docs: TransformersImageToText- inform about supported models, better exception handling (#4310)

* better docs, exception handling and tests

* Update lg

* fix little error

---------

Co-authored-by: agnieszka-m <amarzec13@gmail.com>

haystack/nodes/image_to_text/transformers.py

@@ -19,12 +19,21 @@ logger = logging.getLogger(__name__)
 # see https://github.com/huggingface/transformers/issues/21110
 SUPPORTED_MODELS_CLASSES = ["VisionEncoderDecoderModel"]
 
+UNSUPPORTED_MODEL_MESSAGE = (
+    f"The supported classes are: {SUPPORTED_MODELS_CLASSES}. \n"
+    f"To find the supported models: \n"
+    f"1. Visit [image-to-text models on Hugging Face](https://huggingface.co/models?pipeline_tag=image-to-text). \n"
+    f"2. Open a model you want to check.  \n"
+    f'3. On the model page, go to the "Files and Versions" tab \n'
+    f"4. Open the `config.json` file, and make sure the `architectures` field contains one of the supported classes: {SUPPORTED_MODELS_CLASSES}."
+)
+
 
 class TransformersImageToText(BaseImageToText):
     """
     A transformer-based model to generate captions for images using the Hugging Face's transformers framework.
 
-    For an up-to-date list of available models, see [Hugging Face image to text models](https://huggingface.co/models?pipeline_tag=image-to-text)`__
+    Currently, this node supports `VisionEncoderDecoderModel` models.
 
     **Example**
 
@@ -65,11 +74,15 @@ class TransformersImageToText(BaseImageToText):
         devices: Optional[List[Union[str, torch.device]]] = None,
     ):
         """
-        Load an image-to-text model from transformers.
+        Load a `VisionEncoderDecoderModel` model from transformers.
-        For an up-to-date list of available models, see [Hugging Face image-to-text models](https://huggingface.co/models?pipeline_tag=image-to-text).
 
         :param model_name_or_path: Directory of a saved model or the name of a public model.
-                                   For a full list of models, see [Hugging Face image-to-text models](https://huggingface.co/models?pipeline_tag=image-to-text).
+                                   Currently, only `VisionEncoderDecoderModel` models are supported.
+                                   To find these models:
+                                   1. Visit [Hugging Face image to text models](https://huggingface.co/models?pipeline_tag=image-to-text).`
+                                   2. Open the model you want to check.
+                                   3. On the model page, go to the "Files and Versions" tab.
+                                   4. Open the `config.json` file and make sure the `architectures` field contains `VisionEncoderDecoderModel`.
         :param model_version: The version of the model to use from the Hugging Face model hub. This can be the tag name, branch name, or commit hash.
         :param generation_kwargs: Dictionary containing arguments for the `generate()` method of the Hugging Face model.
                                 See [generate()](https://huggingface.co/docs/transformers/en/main_classes/text_generation#transformers.GenerationMixin.generate) in Hugging Face documentation.
@@ -95,6 +108,7 @@ class TransformersImageToText(BaseImageToText):
                 self.devices[0],
             )
 
+        try:
             self.model = pipeline(
                 task="image-to-text",
                 model=model_name_or_path,
@@ -102,13 +116,17 @@ class TransformersImageToText(BaseImageToText):
                 device=self.devices[0],
                 use_auth_token=use_auth_token,
             )
+        except KeyError as err:
+            raise ValueError(
+                f"The model '{model_name_or_path}' is not supported for ImageToText. " f"{UNSUPPORTED_MODEL_MESSAGE}"
+            ) from err
 
+        # for some unsupported models, initializing the HF pipeline doesn't raise errors but does not work
         model_class_name = self.model.model.__class__.__name__
         if model_class_name not in SUPPORTED_MODELS_CLASSES:
             raise ValueError(
-                f"The model of class '{model_class_name}' is not supported for ImageToText."
+                f"The model '{model_name_or_path}' (class '{model_class_name}') is not supported for ImageToText. "
-                f"The supported classes are: {SUPPORTED_MODELS_CLASSES}."
+                f"{UNSUPPORTED_MODEL_MESSAGE}"
-                f"You can find the availaible models here: https://huggingface.co/models?pipeline_tag=image-to-text."
             )
 
         self.generation_kwargs = generation_kwargs

test/nodes/test_image_to_text.py

@@ -82,8 +82,17 @@ def test_image_to_text_not_image_document(image_to_text):
 
 
 @pytest.mark.integration
-def test_image_to_text_unsupported_model():
+def test_image_to_text_unsupported_model_after_loading():
     with pytest.raises(
-        ValueError, match="The model of class 'BertForQuestionAnswering' is not supported for ImageToText"
+        ValueError,
+        match="The model 'deepset/minilm-uncased-squad2' \(class 'BertForQuestionAnswering'\) is not supported for ImageToText",
     ):
         _ = TransformersImageToText(model_name_or_path="deepset/minilm-uncased-squad2")
+
+
+@pytest.mark.integration
+def test_image_to_text_unsupported_model_before_loading():
+    with pytest.raises(
+        ValueError, match="The model 'Salesforce/blip-image-captioning-base' is not supported for ImageToText"
+    ):
+        _ = TransformersImageToText(model_name_or_path="Salesforce/blip-image-captioning-base")