add and fix examples (#12810)

docs/source/model_doc/clip.rst

@@ -60,7 +60,6 @@ encode the text and prepare the images. The following example shows how to get t
 
 .. code-block::
 
-        >>> import torch
         >>> from PIL import Image
         >>> import requests
 

src/transformers/models/clip/modeling_clip.py

@@ -699,6 +699,18 @@ class CLIPTextModel(CLIPPreTrainedModel):
         r"""
         Returns:
 
+        Examples::
+
+            >>> from transformers import CLIPTokenizer, CLIPTextModel
+
+            >>> model = CLIPTextModel.from_pretrained("openai/clip-vit-base-patch32")
+            >>> tokenizer = CLIPTokenizer.from_pretrained("openai/clip-vit-base-patch32")
+
+            >>> inputs = tokenizer(["a photo of a cat", "a photo of a dog"],  padding=True, return_tensors="pt")
+
+            >>> outputs = model(**inputs)
+            >>> last_hidden_state = outputs.last_hidden_state
+            >>> pooled_output = outputs.pooled_output # pooled (EOS token) states
         """
         return self.text_model(
             input_ids=input_ids,
@@ -791,6 +803,23 @@ class CLIPVisionModel(CLIPPreTrainedModel):
         r"""
         Returns:
 
+        Examples::
+
+            >>> from PIL import Image
+            >>> import requests
+            >>> from transformers import CLIPProcessor, CLIPVisionModel
+
+            >>> model = CLIPVisionModel.from_pretrained("openai/clip-vit-base-patch32")
+            >>> processor = CLIPProcessor.from_pretrained("openai/clip-vit-base-patch32")
+
+            >>> url = "http://images.cocodataset.org/val2017/000000039769.jpg"
+            >>> image = Image.open(requests.get(url, stream=True).raw)
+
+            >>> inputs = processor(images=image, return_tensors="pt")
+
+            >>> outputs = model(**inputs)
+            >>> last_hidden_state = outputs.last_hidden_state
+            >>> pooled_output = outputs.pooled_output # pooled CLS states
         """
         return self.vision_model(
             pixel_values=pixel_values,
@@ -847,6 +876,16 @@ class CLIPModel(CLIPPreTrainedModel):
         Returns:
             text_features (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, output_dim`): The text embeddings
             obtained by applying the projection layer to the pooled output of :class:`~transformers.CLIPTextModel`.
+
+        Examples::
+
+            >>> from transformers import CLIPTokenizer, CLIPModel
+
+            >>> model = CLIPModel.from_pretrained("openai/clip-vit-base-patch32")
+            >>> tokenizer = CLIPTokenizer.from_pretrained("openai/clip-vit-base-patch32")
+
+            >>> inputs = tokenizer(["a photo of a cat", "a photo of a dog"],  padding=True, return_tensors="pt")
+            >>> text_features = model.get_text_features(**inputs)
         """
         text_outputs = self.text_model(
             input_ids=input_ids,
@@ -874,6 +913,22 @@ class CLIPModel(CLIPPreTrainedModel):
         Returns:
             image_features (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, output_dim`): The image embeddings
             obtained by applying the projection layer to the pooled output of :class:`~transformers.CLIPVisionModel`.
+
+        Examples::
+
+            >>> from PIL import Image
+            >>> import requests
+            >>> from transformers import CLIPProcessor, CLIPModel
+
+            >>> model = CLIPModel.from_pretrained("openai/clip-vit-base-patch32")
+            >>> processor = CLIPProcessor.from_pretrained("openai/clip-vit-base-patch32")
+
+            >>> url = "http://images.cocodataset.org/val2017/000000039769.jpg"
+            >>> image = Image.open(requests.get(url, stream=True).raw)
+
+            >>> inputs = processor(images=image, return_tensors="pt")
+
+            >>> image_features = model.get_image_features(**inputs)
         """
         vision_outputs = self.vision_model(
             pixel_values=pixel_values,
@@ -903,6 +958,24 @@ class CLIPModel(CLIPPreTrainedModel):
         r"""
         Returns:
 
