py: Add description to base model authorship metadata

Author: mofosyne

examples/convert_legacy_llama.py

@@ -840,6 +840,8 @@ def add_meta_model(self, params: Params, metadata: gguf.Metadata | None) -> None
                         self.gguf.add_base_model_version(key, base_model_entry["version"])
                     if "organization" in base_model_entry:
                         self.gguf.add_base_model_organization(key, base_model_entry["organization"])
+                    if "description" in base_model_entry:
+                        self.gguf.add_base_model_description(key, base_model_entry["description"])
                     if "url" in base_model_entry:
                         self.gguf.add_base_model_url(key, base_model_entry["url"])
                     if "doi" in base_model_entry:

gguf-py/gguf/constants.py

@@ -64,6 +64,7 @@ class General:
         BASE_MODEL_AUTHOR          = "general.base_model.{id}.author"
         BASE_MODEL_VERSION         = "general.base_model.{id}.version"
         BASE_MODEL_ORGANIZATION    = "general.base_model.{id}.organization"
+        BASE_MODEL_DESCRIPTION     = "general.base_model.{id}.description"
         BASE_MODEL_URL             = "general.base_model.{id}.url" # Model Website/Paper
         BASE_MODEL_DOI             = "general.base_model.{id}.doi"
         BASE_MODEL_UUID            = "general.base_model.{id}.uuid"
@@ -74,6 +75,18 @@ class General:
         LANGUAGES                  = "general.languages"
         DATASETS                   = "general.datasets"
 
+        # Dataset Source
+        DATASET_COUNT           = "general.dataset.count"
+        DATASET_NAME            = "general.dataset.{id}.name"
+        DATASET_AUTHOR          = "general.dataset.{id}.author"
+        DATASET_VERSION         = "general.dataset.{id}.version"
+        DATASET_ORGANIZATION    = "general.dataset.{id}.organization"
+        DATASET_DESCRIPTION     = "general.dataset.{id}.description"
+        DATASET_URL             = "general.dataset.{id}.url" # Model Website/Paper
+        DATASET_DOI             = "general.dataset.{id}.doi"
+        DATASET_UUID            = "general.dataset.{id}.uuid"
+        DATASET_REPO_URL        = "general.dataset.{id}.repo_url" # Model Source Repository (git/svn/etc...)
+
     class LLM:
         VOCAB_SIZE                        = "{arch}.vocab_size"
         CONTEXT_LENGTH                    = "{arch}.context_length"

gguf-py/gguf/gguf_writer.py

@@ -568,6 +568,9 @@ def add_base_model_version(self, source_id: int, version: str) -> None:
     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)
 

gguf-py/gguf/metadata.py

@@ -493,6 +493,8 @@ def set_gguf_meta_model(self, gguf_writer: gguf.GGUFWriter):
                     gguf_writer.add_base_model_version(key, base_model_entry["version"])
                 if "organization" in base_model_entry:
                     gguf_writer.add_base_model_organization(key, base_model_entry["organization"])
+                if "description" in base_model_entry:
+                    gguf_writer.add_base_model_description(key, base_model_entry["description"])
                 if "url" in base_model_entry:
                     gguf_writer.add_base_model_url(key, base_model_entry["url"])
                 if "doi" in base_model_entry: