vikigenius/conditional_text_generation

Adversarial Latent Space model for dialog generation

conditional_text_generation's Issues

GAN model cannot converge.

I found that the GAN model cannot converge during training. I followed the data format in tests/fixtures folder to create the train_setences.tsv file and train_dialog.tsv file from the de-duplicated version of DailyDialog dataset which is proposed by Bahuleyan et al. in their GitHub https://github.com/HareeshBahuleyan/probabilistic_nlg/tree/master/dialog/data/DailyDial/de_duplicated. I did not change any code and the VAE model is well trained whereas the GAN model will early stop after training 5 epochs. The training loss for GAN model did not decrease and the GAN model can only produce some meaningless sentences.

Are there any suggestions for solving this?

Update Codebase for compatibility with allennlp changes to callbacks

Allennlp has made significant changes to their codebase for callbacks in master. I will have to take some time to update my code to take care of it.

As a temporary workaround use the following commit: allenai/allennlp@9b929b2 to install using pip.

RuntimeError: cuDNN error: CUDNN_STATUS_EXECUTION_FAILED

I am getting this error when I run allennlp train experiments/vae.jsonnet -s models/dialog_vae --include-package src I tried using different GPU IDs but it didn't work.

2019-07-21 17:45:17,012 - INFO - pytorch_pretrained_bert.modeling - Better speed can be achieved with apex installed from https://www.github.com/nvidia/apex .
2019-07-21 17:45:17,665 - INFO - allennlp.common.params - random_seed = 0
2019-07-21 17:45:17,665 - INFO - allennlp.common.params - numpy_seed = 0
2019-07-21 17:45:17,665 - INFO - allennlp.common.params - pytorch_seed = 0
2019-07-21 17:45:17,720 - INFO - allennlp.common.checks - Pytorch version: 1.1.0
2019-07-21 17:45:17,722 - INFO - allennlp.common.params - evaluate_on_test = False
2019-07-21 17:45:17,722 - INFO - allennlp.common.params - trainer.type = callback
2019-07-21 17:45:17,722 - INFO - allennlp.common.params - validation_dataset_reader = None
2019-07-21 17:45:17,722 - INFO - allennlp.common.from_params - instantiating class <class 'allennlp.data.dataset_readers.dataset_reader.DatasetReader'> from params {'type': 'autoencoder'} and extras set()
2019-07-21 17:45:17,722 - INFO - allennlp.common.params - dataset_reader.type = autoencoder
2019-07-21 17:45:17,722 - INFO - allennlp.common.from_params - instantiating class <class 'src.data.dataset.AutoencoderDatasetReader'> from params {} and extras set()
2019-07-21 17:45:17,722 - INFO - allennlp.common.params - dataset_reader.source_token_indexers = <allennlp.common.params.Params object at 0x7fa15c5736d8>
2019-07-21 17:45:17,722 - INFO - allennlp.common.params - dataset_reader.target_token_indexers = <allennlp.common.params.Params object at 0x7fa15c5736d8>
2019-07-21 17:45:17,722 - INFO - allennlp.common.params - dataset_reader.source_add_start_token = True
2019-07-21 17:45:17,722 - INFO - allennlp.common.params - dataset_reader.delimiter = 	
2019-07-21 17:45:17,722 - INFO - allennlp.common.params - dataset_reader.lazy = False
2019-07-21 17:45:17,806 - INFO - allennlp.common.params - train_data_path = data/interim/dialog/train_sentences.tsv
2019-07-21 17:45:17,806 - INFO - allennlp.training.util - Reading training data from data/interim/dialog/train_sentences.tsv
0it [00:00, ?it/s]2019-07-21 17:45:17,806 - INFO - src.data.dataset - Reading instances from lines in file at: data/interim/dialog/train_sentences.tsv
152104it [00:26, 5848.90it/s]
2019-07-21 17:45:43,812 - INFO - allennlp.common.params - validation_data_path = data/interim/dialog/valid_sentences.tsv
