Data sampling is a technique to select a subset of the training set at each epoch. This could be a way to make the epoch unit smaller or select relevant training sequences at each epoch. There are different types of sampling that are selected using
-sample_type option as defined below.
When sampling, with the option
-sample_tgt_vocab it is also possible to restrict the target vocabulary to the current sample which gives an approximate of the full generator as defined here Jean et al, 2015 through a so-called "Importance Sampling" approach.
Importance Sampling is particularly useful when training systems with very large target vocabulary
The simplest data sampling is to uniformly select a subset of the training data. Using the
-sample N option, the training will randomly choose training sequences at each epoch.
A typical use case is to reduce the length of the epochs for more frequent learning rate updates and validation perplexity computation.
This approach is an attempt to feed relevant training data at each epoch. When using the flag
-sample_type perplexity, the perplexity of each sequence is used to generate a multinomial probability distribution over the training sequences. The higher the perplexity, the more likely the sequence is selected.
Alternatively, perplexity-based sampling can be enabled when an average training perplexity is met with the
This perplexity-based approach is experimental and effects are to be experimented. This also results in a ~10% slowdown as the perplexity of each sequence has to be independently computed.
When using the flag
-sample_type partition, samples are drawn without random, uniformally and incrementally from the corpus training. Use this mode for making sure all training sequences will be sent the same number of time.