org.datavec.api.records.metadata.RecordMetaDataIndex Java Examples
The following examples show how to use
org.datavec.api.records.metadata.RecordMetaDataIndex.
You can vote up the ones you like or vote down the ones you don't like,
and go to the original project or source file by following the links above each example. You may check out the related API usage on the sidebar.
Example #1
Source File: CollectionSequenceRecordReader.java From DataVec with Apache License 2.0 | 5 votes |
@Override public List<SequenceRecord> loadSequenceFromMetaData(List<RecordMetaData> recordMetaDatas) throws IOException { Set<Integer> toLoad = new LinkedHashSet<>(); for (RecordMetaData recordMetaData : recordMetaDatas) { if (!(recordMetaData instanceof RecordMetaDataIndex)) { throw new IllegalArgumentException("Expected RecordMetaDataIndex; got: " + recordMetaData); } long idx = ((RecordMetaDataIndex) recordMetaData).getIndex(); if (idx >= original.size()) { throw new IllegalStateException( "Cannot get index " + idx + " from collection: contains " + original + " elements"); } toLoad.add((int) idx); } List<SequenceRecord> out = new ArrayList<>(); Iterator<? extends Collection<? extends Collection<Writable>>> iter = original.iterator(); int i = 0; while (iter.hasNext()) { Collection<? extends Collection<Writable>> c = iter.next(); if (!toLoad.contains(i++)) { continue; } List<List<Writable>> record = toList(c); SequenceRecord r = new org.datavec.api.records.impl.SequenceRecord(record, new RecordMetaDataIndex(i - 1, null, CollectionSequenceRecordReader.class)); out.add(r); } return out; }
Example #2
Source File: CollectionSequenceRecordReader.java From deeplearning4j with Apache License 2.0 | 5 votes |
@Override public List<SequenceRecord> loadSequenceFromMetaData(List<RecordMetaData> recordMetaDatas) throws IOException { Set<Integer> toLoad = new LinkedHashSet<>(); for (RecordMetaData recordMetaData : recordMetaDatas) { if (!(recordMetaData instanceof RecordMetaDataIndex)) { throw new IllegalArgumentException("Expected RecordMetaDataIndex; got: " + recordMetaData); } long idx = ((RecordMetaDataIndex) recordMetaData).getIndex(); if (idx >= original.size()) { throw new IllegalStateException( "Cannot get index " + idx + " from collection: contains " + original + " elements"); } toLoad.add((int) idx); } List<SequenceRecord> out = new ArrayList<>(); Iterator<? extends Collection<? extends Collection<Writable>>> iter = original.iterator(); int i = 0; while (iter.hasNext()) { Collection<? extends Collection<Writable>> c = iter.next(); if (!toLoad.contains(i++)) { continue; } List<List<Writable>> record = toList(c); SequenceRecord r = new org.datavec.api.records.impl.SequenceRecord(record, new RecordMetaDataIndex(i - 1, null, CollectionSequenceRecordReader.class)); out.add(r); } return out; }
Example #3
Source File: ArrowConverterTest.java From deeplearning4j with Apache License 2.0 | 5 votes |
@Test public void testRecordReaderMetaData() throws Exception { val recordsToWrite = recordToWrite(); //send file File tmp = tmpDataFile(recordsToWrite); RecordReader recordReader = new ArrowRecordReader(); RecordMetaDataIndex recordMetaDataIndex = new RecordMetaDataIndex(0,tmp.toURI(),ArrowRecordReader.class); recordReader.loadFromMetaData(recordMetaDataIndex); Record record = recordReader.nextRecord(); assertEquals(2,record.getRecord().size()); }
Example #4
Source File: ArrowConverterTest.java From deeplearning4j with Apache License 2.0 | 5 votes |
@Test public void testRecordReaderMetaDataList() throws Exception { val recordsToWrite = recordToWrite(); //send file File tmp = tmpDataFile(recordsToWrite); RecordReader recordReader = new ArrowRecordReader(); RecordMetaDataIndex recordMetaDataIndex = new RecordMetaDataIndex(0,tmp.toURI(),ArrowRecordReader.class); recordReader.loadFromMetaData(Arrays.<RecordMetaData>asList(recordMetaDataIndex)); Record record = recordReader.nextRecord(); assertEquals(2,record.getRecord().size()); }
Example #5
Source File: ArrowRecordReader.java From deeplearning4j with Apache License 2.0 | 5 votes |
@Override public List<Record> loadFromMetaData(List<RecordMetaData> recordMetaDatas) { Map<String,List<RecordMetaData>> metaDataByUri = new HashMap<>(); //gather all unique locations for the metadata //this will prevent initialization multiple times of the record for(RecordMetaData recordMetaData : recordMetaDatas) { if(!(recordMetaData instanceof RecordMetaDataIndex)) { throw new IllegalArgumentException("Unable to load from meta data. No index specified for record"); } List<RecordMetaData> recordMetaData1 = metaDataByUri.get(recordMetaData.getURI().toString()); if(recordMetaData1 == null) { recordMetaData1 = new ArrayList<>(); metaDataByUri.put(recordMetaData.getURI().toString(),recordMetaData1); } recordMetaData1.add(recordMetaData); } List<Record> ret = new ArrayList<>(); for(String uri : metaDataByUri.keySet()) { List<RecordMetaData> metaData = metaDataByUri.get(uri); InputSplit fileSplit = new FileSplit(new File(URI.create(uri))); initialize(fileSplit); for(RecordMetaData index : metaData) { RecordMetaDataIndex index2 = (RecordMetaDataIndex) index; this.currIdx = (int) index2.getIndex(); ret.add(nextRecord()); } } return ret; }
