Update app.py

app.py

@@ -7,14 +7,15 @@ import random
 url = 'https://huggingface.co/TheBloke/WizardLM-7B-uncensored-GGML/resolve/main/WizardLM-7B-uncensored.ggmlv3.q2_K.bin'
 filename = wget.download(url)
 llm2 = Llama(model_path=filename, seed=random.randint(1, 2**31))
+filename = wget.download(url)
 theme = gr.themes.Soft(
     primary_hue=gr.themes.Color("#ededed", "#fee2e2", "#fecaca", "#fca5a5", "#f87171", "#ef4444", "#dc2626", "#b91c1c", "#991b1b", "#7f1d1d", "#6c1e1e"),
     neutral_hue="red",
 )
-title = """<h1 align="center">Chat with awesome WizardLM model!</h1><br>"""
+title = """<h1 align="center">Chat with awesome WizardLM 7b model!</h1><br>"""
 with gr.Blocks(theme=theme) as demo:
     gr.HTML(title)
-    gr.HTML("This model is awesome for its size! It's 20 times smaller than ChatGPT but seems to be very smart. However, this model like all models, can output factually incorrect information. Please do not rely on it for high stakes decisions.")
+    gr.HTML("This model is awesome for its size! It is only 20th the size of Chatgpt but is around 90% as good as Chatgpt. However, please don't rely on WizardLM to provide 100% true information as it might be wrong sometimes. ")
     chatbot = gr.Chatbot()
     msg = gr.Textbox()
     clear = gr.ClearButton([msg, chatbot])
@@ -29,9 +30,9 @@ with gr.Blocks(theme=theme) as demo:
         #token1 = llm.tokenize(b"### Instruction: ")
         #token2 = llm.tokenize(instruction.encode())
         #token3 = llm2.tokenize(b"USER: ")
-        tokens5 = llm2.tokenize(user_message.encode())
+        tokens3 = llm2.tokenize(user_message.encode())
         token4 = llm2.tokenize(b"\n\n### Response:")
-        tokens = tokens5 + token4
+        tokens = tokens3 + token4
         history[-1][1] = ""
         count = 0
         output = ""
@@ -52,3 +53,4 @@ with gr.Blocks(theme=theme) as demo:
 
 demo.queue()
 demo.launch(debug=True)
+