Don't push checkpoints to hub in no_trainer scripts

examples/pytorch/language-modeling/run_clm_no_trainer.py

@@ -39,6 +39,7 @@
 
 import transformers
 from accelerate import Accelerator, DistributedType
+from accelerate.utils import set_seed
 from huggingface_hub import Repository
 from transformers import (
     CONFIG_MAPPING,
@@ -50,7 +51,6 @@
     SchedulerType,
     default_data_collator,
     get_scheduler,
-    set_seed,
 )
 from transformers.utils import get_full_repo_name
 from transformers.utils.versions import require_version
@@ -258,6 +258,12 @@ def main():
             else:
                 repo_name = args.hub_model_id
             repo = Repository(args.output_dir, clone_from=repo_name)
+
+            with open(os.path.join(args.output_dir, ".gitignore"), "w+") as gitignore:
+                if "step_*" not in gitignore:
+                    gitignore.write("step_*\n")
+                if "epoch_*" not in gitignore:
+                    gitignore.write("epoch_*\n")
         elif args.output_dir is not None:
             os.makedirs(args.output_dir, exist_ok=True)
     accelerator.wait_for_everyone()
@@ -542,7 +548,6 @@ def group_texts(examples):
                     if args.output_dir is not None:
                         output_dir = os.path.join(args.output_dir, output_dir)
                     accelerator.save_state(output_dir)
-
             if completed_steps >= args.max_train_steps:
                 break
 

examples/pytorch/language-modeling/run_mlm_no_trainer.py

@@ -39,6 +39,7 @@
 
 import transformers
 from accelerate import Accelerator, DistributedType
+from accelerate.utils import set_seed
 from huggingface_hub import Repository
 from transformers import (
     CONFIG_MAPPING,
@@ -50,7 +51,6 @@
     DataCollatorForLanguageModeling,
     SchedulerType,
     get_scheduler,
-    set_seed,
 )
 from transformers.utils import get_full_repo_name
 from transformers.utils.versions import require_version
@@ -269,6 +269,12 @@ def main():
             else:
                 repo_name = args.hub_model_id
             repo = Repository(args.output_dir, clone_from=repo_name)
+
+            with open(os.path.join(args.output_dir, ".gitignore"), "w+") as gitignore:
+                if "step_*" not in gitignore:
+                    gitignore.write("step_*\n")
+                if "epoch_*" not in gitignore:
+                    gitignore.write("epoch_*\n")
         elif args.output_dir is not None:
             os.makedirs(args.output_dir, exist_ok=True)
     accelerator.wait_for_everyone()

examples/pytorch/multiple-choice/run_swag_no_trainer.py

@@ -37,6 +37,7 @@
 
 import transformers
 from accelerate import Accelerator
+from accelerate.utils import set_seed
 from huggingface_hub import Repository
 from transformers import (
     CONFIG_MAPPING,
@@ -49,7 +50,6 @@
     SchedulerType,
     default_data_collator,
     get_scheduler,
-    set_seed,
 )
 from transformers.utils import PaddingStrategy, get_full_repo_name
 
@@ -296,6 +296,12 @@ def main():
             else:
                 repo_name = args.hub_model_id
             repo = Repository(args.output_dir, clone_from=repo_name)
+
+            with open(os.path.join(args.output_dir, ".gitignore"), "w+") as gitignore:
+                if "step_*" not in gitignore:
+                    gitignore.write("step_*\n")
+                if "epoch_*" not in gitignore:
+                    gitignore.write("epoch_*\n")
         elif args.output_dir is not None:
             os.makedirs(args.output_dir, exist_ok=True)
     accelerator.wait_for_everyone()

examples/pytorch/question-answering/run_qa_beam_search_no_trainer.py

@@ -34,6 +34,7 @@
 
 import transformers
 from accelerate import Accelerator
+from accelerate.utils import set_seed
 from huggingface_hub import Repository
 from transformers import (
     AdamW,
@@ -45,7 +46,6 @@
     XLNetTokenizerFast,
     default_data_collator,
     get_scheduler,
-    set_seed,
 )
 from transformers.utils import check_min_version, get_full_repo_name
 from transformers.utils.versions import require_version
@@ -290,6 +290,12 @@ def main():
             else:
                 repo_name = args.hub_model_id
             repo = Repository(args.output_dir, clone_from=repo_name)
+
+            with open(os.path.join(args.output_dir, ".gitignore"), "w+") as gitignore:
+                if "step_*" not in gitignore:
+                    gitignore.write("step_*\n")
+                if "epoch_*" not in gitignore:
+                    gitignore.write("epoch_*\n")
         elif args.output_dir is not None:
             os.makedirs(args.output_dir, exist_ok=True)
     accelerator.wait_for_everyone()

