Second changes - includes the generated text in images and videos

app.py

@@ -13,18 +13,20 @@ with gr.Blocks() as demo:
         vid_input = gr.Video()
         vid_output = gr.Textbox()
         vid_output_interpret = gr.TextArea()
+        vid_audio_stmt = gr.Textbox()
         vid_button = gr.Button("Check Bias in your Video")
 
     with gr.Tab("Image"):
         img_input = gr.Image()
         img_output = gr.Textbox()
+        img_gen_stmt = gr.Textbox()
         img_output_interpret = gr.TextArea()
         img_button = gr.Button("Check Bias in your Image")
 
     text_button.click(combine_modules.app_nlp_start, inputs=text_input, outputs=[text_output, text_output_interpret])
-    vid_button.click(combine_modules.app_video_start, inputs=vid_input, outputs=[vid_output, vid_output_interpret])
-    img_button.click(combine_modules.app_image_start, inputs=img_input, outputs=[img_output, img_output_interpret])
+    vid_button.click(combine_modules.app_video_start, inputs=vid_input, outputs=[vid_output, vid_audio_stmt, vid_output_interpret])
+    img_button.click(combine_modules.app_image_start, inputs=img_input, outputs=[img_output,  img_gen_stmt, img_output_interpret])
 
 
 if __name__=="__main__":
-    demo.launch()
+    demo.launch(share=True)

combine_modules.py

@@ -8,10 +8,10 @@ def app_nlp_start(statement):
 
 def app_video_start(video_path):
     return_text = rtxt(video_path)
-    output_txt = bias_checker(return_text)
-    return output_txt
+    output_txt, interpreted_txt = bias_checker(return_text)
+    return output_txt, interpreted_txt, return_text
 
 def app_image_start(image_path):
     text_generated = img_pipe(image_path)
-    output_txt = bias_checker(text_generated)
-    return output_txt
+    output_txt, interpreted_txt = bias_checker(text_generated)
+    return output_txt, interpreted_txt, text_generated

interpret_model_pt.py

@@ -1,8 +1,8 @@
 from transformers_interpret import SequenceClassificationExplainer, MultiLabelClassificationExplainer
 
 def explainer(input_statement, model, tokenizer):
-    # cls_explainer = SequenceClassificationExplainer(model, tokenizer)
-    cls_explainer = MultiLabelClassificationExplainer(model, tokenizer)
+    cls_explainer = SequenceClassificationExplainer(model, tokenizer)
+    # cls_explainer = MultiLabelClassificationExplainer(model, tokenizer)
     word_attributions = cls_explainer(input_statement)
     return dict(word_attributions)