update cap

app.py

@@ -244,7 +244,7 @@ def gradio_ask(user_message, chatbot, chat_state,radio):
 
     chat.ask(user_message, chat_state,radio)
     chatbot = chatbot + [[user_message, None]]
-    return '', chatbot, chat_state
+    return chatbot, chat_state
 
 
 def gradio_answer(chatbot, chat_state, img_list,  radio, text,num_beams, temperature):
@@ -320,7 +320,7 @@ with gr.Blocks() as demo:
     upload_button.click(upload_img, [image, text_input, chat_state,chatbot],
                         [image, text_input, upload_button, chat_state, img_list,chatbot])
 
-    text_input.submit(gradio_ask, [text_input, chatbot, chat_state,radio], [text_input, chatbot, chat_state]).then(
+    text_input.submit(gradio_ask, [text_input, chatbot, chat_state,radio], [chatbot, chat_state]).then(
         gradio_answer, [chatbot, chat_state, img_list,  radio, text_input,num_beams, temperature], [text_input,chatbot, chat_state, img_list]
     )
     clear.click(gradio_reset, [chat_state, img_list], [chatbot, image, text_input, upload_button, chat_state, img_list],

multimodal/open_flamingo/chat/conversation.py

@@ -457,7 +457,9 @@ class Chat:
             # print(
             #     f"### Assistant: {tokenizer.decode(outputs[0, input_ids.shape[1]:], skip_special_tokens=True).strip()}")
             output_text = self.tokenizer.decode(text_outputs[0])
-            output_text = re.findall(r'Assistant:(.+)', output_text)[-1]
+            print(output_text)
+            output_text = re.findall(r'Assistant:(.+)', output_text)
+            print(output_text)
 
         return output_text, out_image
 

multimodal/open_flamingo/eval/task/caption_chat.py

@@ -171,7 +171,7 @@ def captioner(
         else:
             break
     outputs = outputs[:, ori_prompt_length:]
-    outputs = postprocess_captioning_generation(tokenizer.batch_decode(outputs, skip_special_tokens=True)[0]).replace('"', "")
+    outputs = tokenizer.batch_decode(outputs, skip_special_tokens=True)[0].replace('"', "")
     # new_predictions = [
     #     postprocess_captioning_generation(out).replace('"', "")
     #     for out in tokenizer.batch_decode(outputs, skip_special_tokens=True)