examples/pytorch/question-answering/run_qa_no_trainer.py

@@ -35,6 +35,7 @@
 
 import transformers
 from accelerate import Accelerator
+from accelerate.utils import set_seed
 from huggingface_hub import Repository
 from transformers import (
     CONFIG_MAPPING,
@@ -48,7 +49,6 @@
     SchedulerType,
     default_data_collator,
     get_scheduler,
-    set_seed,
 )
 from transformers.utils import check_min_version, get_full_repo_name
 from transformers.utils.versions import require_version
@@ -320,6 +320,12 @@ def main():
             else:
                 repo_name = args.hub_model_id
             repo = Repository(args.output_dir, clone_from=repo_name)
+
+            with open(os.path.join(args.output_dir, ".gitignore"), "w+") as gitignore:
+                if "step_*" not in gitignore:
+                    gitignore.write("step_*\n")
+                if "epoch_*" not in gitignore:
+                    gitignore.write("epoch_*\n")
         elif args.output_dir is not None:
             os.makedirs(args.output_dir, exist_ok=True)
     accelerator.wait_for_everyone()

examples/pytorch/summarization/run_summarization_no_trainer.py

@@ -36,6 +36,7 @@
 
 import transformers
 from accelerate import Accelerator
+from accelerate.utils import set_seed
 from filelock import FileLock
 from huggingface_hub import Repository
 from transformers import (
@@ -48,7 +49,6 @@
     DataCollatorForSeq2Seq,
     SchedulerType,
     get_scheduler,
-    set_seed,
 )
 from transformers.utils import get_full_repo_name, is_offline_mode
 from transformers.utils.versions import require_version
@@ -346,6 +346,12 @@ def main():
             else:
                 repo_name = args.hub_model_id
             repo = Repository(args.output_dir, clone_from=repo_name)
+
+            with open(os.path.join(args.output_dir, ".gitignore"), "w+") as gitignore:
+                if "step_*" not in gitignore:
+                    gitignore.write("step_*\n")
+                if "epoch_*" not in gitignore:
+                    gitignore.write("epoch_*\n")
         elif args.output_dir is not None:
             os.makedirs(args.output_dir, exist_ok=True)
     accelerator.wait_for_everyone()

examples/pytorch/text-classification/run_glue_no_trainer.py

@@ -28,6 +28,7 @@
 
 import transformers
 from accelerate import Accelerator
+from accelerate.utils import set_seed
 from huggingface_hub import Repository
 from transformers import (
     AdamW,
@@ -39,7 +40,6 @@
     SchedulerType,
     default_data_collator,
     get_scheduler,
-    set_seed,
 )
 from transformers.utils import get_full_repo_name
 from transformers.utils.versions import require_version
@@ -223,6 +223,12 @@ def main():
             else:
                 repo_name = args.hub_model_id
             repo = Repository(args.output_dir, clone_from=repo_name)
+
+            with open(os.path.join(args.output_dir, ".gitignore"), "w+") as gitignore:
+                if "step_*" not in gitignore:
+                    gitignore.write("step_*\n")
+                if "epoch_*" not in gitignore:
+                    gitignore.write("epoch_*\n")
         elif args.output_dir is not None:
             os.makedirs(args.output_dir, exist_ok=True)
     accelerator.wait_for_everyone()

examples/pytorch/token-classification/run_ner_no_trainer.py

@@ -34,6 +34,7 @@
 
 import transformers
 from accelerate import Accelerator
+from accelerate.utils import set_seed
 from huggingface_hub import Repository
 from transformers import (
     CONFIG_MAPPING,
@@ -47,7 +48,6 @@
     SchedulerType,
     default_data_collator,
     get_scheduler,
-    set_seed,
 )
 from transformers.utils import get_full_repo_name
 from transformers.utils.versions import require_version
@@ -277,6 +277,12 @@ def main():
             else:
                 repo_name = args.hub_model_id
             repo = Repository(args.output_dir, clone_from=repo_name)
+
+            with open(os.path.join(args.output_dir, ".gitignore"), "w+") as gitignore:
+                if "step_*" not in gitignore:
+                    gitignore.write("step_*\n")
+                if "epoch_*" not in gitignore:
+                    gitignore.write("epoch_*\n")
         elif args.output_dir is not None:
             os.makedirs(args.output_dir, exist_ok=True)
     accelerator.wait_for_everyone()

examples/pytorch/translation/run_translation_no_trainer.py

@@ -35,6 +35,7 @@
 
 import transformers
 from accelerate import Accelerator
+from accelerate.utils import set_seed
 from huggingface_hub import Repository
 from transformers import (
     CONFIG_MAPPING,
@@ -49,7 +50,6 @@
     SchedulerType,
     default_data_collator,
     get_scheduler,
-    set_seed,
 )
 from transformers.utils import get_full_repo_name
 from transformers.utils.versions import require_version
@@ -319,6 +319,12 @@ def main():
             else:
                 repo_name = args.hub_model_id
             repo = Repository(args.output_dir, clone_from=repo_name)
+
+            with open(os.path.join(args.output_dir, ".gitignore"), "w+") as gitignore:
+                if "step_*" not in gitignore:
+                    gitignore.write("step_*\n")
+                if "epoch_*" not in gitignore:
+                    gitignore.write("epoch_*\n")
         elif args.output_dir is not None:
             os.makedirs(args.output_dir, exist_ok=True)
     accelerator.wait_for_everyone()