Update app.py

app.py

@@ -25,6 +25,7 @@ import streamlit as st
 import pandas as pd
 import plotly.express as px
 import plotly.graph_objects as go
+from sklearn.preprocessing import LabelEncoder
 
 # Add src directory to path for imports
 sys.path.insert(0, str(Path(__file__).parent / 'src'))
@@ -143,6 +144,8 @@ def init_session_state():
         # Data
         'uploaded_data': None,
         'preprocessed_data': None,
+        'label_encoder': None,
+        'label_classes': None,
         
         # Evaluation
         'evaluation_results': None,
@@ -309,10 +312,10 @@ tab1, tab2, tab3, tab4, tab5 = st.tabs([
 # ==================== TAB 1: Prerequisites ====================
 
 with tab1:
-    st.markdown("## 🔧 System Prerequisites")
+    st.markdown("## System Prerequisites")
     
     create_info_box(
-        "⚠️ <b>Important:</b> Complete all prerequisite checks before proceeding to training.<br>"
+        "<b>Important:</b> Complete all prerequisite checks before proceeding to training.<br>"
         "This ensures your system is properly configured and all required models are downloaded.",
         "warning"
     )
@@ -321,13 +324,13 @@ with tab1:
     system_checker = SystemChecker(models_dir="models")
     
     # ===== CUDA/GPU Check =====
-    st.markdown("### 🎮 1. CUDA/GPU Check")
+    st.markdown("### 1. CUDA/GPU Check")
     
     col1, col2 = st.columns([3, 1])
     with col1:
         st.markdown("Check if CUDA-capable GPU is available for faster training.")
     with col2:
-        if st.button("🔍 Check CUDA", width="stretch"):
+        if st.button("Check CUDA", width="stretch"):
             with st.spinner("Checking CUDA availability..."):
                 cuda_status = system_checker.check_cuda()
                 st.session_state.cuda_status = cuda_status
@@ -337,24 +340,24 @@ with tab1:
         cuda = st.session_state.cuda_status
         
         if cuda['available']:
-            st.success(f"✅ CUDA Available - {cuda['device_count']} GPU(s) detected")
+            st.success(f"CUDA Available - {cuda['device_count']} GPU(s) detected")
             
             for device in cuda['devices']:
-                with st.expander(f"📊 {device['name']} Details"):
+                with st.expander(f"Device: {device['name']} Details"):
                     col1, col2, col3 = st.columns(3)
                     col1.metric("Memory", f"{device['memory_total']:.2f} GB")
                     col2.metric("Compute", device['compute_capability'])
                     col3.metric("CUDA Version", cuda['cuda_version'])
             
             create_info_box(
-                "💡 <b>Recommendation:</b> Your GPU is ready for training! "
+                "<b>Recommendation:</b> Your GPU is ready for training! "
                 "You can use any model from the list. XLM-RoBERTa and RoBERTa are recommended for best accuracy.",
                 "success"
             )
         else:
-            st.warning("⚠️ No CUDA-capable GPU detected - Training will use CPU")
+            st.warning("No CUDA-capable GPU detected - Training will use CPU")
             create_info_box(
-                "💡 <b>Recommendation:</b> For CPU training, we recommend using <b>distilbert-base-multilingual-cased</b> "
+                "<b>Recommendation:</b> For CPU training, we recommend using <b>distilbert-base-multilingual-cased</b> "
                 "as it's significantly faster while maintaining good accuracy.",
                 "warning"
             )
@@ -362,13 +365,13 @@ with tab1:
     st.markdown("---")
     
     # ===== Environment Check =====
-    st.markdown("### 🐍 2. Environment Check")
+    st.markdown("### 2. Environment Check")
     
     col1, col2 = st.columns([3, 1])
     with col1:
         st.markdown("Verify all required Python packages are installed with correct versions.")
     with col2:
-        if st.button("🔍 Check Environment", width="stretch"):
+        if st.button("Check Environment", width="stretch"):
             with st.spinner("Checking environment..."):
                 env_status = system_checker.check_environment()
                 st.session_state.env_status = env_status
@@ -378,22 +381,22 @@ with tab1:
         env = st.session_state.env_status
         
         if env['all_satisfied']:
-            st.success("✅ All required packages are installed")
+            st.success("All required packages are installed")
         else:
-            st.error(f"❌ Missing packages: {', '.join(env['missing_packages'])}")
+            st.error(f"Missing packages: {', '.join(env['missing_packages'])}")
             create_info_box(
                 f"<b>To install missing packages, run:</b><br>"
                 f"<code>pip install {' '.join(env['missing_packages'])}</code>",
                 "error"
             )
         