2019-07-21 17:45:43,812 - INFO - allennlp.training.util - Reading validation data from data/interim/dialog/valid_sentences.tsv
0it [00:00, ?it/s]2019-07-21 17:45:43,812 - INFO - src.data.dataset - Reading instances from lines in file at: data/interim/dialog/valid_sentences.tsv
8932it [00:01, 6413.16it/s]
2019-07-21 17:45:45,205 - INFO - allennlp.common.params - test_data_path = None
2019-07-21 17:45:45,381 - INFO - allennlp.training.trainer_pieces - From dataset instances, train, validation will be considered for vocabulary creation.
2019-07-21 17:45:45,381 - INFO - allennlp.common.params - vocabulary.type = None
2019-07-21 17:45:45,381 - INFO - allennlp.common.params - vocabulary.extend = False
2019-07-21 17:45:45,381 - INFO - allennlp.common.params - vocabulary.directory_path = None
2019-07-21 17:45:45,381 - INFO - allennlp.common.params - vocabulary.min_count = None
2019-07-21 17:45:45,381 - INFO - allennlp.common.params - vocabulary.max_vocab_size = None
2019-07-21 17:45:45,381 - INFO - allennlp.common.params - vocabulary.non_padded_namespaces = ('*tags', '*labels')
2019-07-21 17:45:45,382 - INFO - allennlp.common.params - vocabulary.min_pretrained_embeddings = None
2019-07-21 17:45:45,382 - INFO - allennlp.common.params - vocabulary.only_include_pretrained_words = False
2019-07-21 17:45:45,382 - INFO - allennlp.common.params - vocabulary.tokens_to_add = None
2019-07-21 17:45:45,382 - INFO - allennlp.data.vocabulary - Fitting token dictionary from dataset.
161036it [00:01, 85823.61it/s]
2019-07-21 17:45:47,290 - INFO - allennlp.common.from_params - instantiating class <class 'allennlp.models.model.Model'> from params {'decoder': {'latent_dim': 128, 'rnn': {'hidden_size': 512, 'input_size': 428, 'num_layers': 1, 'type': 'lstm'}, 'target_embedder': {'token_embedders': {'tokens': {'embedding_dim': 300, 'type': 'embedding', 'vocab_namespace': 'tokens'}}}, 'type': 'variational_decoder'}, 'initializer': [['.*', {'a': -0.1, 'b': 0.1, 'type': 'uniform'}]], 'kl_weight': {'max_weight': 0.5, 'min_weight': 0, 'num_iter_to_max': 3500, 'slope': 0.5, 'type': 'sigmoid_annealed', 'warmup': 500}, 'temperature': 1e-05, 'type': 'vae', 'variational_encoder': {'encoder': {'bidirectional': True, 'hidden_size': 512, 'input_size': 300, 'num_layers': 1, 'type': 'lstm'}, 'latent_dim': 128, 'text_field_embedder': {'token_embedders': {'tokens': {'embedding_dim': 300, 'type': 'embedding', 'vocab_namespace': 'tokens'}}}, 'type': 'gaussian'}} and extras {'vocab'}
2019-07-21 17:45:47,290 - INFO - allennlp.common.params - model.type = vae
2019-07-21 17:45:47,290 - INFO - allennlp.common.from_params - instantiating class <class 'src.models.vae.VAE'> from params {'decoder': {'latent_dim': 128, 'rnn': {'hidden_size': 512, 'input_size': 428, 'num_layers': 1, 'type': 'lstm'}, 'target_embedder': {'token_embedders': {'tokens': {'embedding_dim': 300, 'type': 'embedding', 'vocab_namespace': 'tokens'}}}, 'type': 'variational_decoder'}, 'initializer': [['.*', {'a': -0.1, 'b': 0.1, 'type': 'uniform'}]], 'kl_weight': {'max_weight': 0.5, 'min_weight': 0, 'num_iter_to_max': 3500, 'slope': 0.5, 'type': 'sigmoid_annealed', 'warmup': 500}, 'temperature': 1e-05, 'variational_encoder': {'encoder': {'bidirectional': True, 'hidden_size': 512, 'input_size': 300, 'num_layers': 1, 'type': 'lstm'}, 'latent_dim': 128, 'text_field_embedder': {'token_embedders': {'tokens': {'embedding_dim': 300, 'type': 'embedding', 'vocab_namespace': 'tokens'}}}, 'type': 'gaussian'}} and extras {'vocab'}
