Pass datasets trust_remote_code (#31406)

* Pass datasets trust_remote_code

* Pass trust_remote_code in more tests

* Add trust_remote_dataset_code arg to some tests

* Revert "Temporarily pin datasets upper version to fix CI"

This reverts commit b7672826.

* Pass trust_remote_code in librispeech_asr_dummy docstrings

* Revert "Pin datasets<2.20.0 for examples"

This reverts commit 833fc17a.

* Pass trust_remote_code to all examples

* Revert "Add trust_remote_dataset_code arg to some tests" to research_projects

* Pass trust_remote_code to tests

* Pass trust_remote_code to docstrings

* Fix flax examples tests requirements

* Pass trust_remote_dataset_code arg to tests

* Replace trust_remote_dataset_code with trust_remote_code in one example

* Fix duplicate trust_remote_code

* Replace args.trust_remote_dataset_code with args.trust_remote_code

* Replace trust_remote_dataset_code with trust_remote_code in parser

* Replace trust_remote_dataset_code with trust_remote_code in dataclasses

* Replace trust_remote_dataset_code with trust_remote_code arg

examples/pytorch/instance-segmentation/run_instance_segmentation_no_trainer.py

@@ -71,6 +71,15 @@ def parse_args():
         help="Name of the dataset on the hub.",
         default="qubvel-hf/ade20k-mini",
     )
+    parser.add_argument(
+        "--trust_remote_code",
+        action="store_true",
+        help=(
+            "Whether to trust the execution of code from datasets/models defined on the Hub."
+            " This option should only be set to `True` for repositories you trust and in which you have read the"
+            " code, as it will execute code present on the Hub on your local machine."
+        ),
+    )
     parser.add_argument(
         "--image_height",
         type=int,
@@ -425,7 +434,7 @@ def main():
 
     # In distributed training, the load_dataset function guarantees that only one local process can concurrently
     # download the dataset.
-    dataset = load_dataset(args.dataset_name, cache_dir=args.cache_dir)
+    dataset = load_dataset(args.dataset_name, cache_dir=args.cache_dir, trust_remote_code=args.trust_remote_code)
 
     # We need to specify the label2id mapping for the model
     # it is a mapping from semantic class name to class index.

examples/pytorch/language-modeling/run_clm.py

@@ -124,9 +124,9 @@ class ModelArguments:
         default=False,
         metadata={
             "help": (
-                "Whether or not to allow for custom models defined on the Hub in their own modeling files. This option "
-                "should only be set to `True` for repositories you trust and in which you have read the code, as it will "
-                "execute code present on the Hub on your local machine."
+                "Whether to trust the execution of code from datasets/models defined on the Hub."
+                " This option should only be set to `True` for repositories you trust and in which you have read the"
+                " code, as it will execute code present on the Hub on your local machine."
             )
         },
     )
@@ -312,6 +312,7 @@ def main():
             cache_dir=model_args.cache_dir,
             token=model_args.token,
             streaming=data_args.streaming,
+            trust_remote_code=model_args.trust_remote_code,
         )
         if "validation" not in raw_datasets.keys():
             raw_datasets["validation"] = load_dataset(
@@ -321,6 +322,7 @@ def main():
                 cache_dir=model_args.cache_dir,
                 token=model_args.token,
                 streaming=data_args.streaming,
+                trust_remote_code=model_args.trust_remote_code,
             )
             raw_datasets["train"] = load_dataset(
                 data_args.dataset_name,
@@ -329,6 +331,7 @@ def main():
                 cache_dir=model_args.cache_dir,
                 token=model_args.token,
                 streaming=data_args.streaming,
+                trust_remote_code=model_args.trust_remote_code,
             )
     else:
         data_files = {}

