offline mode for firewalled envs (#10407)

* offline mode start

* add specific values

* fix fallback

* add test

* better values check and range

* test that actually works

* document the offline mode

* Apply suggestions from code review

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* more strict check

* cleaner test

* pt-only test

* style

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

src/transformers/configuration_utils.py

@@ -22,7 +22,7 @@ import os
 from typing import Any, Dict, Tuple, Union
 
 from . import __version__
-from .file_utils import CONFIG_NAME, cached_path, hf_bucket_url, is_remote_url
+from .file_utils import CONFIG_NAME, cached_path, hf_bucket_url, is_offline_mode, is_remote_url
 from .utils import logging
 
 
@@ -412,6 +412,10 @@ class PretrainedConfig(object):
         local_files_only = kwargs.pop("local_files_only", False)
         revision = kwargs.pop("revision", None)
 
+        if is_offline_mode() and not local_files_only:
+            logger.info("Offline mode: forcing local_files_only=True")
+            local_files_only = True
+
         pretrained_model_name_or_path = str(pretrained_model_name_or_path)
         if os.path.isdir(pretrained_model_name_or_path):
             config_file = os.path.join(pretrained_model_name_or_path, CONFIG_NAME)

src/transformers/file_utils.py

@@ -234,6 +234,13 @@ PRESET_MIRROR_DICT = {
 }
 
 
+_is_offline_mode = True if os.environ.get("TRANSFORMERS_OFFLINE", "0").upper() in ENV_VARS_TRUE_VALUES else False
+
+
+def is_offline_mode():
+    return _is_offline_mode
+
+
 def is_torch_available():
     return _torch_available
 

src/transformers/modeling_utils.py

@@ -36,6 +36,7 @@ from .file_utils import (
     ModelOutput,
     cached_path,
     hf_bucket_url,
+    is_offline_mode,
     is_remote_url,
     replace_return_docstrings,
 )
@@ -964,6 +965,10 @@ class PreTrainedModel(nn.Module, ModuleUtilsMixin, GenerationMixin):
         revision = kwargs.pop("revision", None)
         mirror = kwargs.pop("mirror", None)
 
+        if is_offline_mode() and not local_files_only:
+            logger.info("Offline mode: forcing local_files_only=True")
+            local_files_only = True
+
         # Load config if we don't provide a configuration
         if not isinstance(config, PretrainedConfig):
             config_path = config if config is not None else pretrained_model_name_or_path

src/transformers/tokenization_utils_base.py

@@ -44,6 +44,7 @@ from .file_utils import (
     cached_path,
     hf_bucket_url,
     is_flax_available,
+    is_offline_mode,
     is_remote_url,
     is_tf_available,
     is_tokenizers_available,
@@ -1597,6 +1598,10 @@ class PreTrainedTokenizerBase(SpecialTokensMixin):
         revision = kwargs.pop("revision", None)
         subfolder = kwargs.pop("subfolder", None)
 
+        if is_offline_mode() and not local_files_only:
+            logger.info("Offline mode: forcing local_files_only=True")
+            local_files_only = True
+
         s3_models = list(cls.max_model_input_sizes.keys())
         pretrained_model_name_or_path = str(pretrained_model_name_or_path)
         vocab_files = {}