Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

fix for custom pipeline configuration #29004

Merged
merged 12 commits into from
May 20, 2024
Merged

fix for custom pipeline configuration #29004

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
12 commits
Select commit Hold shift + click to select a range
3a74092
fix for custom pipeline configuration
not-lain Feb 13, 2024
c03357e
fix for custom pipelines
not-lain Feb 15, 2024
2c3ae80
remove extra exception
not-lain Feb 15, 2024
99ad0e2
Merge branch 'huggingface:main' into fix-config-for-custom-pipelines
not-lain Feb 21, 2024
68e0544
added test for custom pipelines extra tag
not-lain Mar 6, 2024
1283454
Merge branch 'huggingface:main' into fix-config-for-custom-pipelines
not-lain Mar 6, 2024
cd49d7d
format with ruff
not-lain Mar 6, 2024
78e4690
Merge branch 'fix-config-for-custom-pipelines' of https://github.com/…
not-lain Mar 6, 2024
2a30970
limit extra tag for first time only
not-lain Mar 6, 2024
26866d0
format with ruff
not-lain Mar 6, 2024
323b50d
improve tests for custom pipelines
not-lain Mar 29, 2024
091630d
Merge branch 'huggingface:main' into fix-config-for-custom-pipelines
not-lain Mar 29, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 5 additions & 0 deletions src/transformers/configuration_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,6 +31,7 @@
CONFIG_NAME,
PushToHubMixin,
add_model_info_to_auto_map,
add_model_info_to_custom_pipelines,
cached_file,
copy_func,
download_url,
Expand Down Expand Up @@ -729,6 +730,10 @@ def _get_config_dict(
config_dict["auto_map"] = add_model_info_to_auto_map(
config_dict["auto_map"], pretrained_model_name_or_path
)
if "custom_pipelines" in config_dict and not is_local:
config_dict["custom_pipelines"] = add_model_info_to_custom_pipelines(
config_dict["custom_pipelines"], pretrained_model_name_or_path
)
return config_dict, kwargs

@classmethod
Expand Down
14 changes: 10 additions & 4 deletions src/transformers/feature_extraction_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,6 +31,7 @@
PushToHubMixin,
TensorType,
add_model_info_to_auto_map,
add_model_info_to_custom_pipelines,
cached_file,
copy_func,
download_url,
Expand Down Expand Up @@ -539,10 +540,15 @@ def get_feature_extractor_dict(
f"loading configuration file {feature_extractor_file} from cache at {resolved_feature_extractor_file}"
)

if "auto_map" in feature_extractor_dict and not is_local:
feature_extractor_dict["auto_map"] = add_model_info_to_auto_map(
feature_extractor_dict["auto_map"], pretrained_model_name_or_path
)
if not is_local:
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

just curious as to why we only do it if not local whereas before we did it for local and not local? 🤗

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

what does is_local mean ?

if i call a model using this

path = "./folder1/folder2/folder3"
from transformers import AutoModelForImageClassification
model = AutoModelForImageClassification.from_pretrained(path, trust_remote_code=True)

if the path exists in my pc then we load from there

what if the is_local is false ?

then in this case we call the model from the hub and add the tag

is the tag added when we are only calling a custom model from huggingface ?

yes, there is no need to add the tag if we are calling the model from the local pc

hope this answers your questions 🤗

if "auto_map" in feature_extractor_dict:
feature_extractor_dict["auto_map"] = add_model_info_to_auto_map(
feature_extractor_dict["auto_map"], pretrained_model_name_or_path
)
if "custom_pipelines" in feature_extractor_dict:
feature_extractor_dict["custom_pipelines"] = add_model_info_to_custom_pipelines(
feature_extractor_dict["custom_pipelines"], pretrained_model_name_or_path
)

return feature_extractor_dict, kwargs

Expand Down
15 changes: 10 additions & 5 deletions src/transformers/image_processing_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,6 +31,7 @@
IMAGE_PROCESSOR_NAME,
PushToHubMixin,
add_model_info_to_auto_map,
add_model_info_to_custom_pipelines,
cached_file,
copy_func,
download_url,
Expand Down Expand Up @@ -375,11 +376,15 @@ def get_image_processor_dict(
f"loading configuration file {image_processor_file} from cache at {resolved_image_processor_file}"
)

if "auto_map" in image_processor_dict and not is_local:
image_processor_dict["auto_map"] = add_model_info_to_auto_map(
image_processor_dict["auto_map"], pretrained_model_name_or_path
)

if not is_local:
if "auto_map" in image_processor_dict:
image_processor_dict["auto_map"] = add_model_info_to_auto_map(
image_processor_dict["auto_map"], pretrained_model_name_or_path
)
if "custom_pipelines" in image_processor_dict:
image_processor_dict["custom_pipelines"] = add_model_info_to_custom_pipelines(
image_processor_dict["custom_pipelines"], pretrained_model_name_or_path
)
return image_processor_dict, kwargs

