Update app.py

app.py

@@ -82,45 +82,45 @@ def load_models(inp):
 VERBOSE=False
 
 
-def chat_inf(system_prompt,prompt,history,memory,model_state,model_name,seed,temp,tokens,top_p,rep_p,chat_mem,cust_p):
+def chat_inf(prompt,model_state,model_name,seed,temp,tokens,top_p,rep_p,chat_mem,cust_p):
     #token max=8192
-    model_n=models[model_name]
-    print(model_state)
+    #model_n=models[model_name]
+    #print(model_state)
     hist_len=0
     client=model_state
-    if not history:
-        history = []
-        hist_len=0
-    if not memory:
-        memory = []
-        mem_len=0        
-    if memory:
-        for ea in memory[0-chat_mem:]:
-            hist_len+=len(str(ea))
-    in_len=len(system_prompt+prompt)+hist_len
-
-    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
-    else:
-        generate_kwargs = dict(
-            temperature=temp,
-            max_new_tokens=tokens,
-            top_p=top_p,
-            repetition_penalty=rep_p,
-            do_sample=True,
-            seed=seed,
-        )
-
-        stream = client.text_generation(prompt, **generate_kwargs, stream=True, details=True, return_full_text=True)
-        output = ""
-        for response in stream:
-            output += response.token.text
-            yield [(prompt,output)],memory
-        history.append((prompt,output))
-        memory.append((prompt,output))
-        yield history,memory
-        
+    #if not history:
+    #    history = []
+    #    hist_len=0
+    #if not memory:
+    #    memory = []
+    #    mem_len=0        
+    #if memory:
+    #    for ea in memory[0-chat_mem:]:
+    #        hist_len+=len(str(ea))
+    #in_len=len(system_prompt+prompt)+hist_len
+
+    #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
+    #else:
+    generate_kwargs = dict(
+        temperature=temp,
+        max_new_tokens=tokens,
+        top_p=top_p,
+        repetition_penalty=rep_p,
+        do_sample=True,
+        seed=seed,
+    )
+
+    stream = client.text_generation(prompt, **generate_kwargs, stream=True, details=True, return_full_text=True)
+    output = prompt
+    for response in stream:
+        output += response.token.text
+        yield output
+    #history.append((prompt,output))
+    #memory.append((prompt,output))
+    yield output
+    
 
 
 def get_screenshot(chat: list,height=5000,width=600,chatblock=[],theme="light",wait=3000,header=True):
@@ -147,7 +147,10 @@ with gr.Blocks() as app:
     model_state=gr.State()
     memory=gr.State()
     gr.HTML("""<center><h1 style='font-size:xx-large;'>Huggingface Hub InferenceClient</h1><br><h3>Chatbot's</h3></center>""")
-    chat_b = gr.Chatbot(height=500)
+    #chat_b = gr.Chatbot(height=500)
+    chat_b = gr.Textbox(lines=10)
+    
+    
     with gr.Group():
         with gr.Row():
             with gr.Column(scale=3):
@@ -187,7 +190,9 @@ with gr.Blocks() as app:
                     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)
         test_json=gr.JSON(label="Test Output")
-    test_btn.click(test_models,None,test_json)
+    #test_btn.click(test_models,None,test_json)
+
+    c_comp=chat_b.change(chat_inf,[chat_b,model_state,client_choice,seed,temp,tokens,top_p,rep_p,chat_mem,custom_prompt],[chat_b,memory]))
     
     client_choice.change(load_models,client_choice,[chat_b,custom_prompt,model_state])
     app.load(load_models,client_choice,[chat_b,custom_prompt,model_state])