Update app.py

app.py

@@ -1,49 +1,73 @@
-import streamlit as st
-from langchain.prompts import PromptTemplate
-from langchain.llms import CTransformers
 import gradio as gr
-## Function To get response from LLAma 2 model
-
-def getLLamaresponse(message):
-    
-    input_text = "home decoration" 
-    no_words = "100"
-    blog_style = "lifestyle"
-
-    ### LLama2 model
-    llm=CTransformers(model='TheBloke/OpenHermes-2.5-Mistral-7B-GGUF',
-                      model_type='llama',
-                      config={'max_new_tokens':256,
-                              'temperature':0.01})
-    
-    ## Prompt Template
-
-    template="""
-        Write a blog for {blog_style} job profile for a topic {input_text}
-        within {no_words} words.
-            """
-    
-    prompt=PromptTemplate(input_variables=["blog_style","input_text",'no_words'],
-                          template=template)
-    
-    ## Generate the ressponse from the LLama 2 model
-    response=llm(prompt.format(blog_style=blog_style,input_text=input_text,no_words=no_words))
-    print(response)
-    return response
+import os
+from pathlib import Path
+import argparse
+from huggingface_hub import snapshot_download
+
+
+# repo_name = "TheBloke/Mistral-7B-v0.1-GGUF"
+# model_file = "mistral-7b-v0.1.Q6_K.gguf"
+
+
+repo_name = 'HumanityFTW/so_rude'
+model_file = "mistral-comedy-2.0-ckpt-600.Q6_K.gguf"
+
+
+print('Fetching model:', repo_name, model_file)
+snapshot_download(repo_id=repo_name, local_dir=".", allow_patterns=model_file)
+print('Done fetching model:')
+
+DEFAULT_MODEL_PATH = model_file
+
+from llama_cpp import Llama
+llm = Llama(model_path=model_file, model_type="mistral")
+
+
+def predict(input, chatbot, max_length, top_p, temperature, history):
+    chatbot.append((input, ""))
+    response = ""
+    history.append(input)
+
+    for output in llm(input, stream=True, temperature=temperature, top_p=top_p, max_tokens=max_length, ):
+        piece = output['choices'][0]['text']
+        response += piece
+        chatbot[-1] = (chatbot[-1][0], response)
+
+        yield chatbot, history
+
+    history.append(response)
+    yield chatbot, history
+
+
+def reset_user_input():
+    return gr.update(value="")
+
+
+def reset_state():
+    return [], []
 
 
 with gr.Blocks() as demo:
-    gr.Markdown("# AI Patient Chatbot")
-    with gr.Group():
-        with gr.Tab("Patient Chatbot"):
-            chatbot = gr.Chatbot()
-            message = gr.Textbox(label="Enter your message to Barry", placeholder="Type here...", lines=2)
-            send_message = gr.Button("Submit")
-            send_message.click(getLLamaresponse, inputs=[message], outputs=[chatbot])
-            save_chatlog = gr.Button("Save Chatlog")
-            #send_message.click(SaveChatlog, inputs=[message], outputs=[chatbot])
+    gr.HTML("""<h1 align="center">So Rude</h1>""")
+
+    chatbot = gr.Chatbot()
+    with gr.Row():
+        with gr.Column(scale=4):
+            user_input = gr.Textbox(show_label=False, placeholder="Input...", lines=8, elem_id="user_input")
+            submitBtn = gr.Button("Submit", variant="primary", elem_id="submit_btn")
+        with gr.Column(scale=1):
+            max_length = gr.Slider(0, 256, value=64, step=1.0, label="Maximum Length", interactive=True)
+            top_p = gr.Slider(0, 1, value=0.7, step=0.01, label="Top P", interactive=True)
+            temperature = gr.Slider(0, 2.0, value=0.95, step=0.01, label="Temperature", interactive=True)
+            emptyBtn = gr.Button("Clear History")
+
+    history = gr.State([])
 
+    submitBtn.click(
+        predict, [user_input, chatbot, max_length, top_p, temperature, history], [chatbot, history], show_progress=True
+    )
+    submitBtn.click(reset_user_input, [], [user_input])
 
-            #message.submit(AIPatient, inputs=[message], outputs=[chatbot])
+    emptyBtn.click(reset_state, outputs=[chatbot, history], show_progress=True)
 
-demo.launch(debug=True)
+demo.queue().launch(share=False, inbrowser=True)