@classmethod
Expand Down
14 changes: 10 additions & 4 deletions src/transformers/processing_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -30,6 +30,7 @@
PROCESSOR_NAME,
PushToHubMixin,
add_model_info_to_auto_map,
add_model_info_to_custom_pipelines,
cached_file,
copy_func,
direct_transformers_import,
Expand Down Expand Up @@ -355,10 +356,15 @@ def get_processor_dict(
else:
logger.info(f"loading configuration file {processor_file} from cache at {resolved_processor_file}")

if "auto_map" in processor_dict and not is_local:
processor_dict["auto_map"] = add_model_info_to_auto_map(
processor_dict["auto_map"], pretrained_model_name_or_path
)
if not is_local:
if "auto_map" in processor_dict:
processor_dict["auto_map"] = add_model_info_to_auto_map(
processor_dict["auto_map"], pretrained_model_name_or_path
)
if "custom_pipelines" in processor_dict:
processor_dict["custom_pipelines"] = add_model_info_to_custom_pipelines(
processor_dict["custom_pipelines"], pretrained_model_name_or_path
)

return processor_dict, kwargs

Expand Down
20 changes: 13 additions & 7 deletions src/transformers/tokenization_utils_base.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -42,6 +42,7 @@
TensorType,
add_end_docstrings,
add_model_info_to_auto_map,
add_model_info_to_custom_pipelines,
cached_file,
copy_func,
download_url,
Expand Down Expand Up @@ -2152,13 +2153,18 @@ def _from_pretrained(
config_tokenizer_class = None
init_kwargs = init_configuration

if "auto_map" in init_kwargs and not _is_local:
# For backward compatibility with odl format.
if isinstance(init_kwargs["auto_map"], (tuple, list)):
init_kwargs["auto_map"] = {"AutoTokenizer": init_kwargs["auto_map"]}
init_kwargs["auto_map"] = add_model_info_to_auto_map(
init_kwargs["auto_map"], pretrained_model_name_or_path
)
if not _is_local:
if "auto_map" in init_kwargs:
# For backward compatibility with odl format.
if isinstance(init_kwargs["auto_map"], (tuple, list)):
init_kwargs["auto_map"] = {"AutoTokenizer": init_kwargs["auto_map"]}
init_kwargs["auto_map"] = add_model_info_to_auto_map(
init_kwargs["auto_map"], pretrained_model_name_or_path
)
if "custom_pipelines" in init_kwargs:
init_kwargs["custom_pipelines"] = add_model_info_to_custom_pipelines(
init_kwargs["custom_pipelines"], pretrained_model_name_or_path
)

if config_tokenizer_class is None:
# Matt: This entire block is only used to decide if the tokenizer class matches the class in the repo.
Expand Down
1 change: 1 addition & 0 deletions src/transformers/utils/__init__.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -37,6 +37,7 @@
PaddingStrategy,
TensorType,
add_model_info_to_auto_map,
add_model_info_to_custom_pipelines,
cached_property,
can_return_loss,
expand_dims,
Expand Down
13 changes: 13 additions & 0 deletions src/transformers/utils/generic.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -717,6 +717,19 @@ def add_model_info_to_auto_map(auto_map, repo_id):
return auto_map


def add_model_info_to_custom_pipelines(custom_pipeline, repo_id):
"""
Adds the information of the repo_id to a given custom pipeline.
"""
# {custom_pipelines : {task: {"impl": "path.to.task"},...} }
for task in custom_pipeline.keys():
if "impl" in custom_pipeline[task]:
module = custom_pipeline[task]["impl"]
if "--" not in module:
custom_pipeline[task]["impl"] = f"{repo_id}--{module}"
return custom_pipeline


def infer_framework(model_class):
"""
Infers the framework of a given model without using isinstance(), because we cannot guarantee that the relevant
Expand Down
18 changes: 18 additions & 0 deletions tests/pipelines/test_pipelines_common.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -882,6 +882,24 @@ def test_push_to_hub_dynamic_pipeline(self):
# Can't make an isinstance check because the new_classifier is from the PairClassificationPipeline class of a
# dynamic module
self.assertEqual(new_classifier.__class__.__name__, "PairClassificationPipeline")
# check for tag exitence, tag needs to be added when we are calling a custom pipeline from the hub
# useful for cases such as finetuning
self.assertDictEqual(
new_classifier.model.config.custom_pipelines,
{
"pair-classification": {
"impl": f"{USER}/test-dynamic-pipeline--custom_pipeline.PairClassificationPipeline",
"pt": ("AutoModelForSequenceClassification",),
"tf": (),
}
},
)
# test if the pipeline still works after the model is finetuned
# (we are actually testing if the pipeline still works from the final repo)
# this is where the user/repo--module.class is used for
new_classifier.model.push_to_hub(repo_name=f"{USER}/test-pipeline-for-a-finetuned-model", token=self._token)
del new_classifier # free up memory
new_classifier = pipeline(model=f"{USER}/test-pipeline-for-a-finetuned-model", trust_remote_code=True)

results = classifier("I hate you", second_text="I love you")
new_results = new_classifier("I hate you", second_text="I love you")
Expand Down