Update app.py

app.py

@@ -17,28 +17,8 @@ InferenceClient(models[2]),
 InferenceClient(models[3]),
 ]
 
-def compress_history(history,client_choice,seed,temp,tokens,top_p,rep_p,chat_mem):
-    client=clients[int(client_choice)-1]
-    COMPRESS_HISTORY="""You are an Information Summarizer Agent. Your duty is to summarize the following information into a more concise format with far less words.
-    Retain all the main points and provide a brief and concise summary of the conversation.
-    Converstion:
-    {history}"""
-    print("COMPRESSING")
-    formatted_prompt=f"{COMPRESS_HISTORY.format(history=history[0-chat_mem:])}"
-    generate_kwargs = dict(
-        temperature=temp,
-        max_new_tokens=1024,
-        top_p=top_p,
-        repetition_penalty=rep_p,
-        do_sample=True,
-        seed=seed,
-    )
-    stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=False)
-    output = ""
-    for response in stream:
-        output += response.token.text
-    return output
-    
+VERBOSE=True
+
 def format_prompt(message, history):
     prompt = ""
     if history:
@@ -49,10 +29,7 @@ def format_prompt(message, history):
             prompt += f"{bot_response}\n"
             #print(prompt)
     prompt += f"<start_of_turn>user{message}<end_of_turn><start_of_turn>model"
-    #print(prompt)
     return prompt
-result = []
-
 
 def chat_inf(system_prompt,prompt,history,memory,client_choice,seed,temp,tokens,top_p,rep_p,chat_mem):
     #token max=8192
@@ -67,18 +44,11 @@ def chat_inf(system_prompt,prompt,history,memory,client_choice,seed,temp,tokens,
     if memory:
         for ea in memory[0-chat_mem:]:
             hist_len+=len(str(ea))
-        print(hist_len)
     in_len=len(system_prompt+prompt)+hist_len
-    
-    
-    print("\n######### HIST "+str(in_len))
-    print("\n######### TOKENS "+str(tokens))
+
     if (in_len+tokens) > 8000:
         history.append((prompt,"Wait, that's too many tokens, please reduce the 'Chat Memory' value, or reduce the 'Max new tokens' value"))
         yield history,memory
-        #hist=compress_history(history,client_choice,seed,temp,tokens,top_p,rep_p,chat_mem)
-        #yield [(prompt,"History has been compressed, processing request...")]
-        #history.append((prompt,hist))
     else:
         generate_kwargs = dict(
             temperature=temp,
@@ -88,9 +58,7 @@ def chat_inf(system_prompt,prompt,history,memory,client_choice,seed,temp,tokens,
             do_sample=True,
             seed=seed,
         )
-        #formatted_prompt=prompt   
         formatted_prompt = format_prompt(f"{system_prompt}, {prompt}", memory[0-chat_mem:])
-        print("\n######### PROMPT "+str(len(formatted_prompt)))
         stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=False)
         output = ""
         for response in stream:
@@ -99,6 +67,10 @@ def chat_inf(system_prompt,prompt,history,memory,client_choice,seed,temp,tokens,
         history.append((prompt,output))
         memory.append((prompt,output))
         yield history,memory
+    if VERBOSE==True:
+        print("\n######### HIST "+str(in_len))
+        print("\n######### TOKENS "+str(tokens))        
+        print("\n######### PROMPT "+str(len(formatted_prompt)))
 
 def get_screenshot(chat: list,height=5000,width=600,chatblock=[],theme="light",wait=3000,header=True):
     print(chatblock)
@@ -110,18 +82,15 @@ def get_screenshot(chat: list,height=5000,width=600,chatblock=[],theme="light",w
     print(out)
     return out
 
-
-
 def clear_fn():
     return None,None,None
 rand_val=random.randint(1,1111111111111111)
+
 def check_rand(inp,val):
     if inp==True:
         return gr.Slider(label="Seed", minimum=1, maximum=1111111111111111, value=random.randint(1,1111111111111111))
     else:
         return gr.Slider(label="Seed", minimum=1, maximum=1111111111111111, value=int(val))
-
-
     
 with gr.Blocks() as app:
     memory=gr.State()
@@ -140,7 +109,6 @@ with gr.Blocks() as app:
                             stop_btn=gr.Button("Stop")
                             clear_btn=gr.Button("Clear")                
                 client_choice=gr.Dropdown(label="Models",type='index',choices=[c for c in models],value=models[0],interactive=True)
-
             with gr.Column(scale=1):
                 with gr.Group():
                     rand = gr.Checkbox(label="Random Seed", value=True)
@@ -162,14 +130,10 @@ with gr.Blocks() as app:
                     wait_time=gr.Number(label="Wait Time",value=3000)
                     theme=gr.Radio(label="Theme", choices=["light","dark"],value="light")
                     chatblock=gr.Dropdown(label="Chatblocks",info="Choose specific blocks of chat",choices=[c for c in range(1,40)],multiselect=True)
-            
-
 
     im_go=im_btn.click(get_screenshot,[chat_b,im_height,im_width,chatblock,theme,wait_time],img)
     chat_sub=inp.submit(check_rand,[rand,seed],seed).then(chat_inf,[sys_inp,inp,chat_b,memory,client_choice,seed,temp,tokens,top_p,rep_p,chat_mem],[chat_b,memory])
-    
     go=btn.click(check_rand,[rand,seed],seed).then(chat_inf,[sys_inp,inp,chat_b,memory,client_choice,seed,temp,tokens,top_p,rep_p,chat_mem],[chat_b,memory])
-    
     stop_btn.click(None,None,None,cancels=[go,im_go,chat_sub])
-    clear_btn.click(clear_fn,None,[inp,sys_inp,chat_b])
+    clear_btn.click(clear_fn,None,[inp,sys_inp,chat_b,memory])
 app.queue(default_concurrency_limit=10).launch()