config.json

{
  "random_seed": 42,
  "min_samples": 100,
  "serializer": {
    "numeric_precision": 2,
    "model_max_len_tokens": 4096,
    "appx_chars_per_token": 3.5,
    "max_chars": 14336
  },
  "classification": {
    "max_samples_per_dataset": 10000,
    "train_ratio": 0.9,
    "max_label_ratio": 0.1,
    "max_num_labels": 50
  },
  "retrieval": {
    "max_samples_per_dataset": 10000,
    "num_queries_per_type": 10000,
    "min_matches": 5,
    "max_conditions": 3
  }
}