Update app.py

app.py

@@ -7,7 +7,7 @@ import sys
 import os
 import subprocess
 from pathlib import Path
-import gradio as gr  # Import directly, no patch needed
+import gradio as gr  # Import directly, do not use the patch
 
 # Add project paths
 sys.path.insert(0, str(Path(__file__).parent))
@@ -20,16 +20,21 @@ def run_comparison(iterations: int, seed: int, use_deterministic: bool, device:
     """
     
     # Set device environment variable for subprocess
+    # On Hugging Face Spaces with GPU, try to use CUDA
     if device == "cuda":
         try:
             import torch
+            # Check if CUDA is available
             if torch.cuda.is_available():
                 try:
+                    # Try to get device name to verify GPU works
                     gpu_name = torch.cuda.get_device_name(0)
                     gpu_count = torch.cuda.device_count()
                     print(f"✅ GPU available: {gpu_name} (Count: {gpu_count})")
+                    # Keep device as "cuda"
                 except Exception as e:
                     print(f"⚠️ GPU detection failed: {e}")
+                    print("   Attempting to use CUDA anyway (may work)...")
             else:
                 print("⚠️ CUDA not available, falling back to CPU")
                 device = "cpu"
@@ -40,7 +45,7 @@ def run_comparison(iterations: int, seed: int, use_deterministic: bool, device:
             print(f"⚠️ GPU check error: {e}, falling back to CPU")
             device = "cpu"
     
-    # Set environment variable for subprocess
+    # Set environment variable for subprocess to pick up
     os.environ["CUDA_DEVICE"] = device
     print(f"🔧 Using device: {device}")
     
@@ -57,13 +62,14 @@ def run_comparison(iterations: int, seed: int, use_deterministic: bool, device:
         cmd.extend(["--seed", str(int(seed))])
     
     try:
+        # Ensure environment variables are passed to subprocess
         env = os.environ.copy()
         env["CUDA_DEVICE"] = os.environ.get("CUDA_DEVICE", device)
         
         result = subprocess.run(
             cmd,
             cwd=str(Path(__file__).parent),
-            env=env,
+            env=env,  # Pass environment variables
             capture_output=True,
             text=True,
             timeout=3600  # 1 hour timeout
@@ -72,12 +78,13 @@ def run_comparison(iterations: int, seed: int, use_deterministic: bool, device:
         stdout_text = result.stdout
         stderr_text = result.stderr
         
+        # Combine outputs
         full_output = f"=== STDOUT ===\n{stdout_text}\n\n=== STDERR ===\n{stderr_text}"
         
         if result.returncode != 0:
             return f"❌ Error occurred:\n{full_output}", None
         
-        # Check multiple possible locations for the plot
+        # Find output plot (check multiple possible locations)
         plot_paths = [
             Path(__file__).parent / "teacher_agent_dev" / "comparison_all_strategies.png",
             Path(__file__).parent / "comparison_all_strategies.png",
@@ -93,6 +100,7 @@ def run_comparison(iterations: int, seed: int, use_deterministic: bool, device:
         if plot_path:
             return f"✅ Comparison complete!\n\n{stdout_text}", str(plot_path)
         else:
+            # Return output even if plot not found (might still be useful)
             error_msg = f"⚠️ Plot not found at expected locations.\n"
             error_msg += f"Checked: {[str(p) for p in plot_paths]}\n\n"
             error_msg += f"Output:\n{full_output}"
@@ -109,6 +117,8 @@ def check_gpu():
     """Check if GPU is available on Hugging Face Spaces."""
     try:
         import torch
+        
+        # Check CUDA availability
         if torch.cuda.is_available():
             try:
                 gpu_name = torch.cuda.get_device_name(0)
@@ -116,9 +126,12 @@ def check_gpu():
                 cuda_version = torch.version.cuda
                 return f"✅ GPU Available: {gpu_name} (Count: {gpu_count}, CUDA: {cuda_version})"
             except Exception as e:
+                # GPU might be available but not immediately accessible
                 return f"✅ GPU Detected (accessing: {str(e)[:50]}...)"
         else:
+            # On Hugging Face Spaces, check environment
             if os.getenv("SPACE_ID"):
+                # Check if GPU hardware is allocated
                 hf_hardware = os.getenv("SPACE_HARDWARE", "unknown")
                 if "gpu" in hf_hardware.lower() or "t4" in hf_hardware.lower() or "l4" in hf_hardware.lower():
                     return f"⚠️ GPU Hardware ({hf_hardware}) allocated but not accessible yet. Try running anyway."
@@ -141,9 +154,12 @@ with gr.Blocks(title="MentorFlow - Strategy Comparison") as demo:
     3. **Teacher Strategy**: RL teacher agent learns optimal curriculum
     
