Add trust_remote_code for loading datasets in the audio classification example

examples/audio-classification/README.md

@@ -94,7 +94,8 @@ python ../gaudi_spawn.py \
     --use_hpu_graphs_for_inference \
     --gaudi_config_name Habana/wav2vec2 \
     --throughput_warmup_steps 3 \
-    --bf16
+    --bf16 \
+    --trust_remote_code True
 ```
 
 On 8 HPUs, this script should run in ~12 minutes and yield an accuracy of **80.49%**.
@@ -141,7 +142,8 @@ python ../gaudi_spawn.py \
     --use_hpu_graphs_for_inference \
     --gaudi_config_name Habana/wav2vec2 \
     --throughput_warmup_steps 3 \
-    --deepspeed ../../tests/configs/deepspeed_zero_2.json
+    --deepspeed ../../tests/configs/deepspeed_zero_2.json \
+    --trust_remote_code True
 ```
 
 [The documentation](https://huggingface.co/docs/optimum/habana/usage_guides/deepspeed) provides more information about how to use DeepSpeed within Optimum Habana.

examples/audio-classification/run_audio_classification.py

@@ -167,9 +167,9 @@ 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."
             )
         },
     )
@@ -254,12 +254,14 @@ def main():
         data_args.dataset_config_name,
         split=data_args.train_split_name,
         token=model_args.token,
+        trust_remote_code=model_args.trust_remote_code,
     )
     raw_datasets["eval"] = load_dataset(
         data_args.dataset_name,
         data_args.dataset_config_name,
         split=data_args.eval_split_name,
         token=model_args.token,
+        trust_remote_code=model_args.trust_remote_code,
     )
 
     if data_args.audio_column_name not in raw_datasets["train"].column_names:

examples/contrastive-image-text/run_bridgetower.py

@@ -102,9 +102,9 @@ 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."
             )
         },
     )
@@ -203,9 +203,9 @@ def __post_init__(self):
             if self.validation_file is not None:
                 extension = self.validation_file.split(".")[-1]
                 assert extension in ["csv", "json"], "`validation_file` should be a csv or a json file."
-            if self.validation_file is not None:
-                extension = self.validation_file.split(".")[-1]
-                assert extension == "json", "`validation_file` should be a json file."
+            if self.test_file is not None:
+                extension = self.test_file.split(".")[-1]
+                assert extension in ["csv", "json"], "`test_file` should be a csv or a json file."
 
 
 dataset_name_mapping = {
@@ -328,6 +328,7 @@ def main():
             data_dir=data_args.data_dir,
             token=model_args.token,
             revision=data_args.dataset_revision,
+            trust_remote_code=model_args.trust_remote_code,
         )
     else:
         data_files = {}

examples/contrastive-image-text/run_clip.py

@@ -107,9 +107,9 @@ 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."
             )
         },
     )
@@ -201,9 +201,9 @@ def __post_init__(self):
             if self.validation_file is not None:
                 extension = self.validation_file.split(".")[-1]
                 assert extension in ["csv", "json"], "`validation_file` should be a csv or a json file."
-            if self.validation_file is not None:
-                extension = self.validation_file.split(".")[-1]
-                assert extension == "json", "`validation_file` should be a json file."
+            if self.test_file is not None:
+                extension = self.test_file.split(".")[-1]
+                assert extension in ["csv", "json"], "`test_file` should be a csv or a json file."
 
 
 dataset_name_mapping = {
@@ -325,6 +325,7 @@ def main():
             keep_in_memory=False,
             data_dir=data_args.data_dir,
             token=model_args.token,
+            trust_remote_code=model_args.trust_remote_code,
         )
     else:
         data_files = {}

examples/image-classification/run_image_classification.py

@@ -172,9 +172,9 @@ 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."
             )
         },
     )
@@ -259,6 +259,7 @@ 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/language-modeling/run_clm.py

@@ -131,9 +131,9 @@ 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."
             )
         },
     )
@@ -341,6 +341,7 @@ 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(
@@ -350,6 +351,7 @@ 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,
@@ -358,6 +360,7 @@ 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/language-modeling/run_lora_clm.py

@@ -103,7 +103,11 @@ class ModelArguments:
     trust_remote_code: bool = field(
         default=False,
         metadata={
-            "help": "should enable when using custom model architecture that is not yet part of the Hugging Face transformers package like MPT)."
+            "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."
+            )
         },
     )
     use_cache: bool = field(
@@ -502,6 +506,7 @@ 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,
         )
 
         if "validation" not in raw_datasets.keys() and training_args.do_eval:
@@ -511,13 +516,15 @@ def main():
                 split=f"train[:{data_args.validation_split_percentage}%]",
                 cache_dir=model_args.cache_dir,
                 token=model_args.token,
+                trust_remote_code=model_args.trust_remote_code,
             )
             raw_datasets["train"] = load_dataset(
                 data_args.dataset_name,
                 data_args.dataset_config_name,
                 split=f"train[{data_args.validation_split_percentage}%:]",
                 cache_dir=model_args.cache_dir,
                 token=model_args.token,
+                trust_remote_code=model_args.trust_remote_code,
             )
     else:
         data_files = {}

examples/language-modeling/run_mlm.py

@@ -129,9 +129,9 @@ 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."
             )
         },
     )
@@ -340,6 +340,7 @@ 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(
@@ -349,6 +350,7 @@ 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,
@@ -357,6 +359,7 @@ 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/language-modeling/run_prompt_tuning_clm.py

@@ -114,9 +114,9 @@ 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."
             )
         },
     )
@@ -248,6 +248,7 @@ 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 data_args.dataset_name == "ought/raft" and data_args.dataset_config_name == "twitter_complaints":
         text_column = "Tweet text"

examples/question-answering/run_qa.py

@@ -102,9 +102,9 @@ 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."
             )
         },
     )
@@ -319,6 +319,7 @@ 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/question-answering/run_seq2seq_qa.py

@@ -102,9 +102,9 @@ 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."
             )
         },
     )
@@ -364,6 +364,7 @@ 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/speech-recognition/run_speech_recognition_ctc.py

@@ -261,9 +261,9 @@ 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."
             )
         },
     )
@@ -467,6 +467,7 @@ 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:
@@ -492,6 +493,7 @@ 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:

examples/speech-recognition/run_speech_recognition_seq2seq.py

@@ -106,9 +106,9 @@ 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."
             )
         },
     )
@@ -372,6 +372,7 @@ def main():
             split=data_args.train_split_name,
             cache_dir=model_args.cache_dir,
             token=model_args.token,
+            trust_remote_code=model_args.trust_remote_code,
         )
 
     if training_args.do_eval:
@@ -381,6 +382,7 @@ def main():
             split=data_args.eval_split_name,
             cache_dir=model_args.cache_dir,
             token=model_args.token,
+            trust_remote_code=model_args.trust_remote_code,
         )
 
     if data_args.audio_column_name not in next(iter(raw_datasets.values())).column_names:

examples/summarization/run_summarization.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."
             )
         },
     )
@@ -428,6 +428,7 @@ 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 = {}