diff --git a/internlm/data/utils.py b/internlm/data/utils.py index 3eee9d9..724fb9f 100644 --- a/internlm/data/utils.py +++ b/internlm/data/utils.py @@ -5,7 +5,7 @@ import torch from internlm.core.context import global_context as gpc -DATASET_TYPE_IDS_MAP = {"en": 0, "cn": 1} +DATASET_TYPE_IDS_MAP = {"en": 0, "cn": 1, "code": 2} def get_dataset_type_id(path):