Update app_gradio.py

app_gradio.py

@@ -6,6 +6,7 @@ from __future__ import annotations
 import io
 import time
 import pandas as pd
+import numpy as np  # <-- Added numpy for percentiles
 import gradio as gr
 from classify import classify_log, classify_csv
 
@@ -397,21 +398,50 @@ def classify_single(source: str, log_message: str):
 def classify_batch(file):
     if file is None:
         return None, "⚠️ Please upload a CSV file."
+        
+    t0 = time.perf_counter() # Start Total Timer
+    
     try:
         output_path, df = classify_csv(file.name, "/tmp/classified_output.csv")
     except ValueError as e:
         return None, f"⚠️ {e}"
     except Exception as e:
         return None, f"❌ Error: {e}"
+        
+    total_time_sec = time.perf_counter() - t0 # End Total Timer
     total = len(df)
+    
     tier_counts  = df["tier_used"].value_counts().to_dict()
     label_counts = df["predicted_label"].value_counts().to_dict()
+    
     tier_lines  = "\n".join(f"  {TIER_COLORS.get(k,'⚪')} {k}: {v} ({v/total:.0%})" for k, v in tier_counts.items())
     label_lines = "\n".join(f"  • {k}: {v}" for k, v in label_counts.items())
+    
+    # Calculate Latencies (Requires 'latency_ms' column in CSV output from classify_csv)
+    if "latency_ms" in df.columns and not df["latency_ms"].empty:
+        latencies = df["latency_ms"].dropna()
+        p50 = np.percentile(latencies, 50)
+        p95 = np.percentile(latencies, 95)
+        p99 = np.percentile(latencies, 99)
+        latency_stats = (
+            f"⏱️ Performance Metrics:\n"
+            f"  • Total Time: {total_time_sec:.2f} s\n"
+            f"  • P50 Latency: {p50:.1f} ms\n"
+            f"  • P95 Latency: {p95:.1f} ms\n"
+            f"  • P99 Latency: {p99:.1f} ms"
+        )
+    else:
+        latency_stats = (
+            f"⏱️ Performance Metrics:\n"
+            f"  • Total Time: {total_time_sec:.2f} s\n"
+            f"  • (Latency stats unavailable: 'latency_ms' not found in output)"
+        )
+    
     stats = (
         f"✅ Classified {total} logs\n\n"
         f"📊 Tier breakdown:\n{tier_lines}\n\n"
-        f"🏷️ Label distribution:\n{label_lines}"
+        f"🏷️ Label distribution:\n{label_lines}\n\n"
+        f"{latency_stats}"
     )
     return output_path, stats
 
@@ -487,8 +517,7 @@ with gr.Blocks(title="LOG CLASSIFICATION SYSTEM") as demo:
         with gr.Tab("📦 BATCH CSV"):
             gr.Markdown("""
 ### Bulk Classification
-Upload a CSV with columns: **`source`**, **`log_message`**  
-Output includes: `predicted_label`, `tier_used`, `confidence`, `latency_ms`
+Upload a CSV with columns: **`source`**, **`log_message`** Output includes: `predicted_label`, `tier_used`, `confidence`, `latency_ms`
 """)
             with gr.Row():
                 with gr.Column():
@@ -496,7 +525,8 @@ Output includes: `predicted_label`, `tier_used`, `confidence`, `latency_ms`
                     batch_btn  = gr.Button("▶  CLASSIFY ALL", variant="primary")
                 with gr.Column():
                     csv_output = gr.File(label="📥 DOWNLOAD RESULTS")
-                    stats_out  = gr.Textbox(label="📊 STATISTICS", lines=12, interactive=False)
+                    # Increased lines from 12 to 16 to fit the new metrics nicely
+                    stats_out  = gr.Textbox(label="📊 STATISTICS", lines=16, interactive=False)
 
             batch_btn.click(
                 fn=classify_batch,
@@ -505,36 +535,4 @@ Output includes: `predicted_label`, `tier_used`, `confidence`, `latency_ms`
             )
 
             gr.Markdown("""
-**Sample CSV format:**
-```
-source,log_message
-ModernCRM,User User123 logged in.
-LegacyCRM,Case escalation for ticket ID 7324 failed.
-BillingSystem,GET /api/v2/invoice HTTP/1.1 status: 500
-```
-""")
-
-        # ── Tab 3: Architecture ───────────────────────────────────────────
-        with gr.Tab("🏗️ ARCHITECTURE"):
-            gr.Markdown("""
-## 3-Tier Hybrid Pipeline
-
-| Tier | Method | Coverage | Latency | Trigger |
-|------|--------|----------|---------|---------|
-| 🟢 **Regex** | Python `re` patterns | ~21% | < 1ms | Fixed patterns |
-| 🔵 **BERT** | `all-MiniLM-L6-v2` + LogReg | ~79% | 20–80ms | High-volume categories |
-| 🟡 **LLM** | HuggingFace Inference API | ~0.3% | 500–2000ms | LegacyCRM + rare patterns |
-
-## Model Performance
-- **Training data**: 2,410 synthetic enterprise logs
-- **Confidence threshold**: 0.5 (below → escalate to LLM)
-- **Source-aware routing**: `LegacyCRM` → LLM directly
-
-## Environment Variables
-| Secret | Purpose |
-|--------|---------|
-| `HF_TOKEN` | LLM inference for LegacyCRM logs |
-""")
-
-if __name__ == "__main__":
-    demo.launch(server_name="0.0.0.0", server_port=7860, theme=THEME, css=CUSTOM_CSS)
+**Sample CSV format:**