Example #6
Source File: ArrowRecordReader.java From deeplearning4j with Apache License 2.0 | 5 votes |
@Override public Record loadFromMetaData(RecordMetaData recordMetaData) { if(!(recordMetaData instanceof RecordMetaDataIndex)) { throw new IllegalArgumentException("Unable to load from meta data. No index specified for record"); } RecordMetaDataIndex index = (RecordMetaDataIndex) recordMetaData; InputSplit fileSplit = new FileSplit(new File(index.getURI())); initialize(fileSplit); this.currIdx = (int) index.getIndex(); return nextRecord(); }
Example #7
Source File: ArrowConverterTest.java From DataVec with Apache License 2.0 | 5 votes |
@Test public void testRecordReaderMetaData() throws Exception { val recordsToWrite = recordToWrite(); //send file File tmp = tmpDataFile(recordsToWrite); RecordReader recordReader = new ArrowRecordReader(); RecordMetaDataIndex recordMetaDataIndex = new RecordMetaDataIndex(0,tmp.toURI(),ArrowRecordReader.class); recordReader.loadFromMetaData(recordMetaDataIndex); Record record = recordReader.nextRecord(); assertEquals(2,record.getRecord().size()); }
Example #8
Source File: ArrowConverterTest.java From DataVec with Apache License 2.0 | 5 votes |
@Test public void testRecordReaderMetaDataList() throws Exception { val recordsToWrite = recordToWrite(); //send file File tmp = tmpDataFile(recordsToWrite); RecordReader recordReader = new ArrowRecordReader(); RecordMetaDataIndex recordMetaDataIndex = new RecordMetaDataIndex(0,tmp.toURI(),ArrowRecordReader.class); recordReader.loadFromMetaData(Arrays.<RecordMetaData>asList(recordMetaDataIndex)); Record record = recordReader.nextRecord(); assertEquals(2,record.getRecord().size()); }
Example #9
Source File: ArrowRecordReader.java From DataVec with Apache License 2.0 | 5 votes |
@Override public List<Record> loadFromMetaData(List<RecordMetaData> recordMetaDatas) { Map<String,List<RecordMetaData>> metaDataByUri = new HashMap<>(); //gather all unique locations for the metadata //this will prevent initialization multiple times of the record for(RecordMetaData recordMetaData : recordMetaDatas) { if(!(recordMetaData instanceof RecordMetaDataIndex)) { throw new IllegalArgumentException("Unable to load from meta data. No index specified for record"); } List<RecordMetaData> recordMetaData1 = metaDataByUri.get(recordMetaData.getURI().toString()); if(recordMetaData1 == null) { recordMetaData1 = new ArrayList<>(); metaDataByUri.put(recordMetaData.getURI().toString(),recordMetaData1); } recordMetaData1.add(recordMetaData); } List<Record> ret = new ArrayList<>(); for(String uri : metaDataByUri.keySet()) { List<RecordMetaData> metaData = metaDataByUri.get(uri); InputSplit fileSplit = new FileSplit(new File(URI.create(uri))); initialize(fileSplit); for(RecordMetaData index : metaData) { RecordMetaDataIndex index2 = (RecordMetaDataIndex) index; this.currIdx = (int) index2.getIndex(); ret.add(nextRecord()); } } return ret; }
Example #10
Source File: ArrowRecordReader.java From DataVec with Apache License 2.0 | 5 votes |
@Override public Record loadFromMetaData(RecordMetaData recordMetaData) { if(!(recordMetaData instanceof RecordMetaDataIndex)) { throw new IllegalArgumentException("Unable to load from meta data. No index specified for record"); } RecordMetaDataIndex index = (RecordMetaDataIndex) recordMetaData; InputSplit fileSplit = new FileSplit(new File(index.getURI())); initialize(fileSplit); this.currIdx = (int) index.getIndex(); return nextRecord(); }
Example #11
Source File: VasttextExtraMemoryReader.java From scava with Eclipse Public License 2.0 | 5 votes |
private Record processNextRecord() { List<Writable> next = super.next(); //As we increase in next the listIndex, then we need to reduce in one the value of listIndex RecordMetaData meta = new RecordMetaDataIndex(listIndex-1, null, VasttextExtraMemoryReader.class); return new org.datavec.api.records.impl.Record(next, meta); }
Example #12
Source File: VasttextMemoryRecordReader.java From scava with Eclipse Public License 2.0 | 5 votes |
@Override public Record nextRecord() { List<Writable> next = next(); //As we increase in next the listIndex, then we need to reduce in one the value of listIndex RecordMetaData meta = new RecordMetaDataIndex(listIndex-1, null, VasttextMemoryRecordReader.class); return new org.datavec.api.records.impl.Record(next, meta); }
