vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

metadata: Detailed Dataset Authorship Metadata (#8875)

Browse source 
Converter script can now read these two fields as a detailed base model and dataset source.
This was done so that it will be easier for Hugging Face to integrate detailed metadata as needed.

 -  base_model_sources (List[dict], optional)
 -  dataset_sources (List[dict], optional)

Dataset now represented as:

 - general.dataset.count
 - general.dataset.{id}.name
 - general.dataset.{id}.author
 - general.dataset.{id}.version
 - general.dataset.{id}.organization
 - general.dataset.{id}.description
 - general.dataset.{id}.url
 - general.dataset.{id}.doi
 - general.dataset.{id}.uuid
 - general.dataset.{id}.repo_url

This also adds to base model these metadata:

 - general.base_model.{id}.description
This commit is contained in:
Brian 2024-11-13 21:10:38 +11:00 • committed by GitHub
parent 2e82ffa4af
commit a0ec17b32e
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
5 changed files with 233 additions and 25 deletions
Download patch file Download diff file Expand all files Collapse all files

36
gguf-py/gguf/gguf_writer.py
View file

@ -568,6 +568,9 @@ class GGUFWriter:
def add_base_model_organization(self, source_id: int, organization: str) -> None:
self.add_string(Keys.General.BASE_MODEL_ORGANIZATION.format(id=source_id), organization)
def add_base_model_description(self, source_id: int, description: str) -> None:
self.add_string(Keys.General.BASE_MODEL_DESCRIPTION.format(id=source_id), description)
def add_base_model_url(self, source_id: int, url: str) -> None:
self.add_string(Keys.General.BASE_MODEL_URL.format(id=source_id), url)
@ -580,15 +583,42 @@ class GGUFWriter:
def add_base_model_repo_url(self, source_id: int, repo_url: str) -> None:
self.add_string(Keys.General.BASE_MODEL_REPO_URL.format(id=source_id), repo_url)
def add_dataset_count(self, source_count: int) -> None:
self.add_uint32(Keys.General.DATASET_COUNT, source_count)
def add_dataset_name(self, source_id: int, name: str) -> None:
self.add_string(Keys.General.DATASET_NAME.format(id=source_id), name)
def add_dataset_author(self, source_id: int, author: str) -> None:
self.add_string(Keys.General.DATASET_AUTHOR.format(id=source_id), author)
def add_dataset_version(self, source_id: int, version: str) -> None:
self.add_string(Keys.General.DATASET_VERSION.format(id=source_id), version)
def add_dataset_organization(self, source_id: int, organization: str) -> None:
self.add_string(Keys.General.DATASET_ORGANIZATION.format(id=source_id), organization)
def add_dataset_description(self, source_id: int, description: str) -> None:
self.add_string(Keys.General.DATASET_DESCRIPTION.format(id=source_id), description)
def add_dataset_url(self, source_id: int, url: str) -> None:
self.add_string(Keys.General.DATASET_URL.format(id=source_id), url)
def add_dataset_doi(self, source_id: int, doi: str) -> None:
self.add_string(Keys.General.DATASET_DOI.format(id=source_id), doi)
def add_dataset_uuid(self, source_id: int, uuid: str) -> None:
self.add_string(Keys.General.DATASET_UUID.format(id=source_id), uuid)
def add_dataset_repo_url(self, source_id: int, repo_url: str) -> None:
self.add_string(Keys.General.DATASET_REPO_URL.format(id=source_id), repo_url)
def add_tags(self, tags: Sequence[str]) -> None:
self.add_array(Keys.General.TAGS, tags)
def add_languages(self, languages: Sequence[str]) -> None:
self.add_array(Keys.General.LANGUAGES, languages)
def add_datasets(self, datasets: Sequence[str]) -> None:
self.add_array(Keys.General.DATASETS, datasets)
def add_tensor_data_layout(self, layout: str) -> None:
self.add_string(Keys.LLM.TENSOR_DATA_LAYOUT.format(arch=self.arch), layout)