rename model

Signed-off-by: Michele Dolfi <dol@zurich.ibm.com>

docling/datamodel/pipeline_options.py

@@ -184,7 +184,7 @@ class OcrMacOptions(OcrOptions):
     )
 
 
-class PicDescBaseOptions(BaseModel):
+class PictureDescriptionBaseOptions(BaseModel):
     kind: str
     batch_size: int = 8
     scale: float = 2
@@ -194,7 +194,7 @@ class PicDescBaseOptions(BaseModel):
     )
 
 
-class PicDescApiOptions(PicDescBaseOptions):
+class PictureDescriptionApiOptions(PictureDescriptionBaseOptions):
     kind: Literal["api"] = "api"
 
     url: AnyUrl = AnyUrl("http://localhost:8000/v1/chat/completions")
@@ -206,7 +206,7 @@ class PicDescApiOptions(PicDescBaseOptions):
     provenance: str = ""
 
 
-class PicDescVlmOptions(PicDescBaseOptions):
+class PictureDescriptionVlmOptions(PictureDescriptionBaseOptions):
     kind: Literal["vlm"] = "vlm"
 
     repo_id: str
@@ -215,18 +215,11 @@ class PicDescVlmOptions(PicDescBaseOptions):
     generation_config: Dict[str, Any] = dict(max_new_tokens=200, do_sample=False)
 
 
-# class PicDescSmolVlmOptions(PicDescVlmOptions):
+smolvlm_picture_description = PictureDescriptionVlmOptions(
-#     repo_id: str = "HuggingFaceTB/SmolVLM-256M-Instruct"
+    repo_id="HuggingFaceTB/SmolVLM-256M-Instruct"
-
+)
-
+# phi_pic_desc = PictureDescriptionVlmOptions(repo_id="microsoft/Phi-3-vision-128k-instruct")
-# class PicDescGraniteOptions(PicDescVlmOptions):
+granite_picture_description = PictureDescriptionVlmOptions(
-#     repo_id: str = "ibm-granite/granite-vision-3.1-2b-preview"
-#     prompt: str = "What is shown in this image?"
-
-
-smolvlm_pic_desc = PicDescVlmOptions(repo_id="HuggingFaceTB/SmolVLM-256M-Instruct")
-# phi_pic_desc = PicDescVlmOptions(repo_id="microsoft/Phi-3-vision-128k-instruct")
-granite_pic_desc = PicDescVlmOptions(
     repo_id="ibm-granite/granite-vision-3.1-2b-preview",
     prompt="What is shown in this image?",
 )
@@ -282,8 +275,9 @@ class PdfPipelineOptions(PipelineOptions):
         RapidOcrOptions,
     ] = Field(EasyOcrOptions(), discriminator="kind")
     picture_description_options: Annotated[
-        Union[PicDescApiOptions, PicDescVlmOptions], Field(discriminator="kind")
+        Union[PictureDescriptionApiOptions, PictureDescriptionVlmOptions],
-    ] = smolvlm_pic_desc
+        Field(discriminator="kind"),
+    ] = smolvlm_picture_description
 
     images_scale: float = 1.0
     generate_page_images: bool = False

docling/models/pic_description_api_model.py

@@ -11,7 +11,7 @@ from docling_core.types.doc.document import (  # TODO: move import to docling_co
 from PIL import Image
 from pydantic import BaseModel, ConfigDict
 
-from docling.datamodel.pipeline_options import PicDescApiOptions
+from docling.datamodel.pipeline_options import PictureDescriptionApiOptions
 from docling.models.pic_description_base_model import PictureDescriptionBaseModel
 
 _log = logging.getLogger(__name__)
@@ -49,9 +49,9 @@ class ApiResponse(BaseModel):
 class PictureDescriptionApiModel(PictureDescriptionBaseModel):
     # elements_batch_size = 4
 
-    def __init__(self, enabled: bool, options: PicDescApiOptions):
+    def __init__(self, enabled: bool, options: PictureDescriptionApiOptions):
         super().__init__(enabled=enabled, options=options)
-        self.options: PicDescApiOptions
+        self.options: PictureDescriptionApiOptions
 
         if self.enabled:
             if options.url.host != "localhost":

docling/models/pic_description_base_model.py

@@ -13,7 +13,7 @@ from docling_core.types.doc.document import (  # TODO: move import to docling_co
 )
 from PIL import Image
 