examples/pytorch/language-modeling/run_clm_no_trainer.py

@@ -195,12 +195,11 @@ def parse_args():
     parser.add_argument("--hub_token", type=str, help="The token to use to push to the Model Hub.")
     parser.add_argument(
         "--trust_remote_code",
-        type=bool,
-        default=False,
+        action="store_true",
         help=(
-            "Whether or not to allow for custom models defined on the Hub in their own modeling files. This option "
-            "should only be set to `True` for repositories you trust and in which you have read the code, as it will "
-            "execute code present on the Hub on your local machine."
+            "Whether to trust the execution of code from datasets/models defined on the Hub."
+            " This option should only be set to `True` for repositories you trust and in which you have read the"
+            " code, as it will execute code present on the Hub on your local machine."
         ),
     )
     parser.add_argument(
@@ -327,17 +326,21 @@ def main():
     # download the dataset.
     if args.dataset_name is not None:
         # Downloading and loading a dataset from the hub.
-        raw_datasets = load_dataset(args.dataset_name, args.dataset_config_name)
+        raw_datasets = load_dataset(
+            args.dataset_name, args.dataset_config_name, trust_remote_code=args.trust_remote_code
+        )
         if "validation" not in raw_datasets.keys():
             raw_datasets["validation"] = load_dataset(
                 args.dataset_name,
                 args.dataset_config_name,
                 split=f"train[:{args.validation_split_percentage}%]",
+                trust_remote_code=args.trust_remote_code,
             )
             raw_datasets["train"] = load_dataset(
                 args.dataset_name,
                 args.dataset_config_name,
                 split=f"train[{args.validation_split_percentage}%:]",
+                trust_remote_code=args.trust_remote_code,
             )
     else:
         data_files = {}

examples/pytorch/language-modeling/run_fim.py

@@ -127,9 +127,9 @@ class ModelArguments:
         default=False,
         metadata={
             "help": (
-                "Whether or not to allow for custom models defined on the Hub in their own modeling files. This option"
-                "should only be set to `True` for repositories you trust and in which you have read the code, as it will "
-                "execute code present on the Hub on your local machine."
+                "Whether to trust the execution of code from datasets/models defined on the Hub."
+                " This option should only be set to `True` for repositories you trust and in which you have read the"
+                " code, as it will execute code present on the Hub on your local machine."
             )
         },
     )
@@ -382,6 +382,7 @@ def main():
             cache_dir=model_args.cache_dir,
             token=model_args.token,
             streaming=data_args.streaming,
+            trust_remote_code=model_args.trust_remote_code,
         )
         if "validation" not in raw_datasets.keys():
             raw_datasets["validation"] = load_dataset(
@@ -391,6 +392,7 @@ def main():
                 cache_dir=model_args.cache_dir,
                 token=model_args.token,
                 streaming=data_args.streaming,
+                trust_remote_code=model_args.trust_remote_code,
             )
             raw_datasets["train"] = load_dataset(
                 data_args.dataset_name,
@@ -399,6 +401,7 @@ def main():
                 cache_dir=model_args.cache_dir,
                 token=model_args.token,
                 streaming=data_args.streaming,
+                trust_remote_code=model_args.trust_remote_code,
             )
     else:
         data_files = {}

