Quality of life updates (#1)

- Quality of life updates (48919c166ae9f660dea8c2d42c3ca31420f64ab5)


Co-authored-by: yuvraj sharma <ysharma@users.noreply.huggingface.co>

app.py

@@ -86,7 +86,7 @@ def parse_text(text):
     return text
 
 
-def predict(input, chatbot, max_length, top_p, temperature, history, past_key_values):
+def predict(RETRY_FLAG, input, chatbot, max_length, top_p, temperature, history, past_key_values):
     chatbot.append((parse_text(input), ""))
     for response, history, past_key_values in model.stream_chat(tokenizer, input, history, past_key_values=past_key_values,
                                                                 return_past_key_values=True,
@@ -129,9 +129,50 @@ def reset_user_input():
 def reset_state():
     return [], [], None
 
+# Delete last turn
+def delete_last_turn(chat, history):
+    if chat and history:
+        chat.pop(-1)
+        history.pop(-1)
+    return chat, history
+
+
+# Regenerate response
+def retry_last_answer(
+    user_input, 
+    chatbot, 
+    max_length, 
+    top_p, 
+    temperature, 
+    history, 
+    past_key_values
+      ):
+          
+    if chatbot and history:
+        # Removing the previous conversation from chat 
+        chatbot.pop(-1)
+        # Setting up a flag to capture a retry 
+        RETRY_FLAG = True
+        # Getting last message from user
+        user_input = history[-1][0]
+        # Removing bot response from the history 
+        history.pop(-1)
+
+    yield from predict(
+        RETRY_FLAG,
+        user_input, 
+        chatbot, 
+        max_length, 
+        top_p, 
+        temperature, 
+        history, 
+        past_key_values
+        )
 
 with gr.Blocks(theme=gr.themes.Soft()) as demo:
     gr.HTML("""<h1 align="center">ChatGLM2-6B-int4</h1>""")
+    gr.HTML("""<center><a href="https://huggingface.co/spaces/mikeee/chatglm2-6b-4bit?duplicate=true"><img src="https://bit.ly/3gLdBN6" alt="Duplicate Space"></a>To avoid the queue and for faster inference Duplicate this Space and upgrade to GPU</center>""")
+
     with gr.Accordion("Info", open=False):
         _ = """
             A query takes from 30 seconds to a few tens of seconds, dependent on the number of words/characters 
@@ -154,10 +195,14 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
     with gr.Row():
         with gr.Column(scale=4):
             with gr.Column(scale=12):
-                user_input = gr.Textbox(show_label=False, placeholder="Input...", lines=10).style(
+                user_input = gr.Textbox(show_label=False, placeholder="Input...", ).style(
                     container=False)
+                RETRY_FLAG = gr.Checkbox(value=False, visible=False)
             with gr.Column(min_width=32, scale=1):
-                submitBtn = gr.Button("Submit", variant="primary")
+                with gr.Row():
+                    submitBtn = gr.Button("Submit", variant="primary")
+                    deleteBtn = gr.Button("Delete last turn", variant="secondary")
+                    retryBtn = gr.Button("Regenerate", variant="secondary")
         with gr.Column(scale=1):
             emptyBtn = gr.Button("Clear History")
             max_length = gr.Slider(0, 32768, value=8192/2, step=1.0, label="Maximum length", interactive=True)
@@ -175,6 +220,14 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
 
     emptyBtn.click(reset_state, outputs=[chatbot, history, past_key_values], show_progress=True)
 
+    retryBtn.click(
+        retry_last_answer, 
+        inputs = [user_input, chatbot, max_length, top_p, temperature, history, past_key_values],
+        #outputs = [chatbot, history, last_user_message, user_message]
+        outputs=[chatbot, history, past_key_values]
+        )
+    deleteBtn.click(delete_last_turn, [chatbot, history], [chatbot, history])    
+    
     with gr.Accordion("For Translation API", open=False):
         input_text = gr.Text()
         tr_btn = gr.Button("Go", variant="primary")
@@ -182,6 +235,18 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
     tr_btn.click(trans_api, [input_text, max_length, top_p, temperature], out_text, show_progress=True, api_name="tr")       
     input_text.submit(trans_api, [input_text, max_length, top_p, temperature], out_text, show_progress=True, api_name="tr")      
 
+    with gr.Accordion("Example inputs", open=True):
+        examples = gr.Examples(
+            examples=[["Explain the plot of Cinderella in a sentence."],
+                     ["How long does it take to become proficient in French, and what are the best methods for retaining information?"],
+                     ["What are some common mistakes to avoid when writing code?"],
+                     ["Build a prompt to generate a beautiful portrait of a horse"],
+                     ["Suggest four metaphors to describe the benefits of AI"],
+                     ["Write a pop song about leaving home for the sandy beaches."],
+                     ["Write a summary demonstrating my ability to tame lions"]],
+            inputs = [user_input],
+            
+        )    
 # demo.queue().launch(share=False, inbrowser=True)
 # demo.queue().launch(share=True, inbrowser=True, debug=True)