+        Examples::
+
+            >>> from PIL import Image
+            >>> import requests
+            >>> from transformers import CLIPProcessor, CLIPModel
+
+            >>> model = CLIPModel.from_pretrained("openai/clip-vit-base-patch32")
+            >>> processor = CLIPProcessor.from_pretrained("openai/clip-vit-base-patch32")
+
+            >>> url = "http://images.cocodataset.org/val2017/000000039769.jpg"
+            >>> image = Image.open(requests.get(url, stream=True).raw)
+
+            >>> inputs = processor(text=["a photo of a cat", "a photo of a dog"], images=image, return_tensors="pt", padding=True)
+
+            >>> outputs = model(**inputs)
+            >>> logits_per_image = outputs.logits_per_image # this is the image-text similarity score
+            >>> probs = logits_per_image.softmax(dim=1) # we can take the softmax to get the label probabilities
+
         """
         return_dict = return_dict if return_dict is not None else self.config.return_dict
         vision_outputs = self.vision_model(

src/transformers/models/clip/modeling_flax_clip.py

@@ -803,6 +803,16 @@ class FlaxCLIPPreTrainedModel(FlaxPreTrainedModel):
         Returns:
             text_features (:obj:`jax_xla.DeviceArray` of shape :obj:`(batch_size, output_dim`): The text embeddings
             obtained by applying the projection layer to the pooled output of :class:`~transformers.FlaxCLIPTextModel`.
+
+        Examples::
+
+            >>> from transformers import CLIPTokenizer, FlaxCLIPModel
+
+            >>> model = FlaxCLIPModel.from_pretrained("openai/clip-vit-base-patch32")
+            >>> tokenizer = CLIPTokenizer.from_pretrained("openai/clip-vit-base-patch32")
+
+            >>> inputs = tokenizer(["a photo of a cat", "a photo of a dog"],  padding=True, return_tensors="np")
+            >>> text_features = model.get_text_features(**inputs)
         """
         if position_ids is None:
             position_ids = jnp.broadcast_to(jnp.arange(jnp.atleast_2d(input_ids).shape[-1]), input_ids.shape)
@@ -848,6 +858,22 @@ class FlaxCLIPPreTrainedModel(FlaxPreTrainedModel):
             image_features (:obj:`jax_xla.DeviceArray` of shape :obj:`(batch_size, output_dim`): The image embeddings
             obtained by applying the projection layer to the pooled output of
             :class:`~transformers.FlaxCLIPVisionModel`
+
+        Examples::
+
+            >>> from PIL import Image
+            >>> import requests
+            >>> from transformers import CLIPProcessor, FlaxCLIPModel
+
+            >>> model = FlaxCLIPModel.from_pretrained("openai/clip-vit-base-patch32")
+            >>> processor = CLIPProcessor.from_pretrained("openai/clip-vit-base-patch32")
+
+            >>> url = "http://images.cocodataset.org/val2017/000000039769.jpg"
+            >>> image = Image.open(requests.get(url, stream=True).raw)
+
+            >>> inputs = processor(images=image, return_tensors="np")
+
+            >>> image_features = model.get_image_features(**inputs)
         """
         pixel_values = jnp.transpose(pixel_values, (0, 2, 3, 1))
 
@@ -907,6 +933,7 @@ FLAX_CLIP_TEXT_MODEL_DOCSTRING = """
     Returns:
 
     Example::
+
         >>> from transformers import CLIPTokenizer, FlaxCLIPTextModel
 
         >>> model = FlaxCLIPTextModel.from_pretrained("openai/clip-vit-base-patch32")
@@ -957,9 +984,9 @@ FLAX_CLIP_VISION_MODEL_DOCSTRING = """
     Returns:
 
     Example::
+
         >>> from PIL import Image
         >>> import requests
-
         >>> from transformers import CLIPProcessor, FlaxCLIPVisionModel
 
         >>> model = FlaxCLIPVisionModel.from_pretrained("openai/clip-vit-base-patch32")
@@ -1078,10 +1105,10 @@ FLAX_CLIP_MODEL_DOCSTRING = """
     Returns:
 
     Example::
+
         >>> import jax
         >>> from PIL import Image
         >>> import requests
-
         >>> from transformers import CLIPProcessor, FlaxCLIPModel
 
         >>> model = FlaxCLIPModel.from_pretrained("openai/clip-vit-base-patch32")