2019-07-21 17:45:47,290 - INFO - allennlp.common.from_params - instantiating class <class 'src.modules.encoders.variational_encoder.VariationalEncoder'> from params {'encoder': {'bidirectional': True, 'hidden_size': 512, 'input_size': 300, 'num_layers': 1, 'type': 'lstm'}, 'latent_dim': 128, 'text_field_embedder': {'token_embedders': {'tokens': {'embedding_dim': 300, 'type': 'embedding', 'vocab_namespace': 'tokens'}}}, 'type': 'gaussian'} and extras {'vocab'}
2019-07-21 17:45:47,290 - INFO - allennlp.common.params - model.variational_encoder.type = gaussian
2019-07-21 17:45:47,290 - INFO - allennlp.common.from_params - instantiating class <class 'src.modules.encoders.gaussian_encoder.GaussianEncoder'> from params {'encoder': {'bidirectional': True, 'hidden_size': 512, 'input_size': 300, 'num_layers': 1, 'type': 'lstm'}, 'latent_dim': 128, 'text_field_embedder': {'token_embedders': {'tokens': {'embedding_dim': 300, 'type': 'embedding', 'vocab_namespace': 'tokens'}}}} and extras {'vocab'}
2019-07-21 17:45:47,290 - INFO - allennlp.common.from_params - instantiating class <class 'allennlp.modules.text_field_embedders.text_field_embedder.TextFieldEmbedder'> from params {'token_embedders': {'tokens': {'embedding_dim': 300, 'type': 'embedding', 'vocab_namespace': 'tokens'}}} and extras {'vocab'}
2019-07-21 17:45:47,290 - INFO - allennlp.common.params - model.variational_encoder.text_field_embedder.type = basic
2019-07-21 17:45:47,291 - INFO - allennlp.common.params - model.variational_encoder.text_field_embedder.embedder_to_indexer_map = None
2019-07-21 17:45:47,291 - INFO - allennlp.common.params - model.variational_encoder.text_field_embedder.allow_unmatched_keys = False
2019-07-21 17:45:47,291 - INFO - allennlp.common.from_params - instantiating class <class 'allennlp.modules.token_embedders.token_embedder.TokenEmbedder'> from params {'embedding_dim': 300, 'type': 'embedding', 'vocab_namespace': 'tokens'} and extras {'vocab'}
2019-07-21 17:45:47,291 - INFO - allennlp.common.params - model.variational_encoder.text_field_embedder.token_embedders.tokens.type = embedding
2019-07-21 17:45:47,291 - INFO - allennlp.common.params - model.variational_encoder.text_field_embedder.token_embedders.tokens.num_embeddings = None
2019-07-21 17:45:47,291 - INFO - allennlp.common.params - model.variational_encoder.text_field_embedder.token_embedders.tokens.vocab_namespace = tokens
2019-07-21 17:45:47,291 - INFO - allennlp.common.params - model.variational_encoder.text_field_embedder.token_embedders.tokens.embedding_dim = 300
2019-07-21 17:45:47,291 - INFO - allennlp.common.params - model.variational_encoder.text_field_embedder.token_embedders.tokens.pretrained_file = None
2019-07-21 17:45:47,291 - INFO - allennlp.common.params - model.variational_encoder.text_field_embedder.token_embedders.tokens.projection_dim = None
2019-07-21 17:45:47,291 - INFO - allennlp.common.params - model.variational_encoder.text_field_embedder.token_embedders.tokens.trainable = True
2019-07-21 17:45:47,291 - INFO - allennlp.common.params - model.variational_encoder.text_field_embedder.token_embedders.tokens.padding_index = None
2019-07-21 17:45:47,291 - INFO - allennlp.common.params - model.variational_encoder.text_field_embedder.token_embedders.tokens.max_norm = None
2019-07-21 17:45:47,291 - INFO - allennlp.common.params - model.variational_encoder.text_field_embedder.token_embedders.tokens.norm_type = 2.0
2019-07-21 17:45:47,291 - INFO - allennlp.common.params - model.variational_encoder.text_field_embedder.token_embedders.tokens.scale_grad_by_freq = False
2019-07-21 17:45:47,291 - INFO - allennlp.common.params - model.variational_encoder.text_field_embedder.token_embedders.tokens.sparse = False