-        with st.expander("📦 View Package Details"):
+        with st.expander("View Package Details"):
             package_df = pd.DataFrame([
                 {
                     'Package': pkg,
                     'Installed': info['installed'] or 'Not Installed',
                     'Required': info['required'],
-                    'Status': '✅' if info['satisfied'] else '❌'
+                    'Status': 'OK' if info['satisfied'] else 'Missing'
                 }
                 for pkg, info in env['packages'].items()
             ])
@@ -402,10 +405,10 @@ with tab1:
     st.markdown("---")
     
     # ===== Model Selection Guide =====
-    st.markdown("### 📚 3. Model Selection Guide")
+    st.markdown("### 3. Model Selection Guide")
     
     create_info_box(
-        "📖 <b>How to choose the right model:</b><br><br>"
+        "<b>How to choose the right model:</b><br><br>"
         "Consider these factors:<br>"
         "• <b>Language:</b> English only or multilingual?<br>"
         "• <b>Hardware:</b> GPU available or CPU only?<br>"
@@ -430,27 +433,27 @@ with tab1:
     st.dataframe(model_df, width="stretch", hide_index=True)
     
     # Quick recommendations
-    st.markdown("#### 💡 Quick Recommendations:")
+    st.markdown("#### Quick Recommendations:")
     
     rec_col1, rec_col2 = st.columns(2)
     
     with rec_col1:
         st.markdown("**For GPU Training:**")
-        st.markdown("- 🏆 Best: `xlm-roberta-base` (highest accuracy)")
-        st.markdown("- ⚡ Fast: `roberta-base` (English only)")
+        st.markdown("- Best: `xlm-roberta-base` (highest accuracy)")
+        st.markdown("- Fast: `roberta-base` (English only)")
         
     with rec_col2:
         st.markdown("**For CPU Training:**")
-        st.markdown("- 🎯 Recommended: `distilbert-base-multilingual-cased`")
-        st.markdown("- 💨 Fastest training and good performance")
+        st.markdown("- Recommended: `distilbert-base-multilingual-cased`")
+        st.markdown("- Fastest training and good performance")
     
     st.markdown("---")
     
     # ===== Model Download =====
-    st.markdown("### 📥 4. Download Models")
+    st.markdown("### 4. Download Models")
     
     create_info_box(
-        "⬇️ <b>Download models before training:</b><br>"
+        "<b>Download models before training:</b><br>"
         "Models will be downloaded to the <code>models/</code> directory. "
         "This may take several minutes depending on your internet connection.",
         "info"
@@ -466,7 +469,7 @@ with tab1:
     
     col1, col2 = st.columns([3, 1])
     with col2:
-        download_btn = st.button("⬇️ Download Selected", width="stretch", type="primary", disabled=len(selected_models) == 0)
+        download_btn = st.button("Download Selected", width="stretch", type="primary", disabled=len(selected_models) == 0)
     
     if download_btn:
         progress_bar = st.progress(0)
@@ -488,16 +491,16 @@ with tab1:
                 st.error(f"Failed to download {model_id}: {message}")
         
         progress_bar.progress(1.0)
-        status_text.text("✅ Download complete!")
+        status_text.text("Download complete!")
         time.sleep(1)
         st.rerun()
     
     # Show downloaded models
     if st.session_state.models_downloaded:
-        st.markdown("#### ✅ Downloaded Models:")
+        st.markdown("#### Downloaded Models:")
         for model_id in st.session_state.models_downloaded:
             model_info = system_checker.get_model_info(model_id)
-            st.success(f"📦 {MODEL_ARCHITECTURES[model_id]['name']} - {model_info['size_mb']:.0f} MB")
+            st.success(f"{MODEL_ARCHITECTURES[model_id]['name']} - {model_info['size_mb']:.0f} MB")
     
     st.markdown("---")
     