examples/pytorch/language-modeling/run_fim_no_trainer.py

@@ -257,12 +257,11 @@ def parse_args():
     parser.add_argument("--hub_token", type=str, help="The token to use to push to the Model Hub.")
     parser.add_argument(
         "--trust_remote_code",
-        type=bool,
-        default=False,
+        action="store_true",
         help=(
-            "Whether or not to allow for custom models defined on the Hub in their own modeling files. This option"
-            "should only be set to `True` for repositories you trust and in which you have read the code, as it will "
-            "execute code present on the Hub on your local machine."
+            "Whether to trust the execution of code from datasets/models defined on the Hub."
+            " This option should only be set to `True` for repositories you trust and in which you have read the"
+            " code, as it will execute code present on the Hub on your local machine."
         ),
     )
     parser.add_argument(
@@ -395,17 +394,21 @@ def main():
     # download the dataset.
     if args.dataset_name is not None:
         # Downloading and loading a dataset from the hub.
-        raw_datasets = load_dataset(args.dataset_name, args.dataset_config_name)
+        raw_datasets = load_dataset(
+            args.dataset_name, args.dataset_config_name, trust_remote_code=args.trust_remote_code
+        )
         if "validation" not in raw_datasets.keys():
             raw_datasets["validation"] = load_dataset(
                 args.dataset_name,
                 args.dataset_config_name,
                 split=f"train[:{args.validation_split_percentage}%]",
+                trust_remote_code=args.trust_remote_code,
             )
             raw_datasets["train"] = load_dataset(
                 args.dataset_name,
                 args.dataset_config_name,
                 split=f"train[{args.validation_split_percentage}%:]",
+                trust_remote_code=args.trust_remote_code,
             )
     else:
         data_files = {}

examples/pytorch/language-modeling/run_mlm.py

@@ -121,9 +121,9 @@ class ModelArguments:
         default=False,
         metadata={
             "help": (
-                "Whether or not to allow for custom models defined on the Hub in their own modeling files. This option "
-                "should only be set to `True` for repositories you trust and in which you have read the code, as it will "
-                "execute code present on the Hub on your local machine."
+                "Whether to trust the execution of code from datasets/models defined on the Hub."
+                " This option should only be set to `True` for repositories you trust and in which you have read the"
+                " code, as it will execute code present on the Hub on your local machine."
             )
         },
     )
@@ -324,6 +324,7 @@ def main():
             cache_dir=model_args.cache_dir,
             token=model_args.token,
             streaming=data_args.streaming,
+            trust_remote_code=model_args.trust_remote_code,
         )
         if "validation" not in raw_datasets.keys():
             raw_datasets["validation"] = load_dataset(
@@ -333,6 +334,7 @@ def main():
                 cache_dir=model_args.cache_dir,
                 token=model_args.token,
                 streaming=data_args.streaming,
+                trust_remote_code=model_args.trust_remote_code,
             )
             raw_datasets["train"] = load_dataset(
                 data_args.dataset_name,
@@ -341,6 +343,7 @@ def main():
                 cache_dir=model_args.cache_dir,
                 token=model_args.token,
                 streaming=data_args.streaming,
+                trust_remote_code=model_args.trust_remote_code,
             )
     else:
         data_files = {}

examples/pytorch/language-modeling/run_mlm_no_trainer.py

@@ -202,12 +202,11 @@ def parse_args():
     parser.add_argument("--hub_token", type=str, help="The token to use to push to the Model Hub.")
     parser.add_argument(
         "--trust_remote_code",
-        type=bool,
-        default=False,
+        action="store_true",
         help=(
-            "Whether or not to allow for custom models defined on the Hub in their own modeling files. This option "
-            "should only be set to `True` for repositories you trust and in which you have read the code, as it will "
-            "execute code present on the Hub on your local machine."
+            "Whether to trust the execution of code from datasets/models defined on the Hub."
+            " This option should only be set to `True` for repositories you trust and in which you have read the"
+            " code, as it will execute code present on the Hub on your local machine."
         ),
     )
     parser.add_argument(
@@ -334,17 +333,21 @@ def main():
     # download the dataset.
     if args.dataset_name is not None:
         # Downloading and loading a dataset from the hub.
-        raw_datasets = load_dataset(args.dataset_name, args.dataset_config_name)
+        raw_datasets = load_dataset(
+            args.dataset_name, args.dataset_config_name, trust_remote_code=args.trust_remote_code
+        )
         if "validation" not in raw_datasets.keys():
             raw_datasets["validation"] = load_dataset(
                 args.dataset_name,
                 args.dataset_config_name,
                 split=f"train[:{args.validation_split_percentage}%]",
+                trust_remote_code=args.trust_remote_code,
             )
             raw_datasets["train"] = load_dataset(
                 args.dataset_name,
                 args.dataset_config_name,
                 split=f"train[{args.validation_split_percentage}%:]",
+                trust_remote_code=args.trust_remote_code,
             )
     else:
         data_files = {}

