visuals

app.py

@@ -3,6 +3,26 @@ from root import RootSignals
 
 client = None
 custom_judge = None
+MODELS = [
+    "claude-3-5-sonnet",
+    "claude-3-haiku-20240307",
+    "claude-3-opus-20240229",
+    "claude-3-sonnet-20240229",
+    "codestral",
+    "command-r",
+    "command-r-plus",
+    "fireworks_ai/llama-v3-70b-instruct",
+    "gpt-4",
+    "gpt-4o",
+    "gpt-4o-mini",
+    "gpt-4-turbo",
+    "groq/llama3-70b-8192",
+    "mistral-large-latest",
+    "mistral-medium",
+    "o1-mini",
+    "o1-preview",
+    "open-codestral-mamba",
+]
 
 def initialize_client(api_key):
     global client
@@ -21,11 +41,7 @@ def create_judge(api_key, judge_name, intent, judge_prompt):
         model="gpt-4o",
     )
     
-    # Update the visibility of the evaluation and results sections
-    eval_section.visible = True
-    results_section.visible = True
-    evaluate_btn.visible = True
-    return gr.Info(f"Custom LLM-Judge {judge_name} is created successfully!")
+    return gr.Info(f"Custom LLM-Judge '{judge_name}' is created successfully!")
 
 def evaluate_response(llm_response):
     global client, custom_judge
@@ -58,26 +74,41 @@ with gr.Blocks(theme=gr.themes.Default(primary_hue="blue")) as demo:
         show_label=True,
     )
     
+    gr.Markdown("---")  # Divider
+
+    gr.Markdown("### Create Custom Judge")
+    with gr.Row():
+        judge_name = gr.Textbox(label="👨‍⚖️ Judge Name", placeholder="Enter a name for your custom judge...", interactive=True)
+        user_intent = gr.Textbox(label="👤 Intent", placeholder="Enter the high-level intent for this judge...", interactive=True)
+    with gr.Row():
+        judge_prompt = gr.Textbox(
+            label="📝 Custom Judge Prompt",
+            placeholder="Enter the custom judge prompt...",
+            interactive=True,
+            lines=5,
+            max_lines=10
+        )
+        create_judge_btn = gr.Button("✨ CREATE JUDGE", variant="primary")
+    info_message = gr.Info()
+    
+    gr.Markdown("---")  # Divider
+    
     with gr.Row():
-        # Left column - Judge Creation
+        # Left column - Evaluation
         with gr.Column():
-            gr.Markdown("### Create Custom Judge")
-            judge_name = gr.Textbox(label="👨‍⚖️ Judge Name", placeholder="Enter a name for your custom judge...", interactive=True)
-            user_intent = gr.Textbox(label="👤 Intent", placeholder="Enter the high-level intent for this judge...", interactive=True)
-            judge_prompt = gr.Textbox(label="📝 Custom Judge Prompt", placeholder="Enter the custom judge prompt...", interactive=True)
-            create_judge_btn = gr.Button("✨ CREATE JUDGE", variant="primary")
-            info_message = gr.Info()
-            
-            # Evaluation section (initially hidden)
-            eval_section = gr.Column(visible=True)
-            with eval_section:
-                gr.Markdown("### Evaluate Response")
-                llm_response = gr.Textbox(label="🤖 LLM Response", placeholder="Enter the LLM response to be evaluated...", interactive=True)
-                evaluate_btn = gr.Button("🧐 EVALUATE", variant="primary", visible=True)
+            gr.Markdown("### Evaluate Response")
+            llm_response = gr.Textbox(
+                label="🤖 LLM Response", 
+                placeholder="Enter the LLM response to be evaluated...", 
+                interactive=True,
+                lines=5,
+                max_lines=10
+            )
+            evaluate_btn = gr.Button("🧐 EVALUATE", variant="primary", visible=True)
         
         # Right column - Results
-        results_section = gr.Column(visible=True)
-        with results_section:
+        with gr.Column():
+            gr.Markdown("### Results")
             score = gr.Textbox(label="📊 Score (between 0 and 1)", interactive=False)
             justification = gr.TextArea(label="💬 Justification", interactive=False)