ValueError: Missing field 'source_vocabulary' in the data configuration

Support
1

I want to translate using a trained model provided on WMT English-German dataset on opennmt-tf script.
On there prepare_data.sh script created a single vocabulary file and wmt-ende-sp folder there aren’t any vocabulary file for source and target language.
Therefore after putting this command:

onmt-main --config data.yml --model_type Transformer --auto_config infer --features_file newstest2017-ende-src.en --predictions_file predict_1.txt

it’s showing this error

2

This script generates a shared vocabulary.

See the example configuration to see how it is configured:

1 Like