diff --git a/engines/python/setup/djl_python/streaming_utils.py b/engines/python/setup/djl_python/streaming_utils.py index 87241dffb..1d466d97f 100644 --- a/engines/python/setup/djl_python/streaming_utils.py +++ b/engines/python/setup/djl_python/streaming_utils.py @@ -74,7 +74,7 @@ def use_hf_default_streamer(model, tokenizer, inputs, device, **kwargs): if device is not None: input_tokens = input_tokens.to(device) - streamer = HFStreamer(tokenizer, skip_special_token=True) + streamer = HFStreamer(tokenizer, skip_special_tokens=True) generation_kwargs = dict(input_tokens, streamer=streamer, **kwargs) def run_generation(model, **kwargs):