convlab2.e2e.rnn_rollout.engines package¶

Submodules¶

convlab2.e2e.rnn_rollout.engines.engine module¶

Training utilities.

class convlab2.e2e.rnn_rollout.engines.engine.Criterion(dictionary, device_id=None, bad_toks=[], reduction='mean')¶

Bases: object

Weighted CrossEntropyLoss.

class convlab2.e2e.rnn_rollout.engines.engine.Engine(model, args, device_id=None, verbose=False)¶

Bases: object

The training engine.

Performs training and evaluation.

forward(batch, requires_grad=False)¶: A helper function to perform a forward pass on a batch.

get_model()¶: Extracts the model.

iter(N, epoch, lr, traindata, validdata)¶: Performs on iteration of the training. Runs one epoch on the training and validation datasets.

train(corpus)¶: Entry point.

train_pass(N, trainset)¶: Training pass.

train_single(N, trainset)¶: A helper function to train on a random batch.

valid_pass(N, validset, validset_stats)¶: Validation pass.

convlab2.e2e.rnn_rollout.engines.latent_clustering_engine module¶

class convlab2.e2e.rnn_rollout.engines.latent_clustering_engine.BaselineClusteringEngine(model, args, verbose=False)¶

Bases: convlab2.e2e.rnn_rollout.engines.EngineBase

train_batch(batch)¶

train_pass(trainset)¶

valid_batch(batch)¶

valid_pass(validset, validset_stats)¶

class convlab2.e2e.rnn_rollout.engines.latent_clustering_engine.LatentClusteringEngine(model, args, verbose=False)¶

Bases: convlab2.e2e.rnn_rollout.engines.EngineBase

combine_loss(lang_loss, select_loss)¶

train_batch(batch)¶

train_pass(trainset)¶

valid_batch(batch)¶

valid_pass(validset, validset_stats)¶

class convlab2.e2e.rnn_rollout.engines.latent_clustering_engine.LatentClusteringLanguageEngine(model, args, verbose=False)¶

Bases: convlab2.e2e.rnn_rollout.engines.EngineBase

train_batch(batch)¶

valid_batch(batch)¶

class convlab2.e2e.rnn_rollout.engines.latent_clustering_engine.LatentClusteringPredictionEngine(model, args, verbose=False)¶

Bases: convlab2.e2e.rnn_rollout.engines.EngineBase

train_batch(batch)¶

train_pass(trainset)¶

valid_batch(batch)¶

valid_pass(validset, validset_stats)¶

convlab2.e2e.rnn_rollout.engines.rnn_engine module¶

class convlab2.e2e.rnn_rollout.engines.rnn_engine.RnnEngine(model, args, verbose=False)¶

Bases: convlab2.e2e.rnn_rollout.engines.EngineBase

train_batch(batch)¶

valid_batch(batch)¶

convlab2.e2e.rnn_rollout.engines.selection_engine module¶

class convlab2.e2e.rnn_rollout.engines.selection_engine.SelectionEngine(model, args, verbose=False)¶

Bases: convlab2.e2e.rnn_rollout.engines.EngineBase

train_batch(batch)¶

valid_batch(batch)¶

Module contents¶

class convlab2.e2e.rnn_rollout.engines.Criterion(dictionary, device_id=None, bad_toks=[], reduction='mean')¶

Bases: object

Weighted CrossEntropyLoss.

class convlab2.e2e.rnn_rollout.engines.EngineBase(model, args, verbose=False)¶

Bases: object

Base class for training engine.

combine_loss(lang_loss, select_loss)¶

get_model()¶

iter(epoch, lr, traindata, validdata)¶

make_opt(lr)¶

train(corpus)¶

train_batch(batch)¶

train_pass(trainset)¶

valid_batch(batch)¶

valid_pass(validset, validset_stats)¶