Example #13
Source File: CollectionRecordReader.java From DataVec with Apache License 2.0 | 4 votes |
@Override public Record nextRecord() { return new org.datavec.api.records.impl.Record(next(), new RecordMetaDataIndex(count - 1, null, CollectionRecordReader.class)); }
Example #14
Source File: CollectionSequenceRecordReader.java From DataVec with Apache License 2.0 | 4 votes |
@Override public SequenceRecord nextSequence() { return new org.datavec.api.records.impl.SequenceRecord(sequenceRecord(), new RecordMetaDataIndex(count - 1, null, CollectionSequenceRecordReader.class)); }
Example #15
Source File: VasttextTextMemoryReader.java From scava with Eclipse Public License 2.0 | 4 votes |
public Record processNextRecord() { List<Writable> nextProcessed = new ArrayList<>(); List<Writable> next = next(); int label; String text = next.get(0).toString(); //Text part nextProcessed.add(new Text(text)); //Label part if(labelled) { if (multilabel) { String[] labels = next.get(1).toString().split(" "); // We need to create a vector of labels List<Double> transformedLabels = new ArrayList<Double>(); boolean seenLabels = false; for (int i = 0; i < labels.length; i++) { label = vasttextTextVectorizer.getLabelIndex(labels[i]); if (label == -1 && vasttextTextVectorizer.hasfitFinished()) { System.err.println("WARNING: The label \"" + next.get(1).toString() + "\" wasn't seen during the training. It has been removed from the entry " + (listIndex - 1)); continue; } else if (label == -1) throw new UnsupportedOperationException( "Unknown error from the vectorizer. Returned a label of -1 during fitting the dictionary."); transformedLabels.add((double) label); seenLabels = true; } // This means that none of the labels was seen in the training if (!seenLabels) return null; INDArray transformed = Nd4j.create(transformedLabels.stream().mapToDouble(d -> d).toArray(), new int[] { transformedLabels.size(), 1 }); nextProcessed.add(new NDArrayWritable(transformed)); } else { label = vasttextTextVectorizer.getLabelIndex(next.get(1).toString()); if (label == -1 && vasttextTextVectorizer.hasfitFinished()) { System.err.println( "WARNING: The label \"" + next.get(1).toString() + "\" wasn't seen during the training."); return null; } else if (label == -1) throw new UnsupportedOperationException( "Unknown error from the vectorizer. Returned a label of -1 during fitting the dictionary."); nextProcessed.add(new IntWritable(label)); } } //As we increase in next the listIndex, then we need to reduce in one the value of listIndex RecordMetaData meta = new RecordMetaDataIndex(listIndex-1, null, VasttextMemoryRecordReader.class); return new org.datavec.api.records.impl.Record(nextProcessed, meta); }
Example #16
Source File: MapFileSequenceRecordReader.java From DataVec with Apache License 2.0 | 4 votes |
@Override public SequenceRecord loadSequenceFromMetaData(@NonNull RecordMetaData recordMetaData) throws IOException { long idx = ((RecordMetaDataIndex) recordMetaData).getIndex(); return new org.datavec.api.records.impl.SequenceRecord(mapFileReader.getRecord(idx).getSequenceRecord(), recordMetaData); }
Example #17
Source File: MapFileSequenceRecordReader.java From deeplearning4j with Apache License 2.0 | 4 votes |
@Override public SequenceRecord loadSequenceFromMetaData(@NonNull RecordMetaData recordMetaData) throws IOException { long idx = ((RecordMetaDataIndex) recordMetaData).getIndex(); return new org.datavec.api.records.impl.SequenceRecord(mapFileReader.getRecord(idx).getSequenceRecord(), recordMetaData); }
Example #18
Source File: ArrowRecord.java From deeplearning4j with Apache License 2.0 | 4 votes |
@Override public RecordMetaData getMetaData() { RecordMetaData ret = new RecordMetaDataIndex(index,recordUri,ArrowRecordReader.class); return ret; }
Example #19
Source File: ArrowRecord.java From DataVec with Apache License 2.0 | 4 votes |
@Override public RecordMetaData getMetaData() { RecordMetaData ret = new RecordMetaDataIndex(index,recordUri,ArrowRecordReader.class); return ret; }
Example #20
Source File: CollectionRecordReader.java From deeplearning4j with Apache License 2.0 | 4 votes |
@Override public Record nextRecord() { return new org.datavec.api.records.impl.Record(next(), new RecordMetaDataIndex(count - 1, null, CollectionRecordReader.class)); }
Example #21
Source File: CollectionSequenceRecordReader.java From deeplearning4j with Apache License 2.0 | 4 votes |
@Override public SequenceRecord nextSequence() { return new org.datavec.api.records.impl.SequenceRecord(sequenceRecord(), new RecordMetaDataIndex(count - 1, null, CollectionSequenceRecordReader.class)); }