diff --git a/llmfoundry/data/finetuning/dataloader.py b/llmfoundry/data/finetuning/dataloader.py index 97725ce78c..4e1c3bbf9f 100644 --- a/llmfoundry/data/finetuning/dataloader.py +++ b/llmfoundry/data/finetuning/dataloader.py @@ -190,7 +190,7 @@ def build_finetuning_dataloader(cfg: DictConfig, max_seq_len=cfg.dataset.max_seq_len, preprocessing_fn=preprocessing_fn, tokenizer=tokenizer, - hf_kwargs=cfg.dataset.get('kwargs', {})) + hf_kwargs=cfg.dataset.get('hf_kwargs', {})) # Ensure dataset is large enough. if cfg.drop_last: diff --git a/tests/data/test_dataloader.py b/tests/data/test_dataloader.py index 7f99eeda25..44d0442a87 100644 --- a/tests/data/test_dataloader.py +++ b/tests/data/test_dataloader.py @@ -326,7 +326,7 @@ def test_finetuning_dataloader_safe_load(hf_name: str, 'decoder_only_format': True, 'shuffle': True, 'safe_load': True, - 'kwargs': { + 'hf_kwargs': { 'revision': hf_revision } },