Hi
We are including new pipeline in rasa 1.10.3 with DIET classifier by depreciating the old pipeline that has EmbeddingIntentClassifier. When comparing the two pipelines, the accuracy is good for new pipeline compare to the old pipeline. But the training time for 2200 phrases it is taking 9 mins for new pipeline where as 4 mins for old pipeline. Will you suggest the approach or change in component parameters to fasten the model training process with good accuracy to the new pipeline.
old pipeline:
language: “en”
pipeline:
- name: “WhitespaceTokenizer” intent_split_symbol: “_” intent_tokenization_flag: true
- name: “RegexFeaturizer”
- name: “CRFEntityExtractor”
- name: “EntitySynonymMapper”
- name: “CountVectorsFeaturizer”
- name: “EmbeddingIntentClassifier”
- name: DucklingHTTPExtractor dimensions: [“time”, “number”, “email”] locale: “en_US” timezone: “UTC”
new pipeline:
language: “en”
pipeline:
- name: “WhitespaceTokenizer” intent_split_symbol: “_” intent_tokenization_flag: True case_sensitive: False
- name: “RegexFeaturizer”
- name: “LexicalSyntacticFeaturizer”
- name: CountVectorsFeaturizer
- name: CountVectorsFeaturizer analyzer: “char_wb” min_ngram: 1 max_ngram: 4
- name: DIETClassifier random_seed: 11235 epochs: 100
- name: EntitySynonymMapper
- name: DucklingHTTPExtractor dimensions: [“time”, “number”, “email”] locale: “en_US” timezone: “UTC”