examples/pytorch/language-modeling/run_plm.py

@@ -133,6 +133,16 @@ class DataTrainingArguments:
     dataset_config_name: Optional[str] = field(
         default=None, metadata={"help": "The configuration name of the dataset to use (via the datasets library)."}
     )
+    trust_remote_code: bool = field(
+        default=False,
+        metadata={
+            "help": (
+                "Whether to trust the execution of code from datasets/models defined on the Hub."
+                " This option should only be set to `True` for repositories you trust and in which you have read the"
+                " code, as it will execute code present on the Hub on your local machine."
+            )
+        },
+    )
     train_file: Optional[str] = field(default=None, metadata={"help": "The input training data file (a text file)."})
     validation_file: Optional[str] = field(
         default=None,
@@ -292,6 +302,7 @@ def main():
             data_args.dataset_config_name,
             cache_dir=model_args.cache_dir,
             token=model_args.token,
+            trust_remote_code=data_args.trust_remote_code,
         )
         if "validation" not in raw_datasets.keys():
             raw_datasets["validation"] = load_dataset(
@@ -300,6 +311,7 @@ def main():
                 split=f"train[:{data_args.validation_split_percentage}%]",
                 cache_dir=model_args.cache_dir,
                 token=model_args.token,
+                trust_remote_code=data_args.trust_remote_code,
             )
             raw_datasets["train"] = load_dataset(
                 data_args.dataset_name,
@@ -307,6 +319,7 @@ def main():
                 split=f"train[{data_args.validation_split_percentage}%:]",
                 cache_dir=model_args.cache_dir,
                 token=model_args.token,
+                trust_remote_code=data_args.trust_remote_code,
             )
     else:
         data_files = {}

examples/pytorch/multiple-choice/run_swag_no_trainer.py

@@ -184,12 +184,11 @@ def parse_args():
     parser.add_argument("--hub_token", type=str, help="The token to use to push to the Model Hub.")
     parser.add_argument(
         "--trust_remote_code",
-        type=bool,
-        default=False,
+        action="store_true",
         help=(
-            "Whether or not to allow for custom models defined on the Hub in their own modeling files. This option "
-            "should only be set to `True` for repositories you trust and in which you have read the code, as it will "
-            "execute code present on the Hub on your local machine."
+            "Whether to trust the execution of code from datasets/models defined on the Hub."
+            " This option should only be set to `True` for repositories you trust and in which you have read the"
+            " code, as it will execute code present on the Hub on your local machine."
         ),
     )
     parser.add_argument(
@@ -351,7 +350,9 @@ def main():
     # download the dataset.
     if args.dataset_name is not None:
         # Downloading and loading a dataset from the hub.
-        raw_datasets = load_dataset(args.dataset_name, args.dataset_config_name)
+        raw_datasets = load_dataset(
+            args.dataset_name, args.dataset_config_name, trust_remote_code=args.trust_remote_code
+        )
     else:
         data_files = {}
         if args.train_file is not None:

examples/pytorch/object-detection/run_object_detection.py

@@ -313,9 +313,9 @@ class ModelArguments:
         default=False,
         metadata={
             "help": (
-                "Whether or not to allow for custom models defined on the Hub in their own modeling files. This option "
-                "should only be set to `True` for repositories you trust and in which you have read the code, as it will "
-                "execute code present on the Hub on your local machine."
+                "Whether to trust the execution of code from datasets/models defined on the Hub."
+                " This option should only be set to `True` for repositories you trust and in which you have read the"
+                " code, as it will execute code present on the Hub on your local machine."
             )
         },
     )
@@ -383,7 +383,9 @@ def main():
     # Load dataset, prepare splits
     # ------------------------------------------------------------------------------------------------
 