2019-07-21 17:45:47,327 - INFO - allennlp.common.from_params - instantiating class <class 'allennlp.modules.seq2vec_encoders.seq2vec_encoder.Seq2VecEncoder'> from params {'bidirectional': True, 'hidden_size': 512, 'input_size': 300, 'num_layers': 1, 'type': 'lstm'} and extras {'vocab'}
2019-07-21 17:45:47,328 - INFO - allennlp.common.params - model.variational_encoder.encoder.type = lstm
2019-07-21 17:45:47,328 - INFO - allennlp.common.params - model.variational_encoder.encoder.batch_first = True
2019-07-21 17:45:47,328 - INFO - allennlp.common.params - Converting Params object to dict; logging of default values will not occur when dictionary parameters are used subsequently.
2019-07-21 17:45:47,328 - INFO - allennlp.common.params - CURRENTLY DEFINED PARAMETERS: 
2019-07-21 17:45:47,328 - INFO - allennlp.common.params - model.variational_encoder.encoder.bidirectional = True
2019-07-21 17:45:47,328 - INFO - allennlp.common.params - model.variational_encoder.encoder.hidden_size = 512
2019-07-21 17:45:47,328 - INFO - allennlp.common.params - model.variational_encoder.encoder.input_size = 300
2019-07-21 17:45:47,328 - INFO - allennlp.common.params - model.variational_encoder.encoder.num_layers = 1
2019-07-21 17:45:47,328 - INFO - allennlp.common.params - model.variational_encoder.encoder.batch_first = True
2019-07-21 17:45:47,347 - INFO - allennlp.common.params - model.variational_encoder.latent_dim = 128
2019-07-21 17:45:47,349 - INFO - allennlp.common.from_params - instantiating class <class 'src.modules.decoders.decoder.Decoder'> from params {'latent_dim': 128, 'rnn': {'hidden_size': 512, 'input_size': 428, 'num_layers': 1, 'type': 'lstm'}, 'target_embedder': {'token_embedders': {'tokens': {'embedding_dim': 300, 'type': 'embedding', 'vocab_namespace': 'tokens'}}}, 'type': 'variational_decoder'} and extras {'vocab'}
2019-07-21 17:45:47,349 - INFO - allennlp.common.params - model.decoder.type = variational_decoder
2019-07-21 17:45:47,349 - INFO - allennlp.common.from_params - instantiating class <class 'src.modules.decoders.variational_decoder.VariationalDecoder'> from params {'latent_dim': 128, 'rnn': {'hidden_size': 512, 'input_size': 428, 'num_layers': 1, 'type': 'lstm'}, 'target_embedder': {'token_embedders': {'tokens': {'embedding_dim': 300, 'type': 'embedding', 'vocab_namespace': 'tokens'}}}} and extras {'vocab'}
2019-07-21 17:45:47,349 - INFO - allennlp.common.from_params - instantiating class <class 'allennlp.modules.text_field_embedders.text_field_embedder.TextFieldEmbedder'> from params {'token_embedders': {'tokens': {'embedding_dim': 300, 'type': 'embedding', 'vocab_namespace': 'tokens'}}} and extras {'vocab'}
2019-07-21 17:45:47,349 - INFO - allennlp.common.params - model.decoder.target_embedder.type = basic
2019-07-21 17:45:47,349 - INFO - allennlp.common.params - model.decoder.target_embedder.embedder_to_indexer_map = None
2019-07-21 17:45:47,349 - INFO - allennlp.common.params - model.decoder.target_embedder.allow_unmatched_keys = False
2019-07-21 17:45:47,350 - INFO - allennlp.common.from_params - instantiating class <class 'allennlp.modules.token_embedders.token_embedder.TokenEmbedder'> from params {'embedding_dim': 300, 'type': 'embedding', 'vocab_namespace': 'tokens'} and extras {'vocab'}
2019-07-21 17:45:47,350 - INFO - allennlp.common.params - model.decoder.target_embedder.token_embedders.tokens.type = embedding
2019-07-21 17:45:47,350 - INFO - allennlp.common.params - model.decoder.target_embedder.token_embedders.tokens.num_embeddings = None
2019-07-21 17:45:47,350 - INFO - allennlp.common.params - model.decoder.target_embedder.token_embedders.tokens.vocab_namespace = tokens
