veld_step_3_train_fasttext.yaml

x-veld:
  chain:
    description: "Trains a fasttext model on the bible and exports its vectors as a dict serialized 
      into a pkl file. The training data is rather small and the hyperparameteres are simplistic, 
      in order to demonstrate the reproducibility of this chain rather than claiming any deeper 
      insight into the data's words context."
    topic:
      - "ETL"
      - "NLP"
      - "Machine Learning"
      - "Word Embeddings"
      - "Bible Studies"

services:

  veld_step_3_train_fasttext:
    extends:
      file: ./code/veld_code__fasttext/veld_train.yaml
      service: veld_train
    volumes:
      - ./data/training_data/:/veld/input/
      - ./data/models/fasttext/:/veld/output/
    environment:
      in_train_data_file: "bible_processed.txt"
      out_model_file: "m1.bin"
      model_description: "simple bible fasttext model"
      vector_size: 200
      epochs: 100
      window_size: 10

  veld_step_3_export_fasttext:
    extends:
      file: ./code/veld_code__fasttext/veld_export.yaml
      service: veld_export
    volumes:
      - ./data/models/fasttext/:/veld/input/
      - ./data/vectors/fasttext/:/veld/output/
    environment:
      in_model_file: "m1.bin"
      out_vector_file: "m1.pkl"
    depends_on:
      veld_step_3_train_fasttext:
        condition: service_completed_successfully