-    dataset = load_dataset(data_args.dataset_name, cache_dir=model_args.cache_dir)
+    dataset = load_dataset(
+        data_args.dataset_name, cache_dir=model_args.cache_dir, trust_remote_code=model_args.trust_remote_code
+    )
 
     # If we don't have a validation split, split off a percentage of train as validation
     data_args.train_val_split = None if "validation" in dataset.keys() else data_args.train_val_split

examples/pytorch/object-detection/run_object_detection_no_trainer.py

@@ -340,12 +340,11 @@ def parse_args():
     parser.add_argument("--hub_token", type=str, help="The token to use to push to the Model Hub.")
     parser.add_argument(
         "--trust_remote_code",
-        type=bool,
-        default=False,
+        action="store_true",
         help=(
-            "Whether or not to allow for custom models defined on the Hub in their own modeling files. This option "
-            "should only be set to `True` for repositories you trust and in which you have read the code, as it will "
-            "execute code present on the Hub on your local machine."
+            "Whether to trust the execution of code from datasets/models defined on the Hub."
+            " This option should only be set to `True` for repositories you trust and in which you have read the"
+            " code, as it will execute code present on the Hub on your local machine."
         ),
     )
     parser.add_argument(
@@ -445,7 +444,7 @@ def main():
     # Load dataset
     # In distributed training, the load_dataset function guarantees that only one local process can concurrently
     # download the dataset.
-    dataset = load_dataset(args.dataset_name, cache_dir=args.cache_dir)
+    dataset = load_dataset(args.dataset_name, cache_dir=args.cache_dir, trust_remote_code=args.trust_remote_code)
 
     # If we don't have a validation split, split off a percentage of train as validation.
     args.train_val_split = None if "validation" in dataset.keys() else args.train_val_split

examples/pytorch/question-answering/run_qa.py

@@ -93,9 +93,9 @@ class ModelArguments:
         default=False,
         metadata={
             "help": (
-                "Whether or not to allow for custom models defined on the Hub in their own modeling files. This option "
-                "should only be set to `True` for repositories you trust and in which you have read the code, as it will "
-                "execute code present on the Hub on your local machine."
+                "Whether to trust the execution of code from datasets/models defined on the Hub."
+                " This option should only be set to `True` for repositories you trust and in which you have read the"
+                " code, as it will execute code present on the Hub on your local machine."
             )
         },
     )
@@ -301,6 +301,7 @@ def main():
             data_args.dataset_config_name,
             cache_dir=model_args.cache_dir,
             token=model_args.token,
+            trust_remote_code=model_args.trust_remote_code,
         )
     else:
         data_files = {}

examples/pytorch/question-answering/run_qa_beam_search.py

@@ -101,6 +101,16 @@ class DataTrainingArguments:
     dataset_config_name: Optional[str] = field(
         default=None, metadata={"help": "The configuration name of the dataset to use (via the datasets library)."}
     )
+    trust_remote_code: bool = field(
+        default=False,
+        metadata={
+            "help": (
+                "Whether to trust the execution of code from datasets/models defined on the Hub."
+                " This option should only be set to `True` for repositories you trust and in which you have read the"
+                " code, as it will execute code present on the Hub on your local machine."
+            )
+        },
+    )
     train_file: Optional[str] = field(default=None, metadata={"help": "The input training data file (a text file)."})
     validation_file: Optional[str] = field(
         default=None,
@@ -289,6 +299,7 @@ def main():
             data_args.dataset_config_name,
             cache_dir=model_args.cache_dir,
             token=model_args.token,
+            trust_remote_code=data_args.trust_remote_code,
         )
     else:
         data_files = {}