@@ -510,35 +513,35 @@ with tab1:
     )
     
     if can_proceed:
-        if st.button("✅ Prerequisites Complete - Proceed to Data Upload", width="stretch", type="primary"):
+        if st.button("Prerequisites Complete - Proceed to Data Upload", width="stretch", type="primary"):
             st.session_state.prerequisites_checked = True
             add_log("Prerequisites check completed successfully")
-            st.success("🎉 All prerequisites satisfied! You can now proceed to upload your data.")
+            st.success("All prerequisites satisfied! You can now proceed to upload your data.")
             time.sleep(1)
             st.rerun()
     else:
         create_info_box(
-            "⏳ <b>Complete all checks above before proceeding:</b><br>"
-            "✓ CUDA Check<br>"
-            "✓ Environment Check (all packages installed)<br>"
-            "✓ Download at least one model",
+            "<b>Complete all checks above before proceeding:</b><br>"
+            "- CUDA Check<br>"
+            "- Environment Check (all packages installed)<br>"
+            "- Download at least one model",
             "warning"
         )
 
 # ==================== TAB 2: Upload Data ====================
 
 with tab2:
-    st.markdown("## 📤 Upload Training Data")
+    st.markdown("## Upload Training Data")
     
     if not st.session_state.prerequisites_checked:
         create_info_box(
-            "⚠️ Please complete the <b>Prerequisites</b> tab first before uploading data.",
+            "Please complete the <b>Prerequisites</b> tab first before uploading data.",
             "warning"
         )
         st.stop()
     
     create_info_box(
-        "📄 <b>Data Format Requirements:</b><br>"
+        "<b>Data Format Requirements:</b><br>"
         "• CSV file with at least two columns: text and label<br>"
         "• Text column: Contains the text samples to classify<br>"
         "• Label column: Contains the class labels (0/1 for binary, or class names for multi-class)<br>"
@@ -559,17 +562,17 @@ with tab2:
             df = pd.read_csv(uploaded_file)
             st.session_state.uploaded_data = df
             
-            st.success(f"✅ Uploaded {len(df)} samples")
+            st.success(f"Uploaded {len(df)} samples")
             
             # Validate data
             validator = DataValidator()
             is_valid, message = validator.validate_dataframe(df)
             
             if is_valid:
-                st.success(f"✅ Data validation passed: {message}")
+                st.success(f"Data validation passed: {message}")
                 
                 # Show data preview
-                st.markdown("### 📊 Data Preview")
+                st.markdown("### Data Preview")
                 st.dataframe(df.head(10), width="stretch")
                 
                 # Show statistics
@@ -579,7 +582,7 @@ with tab2:
                 col3.metric("Text Columns", len([c for c in df.columns if df[c].dtype == 'object']))
                 
                 # Label distribution
-                st.markdown("### 📈 Label Distribution")
+                st.markdown("### Label Distribution")
                 label_counts = df['label'].value_counts()
                 fig = px.bar(
                     x=label_counts.index.astype(str),
@@ -593,12 +596,12 @@ with tab2:
                 if st.session_state.classification_type == ClassificationType.MULTICLASS:
                     num_classes = df['label'].nunique()
                     st.session_state.config.num_labels = num_classes
-                    st.info(f"ℹ️ Detected {num_classes} classes for multi-class classification")
+                    st.info(f"Detected {num_classes} classes for multi-class classification")
                 
                 add_log(f"Uploaded data with {len(df)} samples and {df['label'].nunique()} labels")
                 
             else:
-                st.error(f"❌ Data validation failed: {message}")
+                st.error(f"Data validation failed: {message}")
                 
         except Exception as e:
             st.error(f"Error reading file: {str(e)}")
@@ -606,28 +609,28 @@ with tab2:
 # ==================== TAB 3: Configure Training ====================
 
 with tab3:
-    st.markdown("## ⚙️ Configure Training Parameters")
+    st.markdown("## Configure Training Parameters")
     
     if st.session_state.uploaded_data is None:
         create_info_box(
-            "⚠️ Please upload your data in the <b>Upload Data</b> tab first.",
+            "Please upload your data in the <b>Upload Data</b> tab first.",
             "warning"
         )
         st.stop()
     
     create_info_box(
-        "🎛️ <b>Configure your training settings:</b><br>"
+        "<b>Configure your training settings:</b><br>"
         "Adjust the parameters below based on your needs. Hover over ⓘ for explanations.",
         "info"
     )
     
