diff --git a/src/llmcompressor/transformers/finetune/data/base.py b/src/llmcompressor/transformers/finetune/data/base.py index ddc348b89..81a3fc95f 100644 --- a/src/llmcompressor/transformers/finetune/data/base.py +++ b/src/llmcompressor/transformers/finetune/data/base.py @@ -253,7 +253,7 @@ def tokenize(self, data: LazyRow) -> Dict[str, Any]: # store unpadded prompt so we can mask out correct number of elements in labels if prompt is not None: data[self.PROMPT_KEY] = self.processor( - prompt, + text=prompt, max_length=self.max_seq_length, truncation=True, )["input_ids"]