Training killed when using DIET with large training set

Rasa Open Source
1

Hello, I got a problem when training DIET

Training set size: 7 intents, 15k examples

Config: language: en

pipeline:

  • name: WhitespaceTokenizer
  • name: RegexFeaturizer
  • name: LexicalSyntacticFeaturizer
  • name: CountVectorsFeaturizer analyzer: char_wb min_ngram: 1 max_ngram: 4
  • name: DIETClassifier epochs: 100
  • name: EntitySynonymMapper
  • name: FallbackClassifier threshold: 0.9 ambiguity_threshold: 0.1

zzzz
zzzz1649×463 144 KB

2

Hi @mfayoumi

Have you tried training with a reduced number of examples? Would be interesting to see if that is the cause.

3

Thank you @jjuzl, I’ve solved it by adding the batch_size: [16, 64] to my pipeline