     # Model selection
-    st.markdown("### 🤖 Model Selection")
+    st.markdown("### Model Selection")
     
     available_models = list(st.session_state.models_downloaded)
     
     if not available_models:
-        st.error("❌ No models downloaded. Please download models in the Prerequisites tab.")
+        st.error("No models downloaded. Please download models in the Prerequisites tab.")
         st.stop()
     
     selected_model = st.selectbox(
@@ -652,7 +655,7 @@ with tab3:
     st.markdown("---")
     
     # Training parameters
-    st.markdown("### 🎯 Training Parameters")
+    st.markdown("### Training Parameters")
     
     col1, col2 = st.columns(2)
     
@@ -722,7 +725,7 @@ with tab3:
     st.markdown("---")
     
     # Show configuration summary
-    st.markdown("### 📋 Configuration Summary")
+    st.markdown("### Configuration Summary")
     
     config_summary = {
         "Classification Type": "Binary" if st.session_state.classification_type == ClassificationType.BINARY else "Multi-class",
@@ -744,29 +747,29 @@ with tab3:
 # ==================== TAB 4: Train Model ====================
 
 with tab4:
-    st.markdown("## 🎯 Train Your Model")
+    st.markdown("## Train Your Model")
     
     if st.session_state.uploaded_data is None:
         create_info_box(
-            "⚠️ Please complete previous steps first.",
+            "Please complete previous steps first.",
             "warning"
         )
         st.stop()
     
     if not st.session_state.training_started:
         create_info_box(
-            "🚀 <b>Ready to train!</b><br>"
+            "<b>Ready to train!</b><br>"
             f"Your {MODEL_ARCHITECTURES[st.session_state.selected_model]['name']} model will be trained on {len(st.session_state.uploaded_data)} samples "
             f"for {st.session_state.config.num_epochs} epochs.",
             "info"
         )
         
-        if st.button("🚀 Start Training", type="primary", width="stretch"):
+        if st.button("Start Training", type="primary", width="stretch"):
             st.session_state.training_started = True
             st.rerun()
     
     if st.session_state.training_started and not st.session_state.training_completed:
-        st.markdown("### ⏳ Training in Progress...")
+        st.markdown("### Training in Progress...")
         
         # Progress display
         progress_bar = st.progress(0)
@@ -778,6 +781,14 @@ with tab4:
             status_text.text("Preparing data...")
             df = st.session_state.uploaded_data
             
+            # Encode labels to integers
+            label_encoder = LabelEncoder()
+            encoded_labels = label_encoder.fit_transform(df['label'])
+            
+            # Store label encoder for later use
+            st.session_state.label_encoder = label_encoder
+            st.session_state.label_classes = label_encoder.classes_.tolist()
+            
             # Initialize trainer with absolute path
             import os
             st.session_state.config.output_dir = os.path.abspath("trained_models")
@@ -804,7 +815,7 @@ with tab4:
             # Train model
             result = trainer.train(
                 texts=df['text'].tolist(),
-                labels=df['label'].tolist(),
+                labels=encoded_labels.tolist(),
                 progress_callback=progress_callback
             )
             
@@ -823,12 +834,12 @@ with tab4:
             progress_bar.progress(1.0)
             status_text.empty()
             
-            st.success("🎉 Training completed successfully!")
+            st.success("Training completed successfully!")
             add_log(f"Training completed successfully. Model saved to: {result.model_path}")
             
             # Show final metrics
             if result.final_metrics:
-                st.markdown("### 📊 Final Training Metrics")
+                st.markdown("### Final Training Metrics")
                 metrics = result.final_metrics.to_dict()
                 
                 col1, col2, col3, col4 = st.columns(4)
@@ -843,49 +854,49 @@ with tab4:
         except Exception as e:
             import traceback
             error_details = traceback.format_exc()
-            st.error(f"❌ Training failed: {str(e)}")
-            with st.expander("🔍 Error Details"):
+            st.error(f"Training failed: {str(e)}")
+            with st.expander("Error Details"):
                 st.code(error_details)
             st.session_state.training_started = False
             add_log(f"Training failed: {str(e)}")
     
     if st.session_state.training_completed:
-        st.success("✅ Training completed!")
+        st.success("Training completed!")
         
