add limt in HF app.py

app.py

@@ -77,7 +77,9 @@ def convert():
     data_docs += f"<hr><h3 style='color:red;'>{pg}</h2><p>{txt}</p><p>{sc}</p>"
   return data_docs
 
-
+# Max values in generation
+DOC_DB_LIMIT = 10
+MAX_NEW_TOKENS = 2048
 
 with gr.Blocks(theme=theme, css=css) as demo:
   with gr.Tab("Chat"):
@@ -93,7 +95,7 @@ with gr.Blocks(theme=theme, css=css) as demo:
         sou = gr.HTML("")
 
   with gr.Tab("Chat Options"):
-    max_docs = gr.inputs.Slider(1, 10, default=3, label="Maximum querys to the DB.", step=1)
+    max_docs = gr.inputs.Slider(1, DOC_DB_LIMIT, default=3, label="Maximum querys to the DB.", step=1)
     row_table = gr.HTML("<hr><h4> </h2>")
     clear_button = gr.Button("CLEAR CHAT HISTORY", )
     link_output = gr.HTML("")
@@ -105,7 +107,7 @@ with gr.Blocks(theme=theme, css=css) as demo:
     gr.HTML("<h3>Only models from the GGML library are accepted.</h3>")
     repo_ = gr.Textbox(label="Repository" ,value="TheBloke/Llama-2-7B-Chat-GGML")
     file_ = gr.Textbox(label="File name" ,value="llama-2-7b-chat.ggmlv3.q2_K.bin")
-    max_tokens = gr.inputs.Slider(1, 2048, default=16, label="Max new tokens", step=1)
+    max_tokens = gr.inputs.Slider(1, MAX_NEW_TOKENS, default=16, label="Max new tokens", step=1)
     temperature = gr.inputs.Slider(0.1, 1., default=0.2, label="Temperature", step=0.1)
     top_k = gr.inputs.Slider(0.01, 1., default=0.95, label="Top K", step=0.01)
     top_p = gr.inputs.Slider(0, 100, default=50, label="Top P", step=1)
@@ -118,9 +120,12 @@ with gr.Blocks(theme=theme, css=css) as demo:
   msg.submit(predict,[msg, chatbot, max_docs],[msg, chatbot]).then(convert,[],[sou])
 
   change_model_button.click(dc.change_llm,[repo_, file_, max_tokens, temperature, top_p, top_k, repeat_penalty, max_docs],[model_verify])
-  falcon_button.click(dc.default_falcon_model, [], [model_verify])
 
+  falcon_button.click(dc.default_falcon_model, [], [model_verify])
 
-DEMO = os.getenv("DEMO")
-print(DEMO)
+# limit in HF, no need to set it
+if "SET_LIMIT" == os.getenv("DEMO"):
+    DOC_DB_LIMIT = 4
+    MAX_NEW_TOKENS = 32
+    
 demo.launch(enable_queue=True)