2019-07-21 17:45:47,350 - INFO - allennlp.common.params - model.decoder.target_embedder.token_embedders.tokens.embedding_dim = 300
2019-07-21 17:45:47,350 - INFO - allennlp.common.params - model.decoder.target_embedder.token_embedders.tokens.pretrained_file = None
2019-07-21 17:45:47,350 - INFO - allennlp.common.params - model.decoder.target_embedder.token_embedders.tokens.projection_dim = None
2019-07-21 17:45:47,350 - INFO - allennlp.common.params - model.decoder.target_embedder.token_embedders.tokens.trainable = True
2019-07-21 17:45:47,350 - INFO - allennlp.common.params - model.decoder.target_embedder.token_embedders.tokens.padding_index = None
2019-07-21 17:45:47,350 - INFO - allennlp.common.params - model.decoder.target_embedder.token_embedders.tokens.max_norm = None
2019-07-21 17:45:47,350 - INFO - allennlp.common.params - model.decoder.target_embedder.token_embedders.tokens.norm_type = 2.0
2019-07-21 17:45:47,350 - INFO - allennlp.common.params - model.decoder.target_embedder.token_embedders.tokens.scale_grad_by_freq = False
2019-07-21 17:45:47,350 - INFO - allennlp.common.params - model.decoder.target_embedder.token_embedders.tokens.sparse = False
2019-07-21 17:45:47,386 - INFO - allennlp.common.from_params - instantiating class <class 'allennlp.modules.seq2seq_encoders.seq2seq_encoder.Seq2SeqEncoder'> from params {'hidden_size': 512, 'input_size': 428, 'num_layers': 1, 'type': 'lstm'} and extras {'vocab'}
2019-07-21 17:45:47,386 - INFO - allennlp.common.params - model.decoder.rnn.type = lstm
2019-07-21 17:45:47,386 - INFO - allennlp.common.params - model.decoder.rnn.batch_first = True
2019-07-21 17:45:47,386 - INFO - allennlp.common.params - model.decoder.rnn.stateful = False
2019-07-21 17:45:47,386 - INFO - allennlp.common.params - Converting Params object to dict; logging of default values will not occur when dictionary parameters are used subsequently.
2019-07-21 17:45:47,386 - INFO - allennlp.common.params - CURRENTLY DEFINED PARAMETERS: 
2019-07-21 17:45:47,386 - INFO - allennlp.common.params - model.decoder.rnn.hidden_size = 512
2019-07-21 17:45:47,386 - INFO - allennlp.common.params - model.decoder.rnn.input_size = 428
2019-07-21 17:45:47,386 - INFO - allennlp.common.params - model.decoder.rnn.num_layers = 1
2019-07-21 17:45:47,386 - INFO - allennlp.common.params - model.decoder.rnn.batch_first = True
2019-07-21 17:45:47,398 - INFO - allennlp.common.params - model.decoder.latent_dim = 128
2019-07-21 17:45:47,398 - INFO - allennlp.common.params - model.decoder.dropout_p = 0.5
2019-07-21 17:45:47,460 - INFO - allennlp.common.from_params - instantiating class <class 'src.modules.annealer.LossWeight'> from params {'max_weight': 0.5, 'min_weight': 0, 'num_iter_to_max': 3500, 'slope': 0.5, 'type': 'sigmoid_annealed', 'warmup': 500} and extras {'vocab'}
2019-07-21 17:45:47,460 - INFO - allennlp.common.params - model.kl_weight.type = sigmoid_annealed
2019-07-21 17:45:47,460 - INFO - allennlp.common.from_params - instantiating class <class 'src.modules.annealer.SigmoidAnnealedWeight'> from params {'max_weight': 0.5, 'min_weight': 0, 'num_iter_to_max': 3500, 'slope': 0.5, 'warmup': 500} and extras {'vocab'}
2019-07-21 17:45:47,460 - INFO - allennlp.common.params - model.kl_weight.min_weight = 0
2019-07-21 17:45:47,460 - INFO - allennlp.common.params - model.kl_weight.max_weight = 0.5
2019-07-21 17:45:47,460 - INFO - allennlp.common.params - model.kl_weight.warmup = 500
2019-07-21 17:45:47,460 - INFO - allennlp.common.params - model.kl_weight.num_iter_to_max = 3500
2019-07-21 17:45:47,460 - INFO - allennlp.common.params - model.kl_weight.slope = 0.5
