CNTK/CMUDict в e8078a2c7688cd9d2222a2bc9455864964682fbc - CNTK

История

Amit Agarwal 8a9020f78e CNTK v2 library: Migrate past_value and future_value to sequence		2017-03-30 12:11:20 -07:00
..
BrainScript	Fixed python commons and eval clients	2016-11-14 16:24:50 +01:00
Data	Restructuring examples and tutorials	2016-11-14 16:24:45 +01:00
Python	CNTK v2 library: Migrate past_value and future_value to sequence	2017-03-30 12:11:20 -07:00
README.md	fixed tutorials and VS links	2016-11-14 16:24:56 +01:00

README.md

This example demonstrates the use of CNTK for grapheme-to-phoneme (letter-to-sound) conversion using a sequence-to-sequence model with attention, using the CMUDict dictionary.

The code supports a number of alternative configurations. As configured currently, it implements

a 3-hidden layer unidirectional LSTM encoder network, all hidden dimensions are 512
a 3-hidden layer unidirectional LSTM decoder network, all hidden dimensions are 512
encoder state is passed to the decoder by means of attention, with projection dimension 128 and maximum input length of 20 tokens
embedding is disabled (because the 'vocabulary' of the task, letters and phonemes, is very small)
beam decoder with beam width 3

To Use

Modify the following in G2P.cntk as needed:

pathnames
deviceId to specify CPU (-1) or GPU (>=0 or "auto")

Run:

command line: cntk configFile=Examples/SequenceToSequence/CMUDict/BrainScript/G2P.cntk RunRootDir=g2p
VS Debugger: configFile=$(SolutionDir)Examples/SequenceToSequence/CMUDict/BrainScript/G2P.cntk RunRootDir=$(SolutionDir)Examples/SequenceToSequence/CMUDict