returns: the output logs, including loss and acc, etc.
Return type: dict

predict_step(inputs: dict) → Tensor[source]¶

One-step prediction on the inputs.

Parameters: inputs (dict like or custom inputs) – the prediction data.
Returns: the output prediction.
Return type: Tensor

predict(inputs: dict, transform: Callable = Softmax(dim=- 1)) → Tensor[source]¶

Parameters

inputs (dict like or custom inputs) – predict data, it is used for predict_step
transform (Callable) – Callable function applied on output predictions.

Example

>>> predict = trainer.predict({'data': data, 'mask': mask_or_not_given}) # prediction

config_optimizer() → Optimizer[source]¶

config_scheduler(optimizer: Optimizer)[source]¶

config_callbacks(verbose, epochs, callbacks=None) → CallbackList[source]¶

property model¶

cache_clear() → Trainer[source]¶: Clear cached inputs or intermediate results of the model.

extra_repr() → str[source]¶

get_trainer(model: Union[str, Module]) → Trainer[source]¶

Get the default trainer using str or a model in graphwar.nn.models

Parameters: model (Union[str, torch.nn.Module]) – the model to be trained in
Return type: Custom trainer or default trainer graphwar.training.Trainer for the model.

Examples

>>> import graphwar
>>> graphwar.training.get_trainer('GCN')
graphwar.training.trainer.Trainer

>>> from graphwar.nn.models import GCN
>>> graphwar.training.get_trainer(GCN)
graphwar.training.trainer.Trainer

>>> # by default, it returns `graphwar.training.Trainer`
>>> graphwar.training.get_trainer('unimplemeted_model')
graphwar.training.trainer.Trainer

>>> graphwar.training.get_trainer('RobustGCN')
graphwar.training.robustgcn_trainer.RobustGCNTrainer

>>> # it is case-sensitive
>>> graphwar.training.get_trainer('robustGCN')
graphwar.training.trainer.Trainer

class RobustGCNTrainer(model: Module, device: Union[str, device] = 'cpu', **cfg)[source]¶

Custom trainer for graphwar.nn.models.RobustGCN

Parameters

model (nn.Module) – the model used for training
device (Union[str, torch.device], optional) – the device used for training, by default ‘cpu’
cfg (other keyword arguments, such as lr and weight_decay.) –

Note

graphwar.training.RobustGCNTrainer accepts the following additional arguments:

kl: trade-off parameter for kl loss

train_step(inputs: dict) → dict[source]¶

One-step training on the input dataloader.

Parameters: inputs (dict) – the training data.
Returns: the output logs, including loss and val_acc, etc.
Return type: dict

class SimPGCNTrainer(model: Module, device: Union[str, device] = 'cpu', **cfg)[source]¶

Custom trainer for graphwar.nn.models.SimPGCN

Parameters

model (nn.Module) – the model used for training
device (Union[str, torch.device], optional) – the device used for training, by default ‘cpu’
cfg (other keyword arguments, such as lr and weight_decay.) –

Note

graphwar.training.SimPGCNTrainer accepts the following additional arguments:

lambda_: trade-off parameter for regression loss

train_step(inputs: dict) → dict[source]¶

One-step training on the input dataloader.

Parameters: inputs (dict) – the training data.
Returns: the output logs, including loss and val_acc, etc.
Return type: dict

class SATTrainer(model: Module, device: Union[str, device] = 'cpu', **cfg)[source]¶

Custom trainer for graphwar.nn.models.SAT

Parameters

model (nn.Module) – the model used for training
device (Union[str, torch.device], optional) – the device used for training, by default ‘cpu’
cfg (other keyword arguments, such as lr and weight_decay.) –

Note

graphwar.training.SATTrainer accepts the following additional arguments:

eps_U: scale of perturbation on eigenvectors
eps_V: scale of perturbation on eigenvalues
lambda_U: trade-off parameters for eigenvectors-specific loss
lambda_V: trade-off parameters for eigenvalues-specific loss

train_step(inputs: dict) → dict[source]¶

One-step training on the input dataloader.

Parameters: inputs (dict) – the training data.
Returns: the output logs, including loss and val_acc, etc.
Return type: dict