2019-07-21 17:45:47,460 - INFO - allennlp.common.params - model.temperature = 1e-05
2019-07-21 17:45:47,460 - INFO - allennlp.common.params - model.initializer = [['.*', {'a': -0.1, 'b': 0.1, 'type': 'uniform'}]]
2019-07-21 17:45:47,460 - INFO - allennlp.common.from_params - instantiating class <class 'allennlp.nn.initializers.Initializer'> from params {'a': -0.1, 'b': 0.1, 'type': 'uniform'} and extras set()
2019-07-21 17:45:47,460 - INFO - allennlp.common.params - model.initializer.0.1.type = uniform
2019-07-21 17:45:47,460 - INFO - allennlp.common.params - Converting Params object to dict; logging of default values will not occur when dictionary parameters are used subsequently.
2019-07-21 17:45:47,460 - INFO - allennlp.common.params - CURRENTLY DEFINED PARAMETERS: 
2019-07-21 17:45:47,460 - INFO - allennlp.common.params - model.initializer.0.1.a = -0.1
2019-07-21 17:45:47,460 - INFO - allennlp.common.params - model.initializer.0.1.b = 0.1
2019-07-21 17:45:47,461 - INFO - allennlp.nn.initializers - Initializing parameters
2019-07-21 17:45:47,461 - INFO - allennlp.nn.initializers - Initializing _encoder._text_field_embedder.token_embedder_tokens.weight using .* initializer
2019-07-21 17:45:47,490 - INFO - allennlp.nn.initializers - Initializing _encoder._encoder._module.weight_ih_l0 using .* initializer
2019-07-21 17:45:47,493 - INFO - allennlp.nn.initializers - Initializing _encoder._encoder._module.weight_hh_l0 using .* initializer
2019-07-21 17:45:47,497 - INFO - allennlp.nn.initializers - Initializing _encoder._encoder._module.bias_ih_l0 using .* initializer
2019-07-21 17:45:47,498 - INFO - allennlp.nn.initializers - Initializing _encoder._encoder._module.bias_hh_l0 using .* initializer
2019-07-21 17:45:47,498 - INFO - allennlp.nn.initializers - Initializing _encoder._encoder._module.weight_ih_l0_reverse using .* initializer
2019-07-21 17:45:47,500 - INFO - allennlp.nn.initializers - Initializing _encoder._encoder._module.weight_hh_l0_reverse using .* initializer
2019-07-21 17:45:47,505 - INFO - allennlp.nn.initializers - Initializing _encoder._encoder._module.bias_ih_l0_reverse using .* initializer
2019-07-21 17:45:47,505 - INFO - allennlp.nn.initializers - Initializing _encoder._encoder._module.bias_hh_l0_reverse using .* initializer
2019-07-21 17:45:47,505 - INFO - allennlp.nn.initializers - Initializing _encoder._latent_to_mean.weight using .* initializer
2019-07-21 17:45:47,506 - INFO - allennlp.nn.initializers - Initializing _encoder._latent_to_mean.bias using .* initializer
2019-07-21 17:45:47,506 - INFO - allennlp.nn.initializers - Initializing _encoder._latent_to_logvar.weight using .* initializer
2019-07-21 17:45:47,507 - INFO - allennlp.nn.initializers - Initializing _encoder._latent_to_logvar.bias using .* initializer
2019-07-21 17:45:47,507 - INFO - allennlp.nn.initializers - Initializing _decoder._target_embedder.token_embedder_tokens.weight using .* initializer
2019-07-21 17:45:47,536 - INFO - allennlp.nn.initializers - Initializing _decoder.rnn._module.weight_ih_l0 using .* initializer
2019-07-21 17:45:47,540 - INFO - allennlp.nn.initializers - Initializing _decoder.rnn._module.weight_hh_l0 using .* initializer
2019-07-21 17:45:47,545 - INFO - allennlp.nn.initializers - Initializing _decoder.rnn._module.bias_ih_l0 using .* initializer
2019-07-21 17:45:47,545 - INFO - allennlp.nn.initializers - Initializing _decoder.rnn._module.bias_hh_l0 using .* initializer
2019-07-21 17:45:47,545 - INFO - allennlp.nn.initializers - Initializing _decoder._latent_to_dec_hidden.weight using .* initializer
