diff --git a/config/train_nougat.yaml b/config/train_nougat.yaml index a27dcdb..e617d84 100644 --- a/config/train_nougat.yaml +++ b/config/train_nougat.yaml @@ -2,6 +2,7 @@ resume_from_checkpoint_path: null result_path: "result" model_path: null dataset_paths: ["path/to/train.jsonl"] +datasets_root: "paired/output" tokenizer: "dataset/tokenizer.json" exp_name: "nougat" train_batch_sizes: [1] diff --git a/train.py b/train.py index ca4f0d0..c89b636 100644 --- a/train.py +++ b/train.py @@ -154,6 +154,7 @@ def train(config): nougat_model=model_module.model, max_length=config.max_length, split=split, + root_name=config.datasets_root ) ) data_module.train_datasets = datasets["train"]