fixed TRADE training

Signed-off-by: Evelina Bakhturina <ebakhturina@nvidia.com>
NVIDIA · Jun 4, 2020 · f9f1d45 · f9f1d45
1 parent db3d410
commit f9f1d45
Show file tree

Hide file tree

Showing 4 changed files with 11 additions and 12 deletions.
diff --git a/examples/nlp/dialogue_state_tracking/dialogue_state_tracking_trade.py b/examples/nlp/dialogue_state_tracking/dialogue_state_tracking_trade.py
@@ -30,13 +30,12 @@
 
 import nemo.core as nemo_core
 from nemo import logging
-from nemo.backends.pytorch.common import EncoderRNN
 from nemo.backends.pytorch.common.losses import CrossEntropyLossNM, LossAggregatorNM
 from nemo.collections.nlp.callbacks.state_tracking_trade_callback import eval_epochs_done_callback, eval_iter_callback
 from nemo.collections.nlp.data.datasets.multiwoz_dataset import MultiWOZDataDesc
 from nemo.collections.nlp.nm.data_layers import MultiWOZDataLayer
 from nemo.collections.nlp.nm.losses import MaskedLogLoss
-from nemo.collections.nlp.nm.trainables import TRADEGenerator
+from nemo.collections.nlp.nm.trainables import EncoderRNN, TRADEGenerator
 from nemo.utils.lr_policies import get_lr_policy
 
 parser = argparse.ArgumentParser(description='Dialogue state tracking with TRADE model on MultiWOZ dataset')
@@ -155,8 +154,8 @@ def create_pipeline(num_samples, batch_size, num_gpus, input_dropout, data_prefi
     point_outputs, gate_outputs = decoder(
         encoder_hidden=hidden,
         encoder_outputs=outputs,
-        input_lens=input_data.src_lens,
-        src_ids=input_data.src_ids,
+        dialog_lens=input_data.src_lens,
+        dialog_ids=input_data.src_ids,
         targets=input_data.tgt_ids,
     )
 

diff --git a/nemo/collections/nlp/nm/data_layers/state_tracking_trade_datalayer.py b/nemo/collections/nlp/nm/data_layers/state_tracking_trade_datalayer.py
@@ -43,7 +43,7 @@
 import nemo
 from nemo.collections.nlp.data.datasets.multiwoz_dataset import MultiWOZDataset
 from nemo.collections.nlp.nm.data_layers.text_datalayer import TextDataLayer
-from nemo.core.neural_types import ChannelType, LabelsType, LengthsType, NeuralType
+from nemo.core.neural_types import LabelsType, Length, NeuralType, TokenIndex
 from nemo.utils.decorators import add_port_docs
 
 __all__ = ['MultiWOZDataLayer']
@@ -83,10 +83,10 @@ def output_ports(self):
 
         """
         return {
-            "src_ids": NeuralType(('B', 'T'), ChannelType()),
-            "src_lens": NeuralType(tuple('B'), LengthsType()),
+            "src_ids": NeuralType(('B', 'T'), TokenIndex()),
+            "src_lens": NeuralType(tuple('B'), Length()),
             "tgt_ids": NeuralType(('B', 'D', 'T'), LabelsType()),
-            "tgt_lens": NeuralType(('B', 'D'), LengthsType()),
+            "tgt_lens": NeuralType(('B', 'D'), Length()),
             "gating_labels": NeuralType(('B', 'D'), LabelsType()),
             "turn_domain": NeuralType(),
         }

diff --git a/nemo/collections/nlp/nm/losses/masked_xentropy_loss.py b/nemo/collections/nlp/nm/losses/masked_xentropy_loss.py
@@ -39,7 +39,7 @@
 import torch
 
 from nemo.backends.pytorch.nm import LossNM
-from nemo.core.neural_types import LabelsType, LengthsType, LogitsType, LossType, NeuralType
+from nemo.core.neural_types import LabelsType, Length, LogitsType, LossType, NeuralType
 from nemo.utils.decorators import add_port_docs
 
 __all__ = ['MaskedLogLoss']
@@ -72,7 +72,7 @@ def input_ports(self):
         return {
             "logits": NeuralType(('B', 'T', 'D', 'D'), LogitsType()),
             "labels": NeuralType(('B', 'D', 'T'), LabelsType()),
-            "length_mask": NeuralType(('B', 'D'), LengthsType()),
+            "length_mask": NeuralType(('B', 'D'), Length()),
         }
 
     @property

diff --git a/nemo/collections/nlp/nm/trainables/dialogue_state_tracking/trade_generator_nm.py b/nemo/collections/nlp/nm/trainables/dialogue_state_tracking/trade_generator_nm.py
@@ -70,8 +70,8 @@ def input_ports(self):
         """Returns definitions of module input ports.
         """
         return {
-            'encoder_hidden': NeuralType(('B', 'T', 'C'), ChannelType()),
-            'encoder_outputs': NeuralType(('B', 'T', 'C'), ChannelType()),
+            'encoder_hidden': NeuralType(('B', 'T', 'D'), ChannelType()),
+            'encoder_outputs': NeuralType(('B', 'T', 'D'), ChannelType()),
             'dialog_ids': NeuralType(('B', 'T'), elements_type=TokenIndex()),
             'dialog_lens': NeuralType(tuple('B'), elements_type=Length()),
             'targets': NeuralType(('B', 'D', 'T'), LabelsType(), optional=True),