added embeddings for almost all interface types

gradio/inputs.py

@@ -10,7 +10,7 @@ import os
 import time
 import warnings
 from gradio.component import Component
-
+from gradio.embeddings import embed_text
 import base64
 import numpy as np
 import PIL
@@ -72,6 +72,16 @@ class InputComponent(Component):
         Returns:
         (List[Any]): Arrangement of interpretation scores for interfaces to render.
         '''
+        pass
+
+    def embed(self, x):
+        """
+        Return a default embedding for the *preprocessed* input to the interface. Used to compute similar inputs.
+        x (Any): Input to interface
+        Returns:
+        (List[Float]): An embedding vector as a list or numpy array of floats
+        """
+        pass
 
 class Textbox(InputComponent):
     """
@@ -170,6 +180,17 @@ class Textbox(InputComponent):
             result.append((self.interpretation_separator, 0))
         return result
 
+    def embed(self, x):
+        """
+        Embeds an arbitrary text based on word frequency
+        """
+        if self.type == "str":
+            return embed_text(x)
+        elif self.type == "number":
+            return [float(x)]
+        else:
+            raise ValueError("Unknown type: " + str(self.type) + ". Please choose from: 'str', 'number'.")
+
 
 class Number(InputComponent):
     """
@@ -238,6 +259,9 @@ class Number(InputComponent):
         interpretation.insert(int(len(interpretation) / 2), [x, None])
         return interpretation
 
+    def embed(self, x):
+        return [float(x)]
+
 
 class Slider(InputComponent):
     """
@@ -306,6 +330,10 @@ class Slider(InputComponent):
         """
         return scores
 
+    def embed(self, x):
+        return [float(x)]
+
+
 
 class Checkbox(InputComponent):
     """
@@ -353,6 +381,10 @@ class Checkbox(InputComponent):
         else:
             return None, scores[0]
 
+    def embed(self, x):
+        return [float(x)]
+
+
 
 class CheckboxGroup(InputComponent):
     """
@@ -417,6 +449,15 @@ class CheckboxGroup(InputComponent):
             final_scores.append(score_set)
         return final_scores
 
+    def embed(self, x):
+        if self.type == "value":
+            return [choice in x for choice in self.choices]
+        elif self.type == "index":
+            return [index in x for index in range(len(choices))]
+        else:
+            raise ValueError("Unknown type: " + str(self.type) + ". Please choose from: 'value', 'index'.")
+
+
 
 class Radio(InputComponent):
     """
@@ -469,6 +510,14 @@ class Radio(InputComponent):
         scores.insert(self.choices.index(x), None)
         return scores
 
+    def embed(self, x):
+        if self.type == "value":
+            return [choice==x for choice in self.choices]
+        elif self.type == "index":
+            return [index==x for index in range(len(choices))]
+        else:
+            raise ValueError("Unknown type: " + str(self.type) + ". Please choose from: 'value', 'index'.")
+
 
 class Dropdown(InputComponent):
     """
@@ -521,6 +570,14 @@ class Dropdown(InputComponent):
         scores.insert(self.choices.index(x), None)
         return scores
 
+    def embed(self, x):
+        if self.type == "value":
+            return [choice==x for choice in self.choices]
+        elif self.type == "index":
+            return [index==x for index in range(len(choices))]
+        else:
+            raise ValueError("Unknown type: " + str(self.type) + ". Please choose from: 'value', 'index'.")
+
 
 class Image(InputComponent):
     """
@@ -641,7 +698,17 @@ class Image(InputComponent):
         return output_scores.tolist()
 
     def embed(self, x):
-        return x.flatten()
+        shape = (100, 100) if self.shape is None else self.shape  
+        if self.type == "pil":
+            im = x
+        elif self.type == "numpy":
+            im = PIL.Image.fromarray(x)
+        elif self.type == "file":
+            im = PIL.Image.open(x)
+        else:
+            raise ValueError("Unknown type: " + str(self.type) + ". Please choose from: 'numpy', 'pil', 'file'.")
+        im = processing_utils.resize_and_crop(im, (shape[0], shape[1]))
+        return np.asarray(im).flatten()
 
 class Audio(InputComponent):
     """
@@ -723,6 +790,9 @@ class Audio(InputComponent):
         """
         return scores
 
+    def embed(self, x):
+        raise NotImplementedError("Audio doesn't currently support embeddings")
+
 
 class File(InputComponent):
     """
@@ -761,6 +831,8 @@ class File(InputComponent):
         else:
             raise ValueError("Unknown type: " + str(self.type) + ". Please choose from: 'file', 'bytes'.")
 
+    def embed(self, x):
+        raise NotImplementedError("File doesn't currently support embeddings")
 
 
 class Dataframe(InputComponent):
@@ -853,6 +925,9 @@ class Dataframe(InputComponent):
         """
         return np.array(scores).reshape((shape)).tolist()
 
+    def embed(self, x):
+        raise NotImplementedError("DataFrame doesn't currently support embeddings")
+
 
 #######################
 # DEPRECATED COMPONENTS

gradio/networking.py

@@ -20,7 +20,7 @@ import sys
 import csv
 import logging
 import gradio as gr
-from gradio.similarity import calculate_similarity
+from gradio.embeddings import calculate_similarity
 from gradio.tunneling import create_tunnel
 
 INITIAL_PORT_VALUE = int(os.getenv(

gradio/similarity.py

@@ -1,8 +0,0 @@
-import numpy as np
-
-def calculate_similarity(embedding1, embedding2):
-    """
-    Scores the similarity between two embeddings by taking the L2 distance
-    """
-    return np.linalg.norm(np.array(embedding1) - np.array(embedding2))
-