     ## Usage
+    
     1. Set parameters below
     2. Click "Run Comparison" to start training
     3. View results and generated plots
+    
+    **Note**: With LM Student, this will take 15-30 minutes for 500 iterations.
     """)
     
     # GPU Status
@@ -156,10 +172,33 @@ with gr.Blocks(title="MentorFlow - Strategy Comparison") as demo:
     # Parameters
     with gr.Row():
         with gr.Column():
-            iterations = gr.Slider(minimum=50, maximum=500, value=100, step=50, label="Iterations")
-            seed = gr.Number(value=42, label="Random Seed")
-            use_deterministic = gr.Checkbox(value=True, label="Deterministic Mode")
-            device = gr.Radio(choices=["cuda", "cpu"], value="cuda", label="Device")
+            iterations = gr.Slider(
+                minimum=50,
+                maximum=500,
+                value=100,
+                step=50,
+                label="Iterations",
+                info="Number of training iterations (higher = longer runtime)"
+            )
+            
+            seed = gr.Number(
+                value=42,
+                label="Random Seed",
+                info="Seed for reproducibility (ignored if deterministic)"
+            )
+            
+            use_deterministic = gr.Checkbox(
+                value=True,
+                label="Deterministic Mode",
+                info="Use fixed seed=42 for reproducible results"
+            )
+            
+            device = gr.Radio(
+                choices=["cuda", "cpu"],
+                value="cuda",  # Default to GPU for HF Spaces with Nvidia 4xL4
+                label="Device",
+                info="GPU (cuda) recommended for Nvidia 4xL4, CPU fallback available"
+            )
         
         with gr.Column():
             run_btn = gr.Button("🚀 Run Comparison", variant="primary", size="lg")
@@ -167,9 +206,19 @@ with gr.Blocks(title="MentorFlow - Strategy Comparison") as demo:
     # Output
     with gr.Row():
         with gr.Column(scale=1):
-            output_text = gr.Textbox(label="Output", lines=15, max_lines=30, interactive=False)
+            output_text = gr.Textbox(
+                label="Output",
+                lines=15,
+                max_lines=30,
+                interactive=False
+            )
+        
         with gr.Column(scale=1):
-            output_plot = gr.Image(label="Comparison Plot", type="filepath", height=500)
+            output_plot = gr.Image(
+                label="Comparison Plot",
+                type="filepath",
+                height=500
+            )
     
     # Run comparison
     run_btn.click(
@@ -178,7 +227,19 @@ with gr.Blocks(title="MentorFlow - Strategy Comparison") as demo:
         outputs=[output_text, output_plot],
         api_name="run_comparison"
     )
+    
+    gr.Markdown("""
+    ## 📊 Understanding Results
+    
+    The comparison plot shows:
+    - **Learning Curves**: How each strategy improves over time
+    - **Difficult Question Performance**: Accuracy on hard questions
+    - **Curriculum Diversity**: Topic coverage over time
+    - **Learning Efficiency**: Iterations to reach target vs final performance
+    
+    The **Teacher Strategy** should ideally outperform Random and Progressive strategies.
+    """)
 
 if __name__ == "__main__":
-    # Standard HF Spaces configuration
+    # Ensure the app binds to all interfaces for HF Spaces
     demo.launch(server_name="0.0.0.0", server_port=7860)