diff --git a/engines/python/setup/djl_python/streaming_utils.py b/engines/python/setup/djl_python/streaming_utils.py index a0f8481ee..c11554690 100644 --- a/engines/python/setup/djl_python/streaming_utils.py +++ b/engines/python/setup/djl_python/streaming_utils.py @@ -81,7 +81,7 @@ def _hf_model_stream_generator(model, tokenizer, inputs, **kwargs): @torch.inference_mode() def _transformers_neuronx_stream_generator(model, tokenizer, inputs, **kwargs): - sequence_length = kwargs.get("seq_len", 50) + sequence_length = kwargs.get("seq_length", 50) top_k = kwargs.get("top_k", 50) input_ids = torch.as_tensor( [tokenizer.encode(text) for text in inputs])