-        model_path_display = st.session_state.model_path if st.session_state.model_path else "⚠️ Path not available"
+        model_path_display = st.session_state.model_path if st.session_state.model_path else "Path not available"
         
         create_info_box(
-            f"🎉 <b>Model trained successfully!</b><br>"
+            f"<b>Model trained successfully!</b><br>"
             f"Model saved to: <code>{model_path_display}</code><br>"
             "Proceed to the <b>Evaluate Model</b> tab to analyze performance.",
             "success" if st.session_state.model_path else "warning"
         )
         
         # Show training logs
-        with st.expander("📜 View Training Logs"):
+        with st.expander("View Training Logs"):
             for log in st.session_state.training_logs[-20:]:  # Show last 20 logs
                 st.text(log)
 
 # ==================== TAB 5: Evaluate Model ====================
 
 with tab5:
-    st.markdown("## 📊 Evaluate Model Performance")
+    st.markdown("## Evaluate Model Performance")
     
     if not st.session_state.training_completed:
         create_info_box(
-            "⚠️ Please train a model first in the <b>Train Model</b> tab.",
+            "Please train a model first in the <b>Train Model</b> tab.",
             "warning"
         )
         st.stop()
     
     create_info_box(
-        "📈 <b>Model Evaluation:</b><br>"
+        "<b>Model Evaluation:</b><br>"
         "Analyze your model's performance with detailed metrics and visualizations.",
         "info"
     )
     
     if st.session_state.evaluation_results is None:
-        if st.button("🔍 Evaluate Model", type="primary", width="stretch"):
+        if st.button("Evaluate Model", type="primary", width="stretch"):
             with st.spinner("Evaluating model..."):
                 try:
                     # Initialize evaluator
@@ -899,10 +910,19 @@ with tab5:
                     test_size = int(len(df) * st.session_state.config.validation_split)
                     test_df = df.tail(test_size)
                     
+                    # Encode labels using the same encoder from training
+                    if 'label_encoder' in st.session_state:
+                        test_labels_encoded = st.session_state.label_encoder.transform(test_df['label']).tolist()
+                    else:
+                        # Fallback: create new encoder if not available
+                        from sklearn.preprocessing import LabelEncoder
+                        label_encoder = LabelEncoder()
+                        test_labels_encoded = label_encoder.fit_transform(test_df['label']).tolist()
+                    
                     # Evaluate
                     results = evaluator.evaluate(
                         texts=test_df['text'].tolist(),
-                        true_labels=test_df['label'].tolist(),
+                        true_labels=test_labels_encoded,
                         batch_size=st.session_state.config.batch_size
                     )
                     
@@ -917,7 +937,7 @@ with tab5:
         results = st.session_state.evaluation_results
         
         # Overall metrics
-        st.markdown("### 📊 Overall Metrics")
+        st.markdown("### Overall Metrics")
         
         col1, col2, col3, col4 = st.columns(4)
         col1.metric("Accuracy", f"{results['accuracy']:.2%}")
@@ -928,7 +948,7 @@ with tab5:
         st.markdown("---")
         
         # Confusion Matrix
-        st.markdown("### 🔢 Confusion Matrix")
+        st.markdown("### Confusion Matrix")
         
         if 'confusion_matrix' in results:
             cm = results['confusion_matrix']
@@ -956,7 +976,7 @@ with tab5:
         st.markdown("---")
         
         # Classification Report
-        st.markdown("### 📋 Detailed Classification Report")
+        st.markdown("### Detailed Classification Report")
         
         if 'classification_report' in results:
             report = results['classification_report']
@@ -964,9 +984,9 @@ with tab5:
         
         # Download results
         st.markdown("---")
-        st.markdown("### 💾 Download Results")
+        st.markdown("### Download Results")
         
-        if st.button("📥 Download Evaluation Report", width="stretch"):
+        if st.button("Download Evaluation Report", width="stretch"):
             # Create downloadable report
             report_text = f"""
 MLOps Training Platform - Evaluation Report
@@ -993,7 +1013,7 @@ Training Configuration:
 """
             
             st.download_button(
-                label="📄 Download Text Report",
+                label="Download Text Report",
                 data=report_text,
                 file_name=f"evaluation_report_{datetime.now().strftime('%Y%m%d_%H%M%S')}.txt",
                 mime="text/plain"