0 ×

02_​Deployment

Workflow

Generate Text Using a Many-To-One LSTM Network (Deployment)
TAG:deep learning keras text generation RNN LSTM text analysis sequence analysis neural network text processing
Deplyment Workflow Deplyment Workflow II Generate Text Using a Many-To-One LSTM Network (Deployment)The workflows generates text in fairy tale style. It reads the previously trained TensorFlow network and predicts a sequences of index-encoded characterswithin a loop. In the “Extract Index” metanode we use the probability distribution over all possible indexes to make the predictions.Here we have two options. We can either always predict the index with the highest probability (Deployment Workflow I) or we can pick the next index basedon the given probability distribution (Deployment Workflow I) . The last node, named Extract Predicted Text translates the sequence of indexes into characters. In order to run the example, please make sure you have the following KNIME extensions installed:* KNIME Deep Learning - Keras Integration (Labs)* KNIME Deep Learning - TensorFlow Integration* KNIME Python Integration Reas modeOutput probabilityfor each characterplus input collectionAppend predicted charseperate inputfrom network outputSplit input collectionOutput probabilityfor each characterplus input collectionSplit input collectionReas modeAppend predicted charseperate inputfrom network output TensorFlowNetwork Reader DL Network Executor Extract Index Create Collection RecursiveLoop Start Column Appender ExtractPredicted Text Column Splitter Delete First Charof Input Sequence Split CollectionColumn Recursive Loop End Extract Index DL Network Executor Delete First Charof Input Sequence Create Collection Split CollectionColumn TensorFlowNetwork Reader RecursiveLoop Start Column Appender Recursive Loop End Read andPre-Process ExtractPredicted Text Column Splitter Read andPre-Process RowID Deplyment Workflow Deplyment Workflow II Generate Text Using a Many-To-One LSTM Network (Deployment)The workflows generates text in fairy tale style. It reads the previously trained TensorFlow network and predicts a sequences of index-encoded characterswithin a loop. In the “Extract Index” metanode we use the probability distribution over all possible indexes to make the predictions.Here we have two options. We can either always predict the index with the highest probability (Deployment Workflow I) or we can pick the next index basedon the given probability distribution (Deployment Workflow I) . The last node, named Extract Predicted Text translates the sequence of indexes into characters. In order to run the example, please make sure you have the following KNIME extensions installed:* KNIME Deep Learning - Keras Integration (Labs)* KNIME Deep Learning - TensorFlow Integration* KNIME Python Integration Reas modeOutput probabilityfor each characterplus input collectionAppend predicted charseperate inputfrom network outputSplit input collectionOutput probabilityfor each characterplus input collectionSplit input collectionReas modeAppend predicted charseperate inputfrom network output TensorFlowNetwork Reader DL Network Executor Extract Index Create Collection RecursiveLoop Start Column Appender ExtractPredicted Text Column Splitter Delete First Charof Input Sequence Split CollectionColumn Recursive Loop End Extract Index DL Network Executor Delete First Charof Input Sequence Create Collection Split CollectionColumn TensorFlowNetwork Reader RecursiveLoop Start Column Appender Recursive Loop End Read andPre-Process ExtractPredicted Text Column Splitter Read andPre-Process RowID

Download

Get this workflow from the following link: Download

Resources

Nodes

02_​Deployment consists of the following 92 nodes(s):

Plugins

02_​Deployment contains nodes provided by the following 5 plugin(s):