train_config.json

{
  "model_name": "google/byt5-small",
  "revision": "68377bdc18a2ffec8a0533fef03b1c513a4dd49d",
  "train_file": "data/processed/train.tsv",
  "valid_file": "data/processed/valid.tsv",
  "output_dir": "models/run",
  "best_dir": "models/best",
  "n_train": 200000,
  "n_eval": 2000,
  "max_source_length": 256,
  "max_target_length": 640,
  "learning_rate": 0.0003,
  "optim": "adafactor",
  "per_device_batch": 16,
  "grad_accum": 2,
  "epochs": 1.0,
  "max_steps": -1,
  "warmup_steps": 200,
  "eval_steps": 1000,
  "eval_batch": 64,
  "logging_steps": 50,
  "seed": 42
}