-from docling.datamodel.pipeline_options import PicDescBaseOptions
+from docling.datamodel.pipeline_options import PictureDescriptionBaseOptions
 from docling.models.base_model import (
     BaseItemAndImageEnrichmentModel,
     ItemAndImageEnrichmentElement,
@@ -26,7 +26,7 @@ class PictureDescriptionBaseModel(BaseItemAndImageEnrichmentModel):
     def __init__(
         self,
         enabled: bool,
-        options: PicDescBaseOptions,
+        options: PictureDescriptionBaseOptions,
     ):
         self.enabled = enabled
         self.options = options

docling/models/pic_description_vlm_model.py

@@ -4,7 +4,10 @@ from typing import Iterable, List, Optional, Union
 
 from PIL import Image
 
-from docling.datamodel.pipeline_options import AcceleratorOptions, PicDescVlmOptions
+from docling.datamodel.pipeline_options import (
+    AcceleratorOptions,
+    PictureDescriptionVlmOptions,
+)
 from docling.models.pic_description_base_model import PictureDescriptionBaseModel
 from docling.utils.accelerator_utils import decide_device
 
@@ -15,11 +18,11 @@ class PictureDescriptionVlmModel(PictureDescriptionBaseModel):
         self,
         enabled: bool,
         artifacts_path: Optional[Union[Path, str]],
-        options: PicDescVlmOptions,
+        options: PictureDescriptionVlmOptions,
         accelerator_options: AcceleratorOptions,
     ):
         super().__init__(enabled=enabled, options=options)
-        self.options: PicDescVlmOptions
+        self.options: PictureDescriptionVlmOptions
 
         if self.enabled:
 

docling/pipeline/standard_pdf_pipeline.py

@@ -13,8 +13,8 @@ from docling.datamodel.pipeline_options import (
     EasyOcrOptions,
     OcrMacOptions,
     PdfPipelineOptions,
-    PicDescApiOptions,
+    PictureDescriptionApiOptions,
-    PicDescVlmOptions,
+    PictureDescriptionVlmOptions,
     RapidOcrOptions,
     TesseractCliOcrOptions,
     TesseractOcrOptions,
@@ -191,14 +191,16 @@ class StandardPdfPipeline(PaginatedPipeline):
         self, artifacts_path: Optional[Path] = None
     ) -> Optional[PictureDescriptionBaseModel]:
         if isinstance(
-            self.pipeline_options.picture_description_options, PicDescApiOptions
+            self.pipeline_options.picture_description_options,
+            PictureDescriptionApiOptions,
         ):
             return PictureDescriptionApiModel(
                 enabled=self.pipeline_options.do_picture_description,
                 options=self.pipeline_options.picture_description_options,
             )
         elif isinstance(
-            self.pipeline_options.picture_description_options, PicDescVlmOptions
+            self.pipeline_options.picture_description_options,
+            PictureDescriptionVlmOptions,
         ):
             return PictureDescriptionVlmModel(
                 enabled=self.pipeline_options.do_picture_description,

docs/examples/pictures_description.py

@@ -4,10 +4,10 @@ from pathlib import Path
 from docling_core.types.doc import PictureItem
 
 from docling.datamodel.base_models import InputFormat
-from docling.datamodel.pipeline_options import (  # PicDescSmolVlmOptions, PicDescGraniteOptions
+from docling.datamodel.pipeline_options import (
     PdfPipelineOptions,
-    granite_pic_desc,
+    granite_picture_description,
-    smolvlm_pic_desc,
+    smolvlm_picture_description,
 )
 from docling.document_converter import DocumentConverter, PdfFormatOption
 
@@ -19,7 +19,7 @@ def main():
 
     pipeline_options = PdfPipelineOptions()
     pipeline_options.do_picture_description = True
-    pipeline_options.picture_description_options = smolvlm_pic_desc
+    pipeline_options.picture_description_options = smolvlm_picture_description
     # pipeline_options.picture_description_options = granite_pic_desc
 
     pipeline_options.picture_description_options.prompt = (

docs/examples/pictures_description_api.py

@@ -4,7 +4,10 @@ from pathlib import Path
 from docling_core.types.doc import PictureItem
 
 from docling.datamodel.base_models import InputFormat
-from docling.datamodel.pipeline_options import PdfPipelineOptions, PicDescApiOptions
+from docling.datamodel.pipeline_options import (
+    PdfPipelineOptions,
+    PictureDescriptionApiOptions,
+)
 from docling.document_converter import DocumentConverter, PdfFormatOption
 
 
@@ -19,7 +22,7 @@ def main():
 
     pipeline_options = PdfPipelineOptions()
     pipeline_options.do_picture_description = True
-    pipeline_options.picture_description_options = PicDescApiOptions(
+    pipeline_options.picture_description_options = PictureDescriptionApiOptions(
         url="http://localhost:8000/v1/chat/completions",
         params=dict(
             model="HuggingFaceTB/SmolVLM-256M-Instruct",