Update app.py

app.py

@@ -8,26 +8,44 @@ client = InferenceClient("mistralai/Mistral-7B-Instruct-v0.3")
 def switch_client(model_name: str):
     return InferenceClient(model_name)
 
+# Define presets for each model
+presets = {
+    "mistralai/Mistral-7B-Instruct-v0.3": {
+        "Fast": {"max_tokens": 256, "temperature": 1.0, "top_p": 0.9},
+        "Normal": {"max_tokens": 512, "temperature": 0.7, "top_p": 0.95},
+        "Quality": {"max_tokens": 1024, "temperature": 0.5, "top_p": 0.90},
+        "Unreal Performance": {"max_tokens": 2048, "temperature": 0.6, "top_p": 0.75},
+    }
+}
+
+# Fixed system message
+SYSTEM_MESSAGE = "Lake 1 Base"
+
 def respond(
     message,
-    history: list[dict],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-    model_name
+    history: list,
+    model_name,
+    preset_name
 ):
     # Switch client based on model selection
     global client
     client = switch_client(model_name)
     
-    messages = [{"role": "system", "content": system_message}]
+    messages = [{"role": "system", "content": SYSTEM_MESSAGE}]
 
+    # Ensure history is a list of dictionaries
     for val in history:
-        messages.append({"role": val['role'], "content": val['content']})
+        if isinstance(val, dict) and 'role' in val and 'content' in val:
+            messages.append({"role": val['role'], "content": val['content']})
 
     messages.append({"role": "user", "content": message})
 
+    # Get the preset settings
+    preset = presets[model_name][preset_name]
+    max_tokens = preset["max_tokens"]
+    temperature = preset["temperature"]
+    top_p = preset["top_p"]
+
     # Get the response from the model
     response = client.chat_completion(
         messages,
@@ -52,37 +70,24 @@ pseudonyms = [model[1] for model in model_choices]
 # Function to handle model selection and pseudonyms
 def respond_with_pseudonym(
     message,
-    history: list[dict],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-    selected_pseudonym
+    history: list,
+    selected_pseudonym,
+    selected_preset
 ):
     # Find the actual model name from the pseudonym
     model_name = next(model[0] for model in model_choices if model[1] == selected_pseudonym)
     
     # Call the existing respond function
-    response = respond(message, history, system_message, max_tokens, temperature, top_p, model_name)
+    response = respond(message, history, model_name, selected_preset)
     
-    # No longer adding the pseudonym at the end of the response
     return response
 
 # Gradio Chat Interface
 demo = gr.ChatInterface(
-    respond_with_pseudonym,
+    fn=respond_with_pseudonym,
     additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new 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=0.95,
-            step=0.05,
-            label="Top-p (nucleus sampling)",
-        ),
-        gr.Dropdown(pseudonyms, label="Select Model", value=pseudonyms[0])  # Pseudonym selection dropdown
+        gr.Dropdown(choices=list(presets["mistralai/Mistral-7B-Instruct-v0.3"].keys()), label="Select Preset", value="Fast"),  # Preset selection dropdown
+        gr.Dropdown(choices=pseudonyms, label="Select Model", value=pseudonyms[0])  # Pseudonym selection dropdown
     ],
 )