config.fiu-eng.opustrainer.yml

####
# Example of a production config
# Change language pair, experiment name, datasets and other settings if needed
# Training low resource languages might require more tuning of pipeline/training/configs
###

experiment:
  dirname: opustrainer
  name: baseline
  langpairs:
    - et-en
    - fi-en
    - hu-en

  #URL to the OPUS-MT model to use as the teacher
  opusmt-teacher: "https://object.pouta.csc.fi/Tatoeba-MT-models/fiu-eng/opus4m-2020-08-12.zip"
  
  # Specify if the teacher and the student are many2one
  one2many-teacher: False
  one2many-student: False

  #URL to the OPUS-MT model to use as the backward model
  opusmt-backward: "https://object.pouta.csc.fi/Tatoeba-MT-models/eng-fiu/opus2m-2020-08-01.zip"
  one2many-backward: True

  teacher-ensemble: 1
  
  parallel-max-sentences: 10000000
  split-length: 1000000

  best-model: perplexity
  
  spm-sample-size: 1000000
  
  opustrainer:
    model: student # Ideally, could be teacher or backward
    path: "configs/opustrainer/config.fiu-eng.opustrainer.stages.yml" # This assumes you already know the paths to the data

datasets:
  train:
    - tc_Tatoeba-Challenge-v2023-09-26
  devtest:
    - flores_dev
  test:
    - flores_devtest