Input type for training

mazida · August 12, 2022, 9:31am

Can we train a model with tagged input be it either POS tags or NE tags ?

ymoslem · August 12, 2022, 10:58am

OpenNMT-tf supports multiple features:

OpenNMT/OpenNMT-tf/blob/master/config/models/multi_features_transformer.py

"""Defines a Transformer model with multiple input features. For example, these
could be words, parts of speech, and lemmas that are embedded in parallel and
concatenated into a single input embedding.

The features are separate data files with separate vocabularies. The YAML
configuration file should look like this:

data:
  train_features_file:
    - features_1.txt
    - features_2.txt
    - features_3.txt
  train_labels_file: target.txt
  source_1_vocabulary: feature_1_vocab.txt
  source_2_vocabulary: feature_2_vocab.txt
  source_3_vocabulary: feature_3_vocab.txt
  target_vocabulary: target_vocab.txt
"""

import opennmt

This file has been truncated. show original

mazida · August 12, 2022, 11:55am

Okay…thank you

francoishernandez · August 30, 2022, 10:47am

Source features support was added by @anderleich some time ago:
https://opennmt.net/OpenNMT-py/FAQ.html#how-can-i-use-source-word-features