Update app.py

app.py

@@ -74,11 +74,11 @@ def respond(
     message,
     history: list[tuple[str, str]],
     system_message,
-    max_tokens,
+    # max_tokens,
     temperature,
-    top_p,
-    top_k,
-    repeat_penalty,
+    # top_p,
+    # top_k,
+    # repeat_penalty,
     model,
 ):
     chat_template = get_messages_formatter_type(model)
@@ -102,10 +102,10 @@ def respond(
     
     settings = provider.get_provider_default_settings()
     settings.temperature = temperature
-    settings.top_k = top_k
-    settings.top_p = top_p
-    settings.max_tokens = max_tokens
-    settings.repeat_penalty = repeat_penalty
+    # settings.top_k = top_k
+    # settings.top_p = top_p
+    # settings.max_tokens = max_tokens
+    # settings.repeat_penalty = repeat_penalty
     settings.stream = True
 
     messages = BasicChatHistory()
@@ -139,28 +139,43 @@ PLACEHOLDER = """
 <div class="message-bubble-border" style="display:flex; max-width: 600px; border-radius: 8px; box-shadow: 0 4px 6px rgba(0, 0, 0, 0.1); backdrop-filter: blur(10px);">
     <div style="padding: .5rem 1.5rem;">
     <img src="https://avatars.githubusercontent.com/u/39557177?v=4" style="width: 80%; max-width: 550px; height: auto; opacity: 0.80;  "> 
-    <h2 dir="rtl" style="text-align: center; font-size: 1.5rem; font-weight: 700; margin-bottom: 0.5rem;">با فرمت‌های GGUF درنا چت کنید!</h2>
+    <h2 dir="rtl" style="text-align: right; font-size: 1.5rem; font-weight: 700; margin-bottom: 0.5rem;">با فرمت‌های GGUF درنا چت کنید!</h2>
 </div>
 """
 demo = gr.ChatInterface(
     respond,
     additional_inputs=[
         gr.Textbox(value="", label="System message", rtl=False),
-        gr.Slider(minimum=0.1, maximum=4.0, value=0.1, step=0.1, label="Temperature"),
-        gr.Slider(
-            minimum=0.1,
-            maximum=1.0,
-            value=0.85,
-            step=0.05,
-            label="Top-p",
-        ),
+        #gr.Slider(minimum=1, maximum=8192, value=2048, step=1, label="Max tokens"),
+        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
+        # gr.Slider(
+        #     minimum=0.1,
+        #     maximum=1.0,
+        #     value=0.85,
+        #     step=0.05,
+        #     label="Top-p",
+        # ),
+        # gr.Slider(
+        #     minimum=0,
+        #     maximum=100,
+        #     value=40,
+        #     step=1,
+        #     label="Top-k",
+        # ),
+        # gr.Slider(
+        #     minimum=0.0,
+        #     maximum=2.0,
+        #     value=1,
+        #     step=0.1,
+        #     label="Repetition penalty",
+        # ),
         gr.Dropdown([
                 'dorna-llama3-8b-instruct.Q8_0.gguf',
                 'dorna-llama3-8b-instruct.Q4_0.gguf',
                 'dorna-llama3-8b-instruct.Q5_0.gguf',
                 'dorna-llama3-8b-instruct.bf16.gguf',
             ],
-            value="dorna-llama3-8b-instruct.Q8_0.gguf",
+            value="dorna-llama3-8b-instruct.Q4_0.gguf",
             label="Model"
         ),
     ],