2019-07-21 17:45:47,545 - INFO - allennlp.nn.initializers - Initializing _decoder._latent_to_dec_hidden.bias using .* initializer
2019-07-21 17:45:47,545 - INFO - allennlp.nn.initializers - Initializing _decoder._dec_linear.weight using .* initializer
2019-07-21 17:45:47,595 - INFO - allennlp.nn.initializers - Initializing _decoder._dec_linear.bias using .* initializer
2019-07-21 17:45:47,595 - INFO - allennlp.nn.initializers - Done initializing parameters; the following parameters are using their default initialization from their code
2019-07-21 17:45:47,626 - INFO - allennlp.common.from_params - instantiating class <class 'allennlp.data.iterators.data_iterator.DataIterator'> from params {'batch_size': 64, 'sorting_keys': [['source_tokens', 'num_tokens']], 'type': 'bucket'} and extras set()
2019-07-21 17:45:47,626 - INFO - allennlp.common.params - iterator.type = bucket
2019-07-21 17:45:47,626 - INFO - allennlp.common.from_params - instantiating class <class 'allennlp.data.iterators.bucket_iterator.BucketIterator'> from params {'batch_size': 64, 'sorting_keys': [['source_tokens', 'num_tokens']]} and extras set()
2019-07-21 17:45:47,626 - INFO - allennlp.common.params - iterator.sorting_keys = [['source_tokens', 'num_tokens']]
2019-07-21 17:45:47,626 - INFO - allennlp.common.params - iterator.padding_noise = 0.1
2019-07-21 17:45:47,626 - INFO - allennlp.common.params - iterator.biggest_batch_first = False
2019-07-21 17:45:47,626 - INFO - allennlp.common.params - iterator.batch_size = 64
2019-07-21 17:45:47,626 - INFO - allennlp.common.params - iterator.instances_per_epoch = None
2019-07-21 17:45:47,626 - INFO - allennlp.common.params - iterator.max_instances_in_memory = None
2019-07-21 17:45:47,626 - INFO - allennlp.common.params - iterator.cache_instances = False
2019-07-21 17:45:47,626 - INFO - allennlp.common.params - iterator.track_epoch = False
2019-07-21 17:45:47,626 - INFO - allennlp.common.params - iterator.maximum_samples_per_batch = None
2019-07-21 17:45:47,626 - INFO - allennlp.common.params - validation_iterator = None
2019-07-21 17:45:47,626 - INFO - allennlp.common.params - trainer.no_grad = ()
2019-07-21 17:45:47,626 - INFO - allennlp.training.trainer_pieces - Following parameters are Frozen  (without gradient):
2019-07-21 17:45:47,626 - INFO - allennlp.training.trainer_pieces - Following parameters are Tunable (with gradient):
2019-07-21 17:45:47,626 - INFO - allennlp.training.trainer_pieces - _encoder._text_field_embedder.token_embedder_tokens.weight
2019-07-21 17:45:47,626 - INFO - allennlp.training.trainer_pieces - _encoder._encoder._module.weight_ih_l0
2019-07-21 17:45:47,626 - INFO - allennlp.training.trainer_pieces - _encoder._encoder._module.weight_hh_l0
2019-07-21 17:45:47,626 - INFO - allennlp.training.trainer_pieces - _encoder._encoder._module.bias_ih_l0
2019-07-21 17:45:47,626 - INFO - allennlp.training.trainer_pieces - _encoder._encoder._module.bias_hh_l0
2019-07-21 17:45:47,626 - INFO - allennlp.training.trainer_pieces - _encoder._encoder._module.weight_ih_l0_reverse
2019-07-21 17:45:47,626 - INFO - allennlp.training.trainer_pieces - _encoder._encoder._module.weight_hh_l0_reverse
2019-07-21 17:45:47,627 - INFO - allennlp.training.trainer_pieces - _encoder._encoder._module.bias_ih_l0_reverse
2019-07-21 17:45:47,627 - INFO - allennlp.training.trainer_pieces - _encoder._encoder._module.bias_hh_l0_reverse
2019-07-21 17:45:47,627 - INFO - allennlp.training.trainer_pieces - _encoder._latent_to_mean.weight
2019-07-21 17:45:47,627 - INFO - allennlp.training.trainer_pieces - _encoder._latent_to_mean.bias
2019-07-21 17:45:47,627 - INFO - allennlp.training.trainer_pieces - _encoder._latent_to_logvar.weight
