- All Implemented Interfaces:
- Closeable, AutoCloseable, JobConfigurable, Reducer<IntWritable,IntWritable,IntWritable,FloatAsBytesWritable>
- Enclosing class:
- WriteRandomVectors
public static class WriteRandomVectors.MyReducer0
extends MapReduceBase
implements Reducer<IntWritable,IntWritable,IntWritable,FloatAsBytesWritable>
- Author:
- ferhanture
(K=size of each random vector) needs to be set manually, as the number of terms (i.e.,
vocabulary size) Ideally, this should be passed through the Configuration object, but
that is not checked right now.
The size of a file is the number of random vectors to be written at a single file.
Multiple files are used if necessary. The parameter FILESIZE should be set through the
ComputeSignatures class' SIZE_OF_FILE field. FILESIZE should be the maximum number of
random vectors such that the size of each file is small enough to fit into memory