examples/pytorch/question-answering/run_qa_beam_search_no_trainer.py

@@ -100,6 +100,15 @@ def parse_args():
         default=None,
         help="The configuration name of the dataset to use (via the datasets library).",
     )
+    parser.add_argument(
+        "--trust_remote_code",
+        action="store_true",
+        help=(
+            "Whether to trust the execution of code from datasets/models defined on the Hub."
+            " This option should only be set to `True` for repositories you trust and in which you have read the"
+            " code, as it will execute code present on the Hub on your local machine."
+        ),
+    )
     parser.add_argument(
         "--train_file", type=str, default=None, help="A csv or a json file containing the training data."
     )
@@ -356,7 +365,9 @@ def main():
     # download the dataset.
     if args.dataset_name is not None:
         # Downloading and loading a dataset from the hub.
-        raw_datasets = load_dataset(args.dataset_name, args.dataset_config_name)
+        raw_datasets = load_dataset(
+            args.dataset_name, args.dataset_config_name, trust_remote_code=args.trust_remote_code
+        )
     else:
         data_files = {}
         if args.train_file is not None:

examples/pytorch/question-answering/run_qa_no_trainer.py

@@ -275,12 +275,11 @@ def parse_args():
     parser.add_argument("--hub_token", type=str, help="The token to use to push to the Model Hub.")
     parser.add_argument(
         "--trust_remote_code",
-        type=bool,
-        default=False,
+        action="store_true",
         help=(
-            "Whether or not to allow for custom models defined on the Hub in their own modeling files. This option "
-            "should only be set to `True` for repositories you trust and in which you have read the code, as it will "
-            "execute code present on the Hub on your local machine."
+            "Whether to trust the execution of code from datasets/models defined on the Hub."
+            " This option should only be set to `True` for repositories you trust and in which you have read the"
+            " code, as it will execute code present on the Hub on your local machine."
         ),
     )
     parser.add_argument(
@@ -404,7 +403,9 @@ def main():
     # download the dataset.
     if args.dataset_name is not None:
         # Downloading and loading a dataset from the hub.
-        raw_datasets = load_dataset(args.dataset_name, args.dataset_config_name)
+        raw_datasets = load_dataset(
+            args.dataset_name, args.dataset_config_name, trust_remote_code=args.trust_remote_code
+        )
     else:
         data_files = {}
         if args.train_file is not None:

examples/pytorch/question-answering/run_seq2seq_qa.py

@@ -93,9 +93,9 @@ class ModelArguments:
         default=False,
         metadata={
             "help": (
-                "Whether or not to allow for custom models defined on the Hub in their own modeling files. This option "
-                "should only be set to `True` for repositories you trust and in which you have read the code, as it will "
-                "execute code present on the Hub on your local machine."
+                "Whether to trust the execution of code from datasets/models defined on the Hub."
+                " This option should only be set to `True` for repositories you trust and in which you have read the"
+                " code, as it will execute code present on the Hub on your local machine."
             )
         },
     )
@@ -346,6 +346,7 @@ def main():
             data_args.dataset_config_name,
             cache_dir=model_args.cache_dir,
             token=model_args.token,
+            trust_remote_code=model_args.trust_remote_code,
         )
     else:
         data_files = {}

examples/pytorch/semantic-segmentation/run_semantic_segmentation.py

@@ -165,9 +165,9 @@ class ModelArguments:
         default=False,
         metadata={
             "help": (
-                "Whether or not to allow for custom models defined on the Hub in their own modeling files. This option "
-                "should only be set to `True` for repositories you trust and in which you have read the code, as it will "
-                "execute code present on the Hub on your local machine."
+                "Whether to trust the execution of code from datasets/models defined on the Hub."
+                " This option should only be set to `True` for repositories you trust and in which you have read the"
+                " code, as it will execute code present on the Hub on your local machine."
             )
         },
     )
@@ -233,7 +233,9 @@ def main():
     # In distributed training, the load_dataset function guarantees that only one local process can concurrently
     # download the dataset.
     # TODO support datasets from local folders
-    dataset = load_dataset(data_args.dataset_name, cache_dir=model_args.cache_dir)
+    dataset = load_dataset(
+        data_args.dataset_name, cache_dir=model_args.cache_dir, trust_remote_code=model_args.trust_remote_code
+    )
 