2019-07-21 17:45:47,627 - INFO - allennlp.training.trainer_pieces - _encoder._latent_to_logvar.bias
2019-07-21 17:45:47,627 - INFO - allennlp.training.trainer_pieces - _decoder._target_embedder.token_embedder_tokens.weight
2019-07-21 17:45:47,627 - INFO - allennlp.training.trainer_pieces - _decoder.rnn._module.weight_ih_l0
2019-07-21 17:45:47,627 - INFO - allennlp.training.trainer_pieces - _decoder.rnn._module.weight_hh_l0
2019-07-21 17:45:47,627 - INFO - allennlp.training.trainer_pieces - _decoder.rnn._module.bias_ih_l0
2019-07-21 17:45:47,627 - INFO - allennlp.training.trainer_pieces - _decoder.rnn._module.bias_hh_l0
2019-07-21 17:45:47,627 - INFO - allennlp.training.trainer_pieces - _decoder._latent_to_dec_hidden.weight
2019-07-21 17:45:47,627 - INFO - allennlp.training.trainer_pieces - _decoder._latent_to_dec_hidden.bias
2019-07-21 17:45:47,627 - INFO - allennlp.training.trainer_pieces - _decoder._dec_linear.weight
2019-07-21 17:45:47,627 - INFO - allennlp.training.trainer_pieces - _decoder._dec_linear.bias
2019-07-21 17:45:47,627 - INFO - allennlp.common.params - trainer.shuffle = True
2019-07-21 17:45:47,627 - INFO - allennlp.common.params - trainer.num_epochs = 30
2019-07-21 17:45:47,627 - INFO - allennlp.common.params - trainer.cuda_device = 1
Traceback (most recent call last):
  File "/home/gsahu/dev/bin/allennlp", line 11, in <module>
    load_entry_point('allennlp', 'console_scripts', 'allennlp')()
  File "/home/gsahu/allennlp/allennlp/run.py", line 18, in run
    main(prog="allennlp")
  File "/home/gsahu/allennlp/allennlp/commands/__init__.py", line 102, in main
    args.func(args)
  File "/home/gsahu/allennlp/allennlp/commands/train.py", line 117, in train_model_from_args
    args.cache_prefix)
  File "/home/gsahu/allennlp/allennlp/commands/train.py", line 161, in train_model_from_file
    cache_directory, cache_prefix)
  File "/home/gsahu/allennlp/allennlp/commands/train.py", line 239, in train_model
    trainer = TrainerBase.from_params(params, serialization_dir, recover)
  File "/home/gsahu/allennlp/allennlp/training/trainer_base.py", line 85, in from_params
    return klass.from_params(params, serialization_dir, recover)
  File "/home/gsahu/allennlp/allennlp/training/callback_trainer.py", line 285, in from_params
    model = model.cuda(model_device)
  File "/home/gsahu/dev/lib/python3.6/site-packages/torch/nn/modules/module.py", line 265, in cuda
    return self._apply(lambda t: t.cuda(device))
  File "/home/gsahu/dev/lib/python3.6/site-packages/torch/nn/modules/module.py", line 193, in _apply
    module._apply(fn)
  File "/home/gsahu/dev/lib/python3.6/site-packages/torch/nn/modules/module.py", line 193, in _apply
    module._apply(fn)
  File "/home/gsahu/dev/lib/python3.6/site-packages/torch/nn/modules/module.py", line 193, in _apply
    module._apply(fn)
  File "/home/gsahu/dev/lib/python3.6/site-packages/torch/nn/modules/rnn.py", line 127, in _apply
    self.flatten_parameters()
  File "/home/gsahu/dev/lib/python3.6/site-packages/torch/nn/modules/rnn.py", line 123, in flatten_parameters
    self.batch_first, bool(self.bidirectional))
RuntimeError: cuDNN error: CUDNN_STATUS_EXECUTION_FAILED

problem of data preprocess

How do you preprocess the data, for example for DailyDialogue?
I see that you are directly using the file "data/interim/dialog/train_sentences.tsv". How did you get it? Thank you

Generating Responses

What is the protocol for generating individual or batch responses using the DialogGan? If you could elaborate more on using the VAE and GAN models in practice together that would be much appreciated. Thanks.