     # Rename column names to standardized names (only "image" and "label" need to be present)
     if "pixel_values" in dataset["train"].column_names:

examples/pytorch/semantic-segmentation/run_semantic_segmentation_no_trainer.py

@@ -180,12 +180,11 @@ def parse_args():
     parser.add_argument("--hub_token", type=str, help="The token to use to push to the Model Hub.")
     parser.add_argument(
         "--trust_remote_code",
-        type=bool,
-        default=False,
+        action="store_true",
         help=(
-            "Whether or not to allow for custom models defined on the Hub in their own modeling files. This option "
-            "should only be set to `True` for repositories you trust and in which you have read the code, as it will "
-            "execute code present on the Hub on your local machine."
+            "Whether to trust the execution of code from datasets/models defined on the Hub."
+            " This option should only be set to `True` for repositories you trust and in which you have read the"
+            " code, as it will execute code present on the Hub on your local machine."
         ),
     )
     parser.add_argument(
@@ -294,7 +293,7 @@ def main():
     # In distributed training, the load_dataset function guarantees that only one local process can concurrently
     # download the dataset.
     # TODO support datasets from local folders
-    dataset = load_dataset(args.dataset_name, cache_dir=args.cache_dir)
+    dataset = load_dataset(args.dataset_name, cache_dir=args.cache_dir, trust_remote_code=args.trust_remote_code)
 
     # Rename column names to standardized names (only "image" and "label" need to be present)
     if "pixel_values" in dataset["train"].column_names:

examples/pytorch/speech-pretraining/run_wav2vec2_pretraining_no_trainer.py

@@ -71,6 +71,15 @@ def parse_args():
         required=True,
         help="The names of the training data set splits to use (via the datasets library).",
     )
+    parser.add_argument(
+        "--trust_remote_code",
+        action="store_true",
+        help=(
+            "Whether to trust the execution of code from datasets/models defined on the Hub."
+            " This option should only be set to `True` for repositories you trust and in which you have read the"
+            " code, as it will execute code present on the Hub on your local machine."
+        ),
+    )
     parser.add_argument(
         "--preprocessing_num_workers",
         type=int,
@@ -446,6 +455,7 @@ def main():
             dataset_config_name,
             split=train_split_name,
             cache_dir=args.cache_dir,
+            trust_remote_code=args.trust_remote_code,
         )
         datasets_splits.append(dataset_split)
 

examples/pytorch/speech-recognition/run_speech_recognition_ctc.py

@@ -255,9 +255,9 @@ class DataTrainingArguments:
         default=False,
         metadata={
             "help": (
-                "Whether or not to allow for custom models defined on the Hub in their own modeling files. This option "
-                "should only be set to `True` for repositories you trust and in which you have read the code, as it will "
-                "execute code present on the Hub on your local machine."
+                "Whether to trust the execution of code from datasets/models defined on the Hub."
+                " This option should only be set to `True` for repositories you trust and in which you have read the"
+                " code, as it will execute code present on the Hub on your local machine."
             )
         },
     )
@@ -454,6 +454,7 @@ def main():
             data_args.dataset_config_name,
             split=data_args.train_split_name,
             token=data_args.token,
+            trust_remote_code=data_args.trust_remote_code,
         )
 
         if data_args.audio_column_name not in raw_datasets["train"].column_names:
@@ -479,6 +480,7 @@ def main():
             data_args.dataset_config_name,
             split=data_args.eval_split_name,
             token=data_args.token,
+            trust_remote_code=data_args.trust_remote_code,
         )
 
         if data_args.max_eval_samples is not None: