Update app.py

app.py

@@ -7,23 +7,47 @@ import re
 # Global datasets - load lazily
 math_samples = None
 
-def load_datasets_lazy():
-    """Load datasets only when needed - prevents startup crashes"""
+def load_sample_problems():
+    """Load sample problems from ALL datasets"""
     global math_samples
     if math_samples is not None:
         return math_samples
     
-    print("🔄 Loading datasets...")
+    samples = []
     try:
+        # GSM8K (math problems)
         gsm8k = load_dataset("openai/gsm8k", "main", streaming=True)
-        samples = []
         for i, item in enumerate(gsm8k["train"]):
             samples.append(item["question"])
             if i >= 50:
                 break
-        print(f"✅ Loaded {len(samples)} GSM8K samples")
+        
+        # Fineweb-edu (educational text - extract math-like questions)
+        fw = load_dataset("HuggingFaceFW/fineweb-edu", name="sample-10BT", split="train", streaming=True)
+        fw_count = 0
+        for item in fw:
+            # Filter for math-related content (simple keyword match)
+            if any(word in item['text'].lower() for word in ['math', 'calculate', 'solve', 'derivative', 'integral', 'triangle', 'equation']):
+                samples.append(item['text'][:200] + " (Solve this math problem.)")  # Truncate for brevity
+                fw_count += 1
+                if fw_count >= 20:
+                    break
+        
+        # Ultrachat_200k (chat-like math queries)
+        ds = load_dataset("HuggingFaceH4/ultrachat_200k", streaming=True)
+        ds_count = 0
+        for item in ds:
+            if 'math' in item['messages'][0]['content'].lower() or 'calculate' in item['messages'][0]['content'].lower():
+                user_msg = item['messages'][0]['content']
+                samples.append(user_msg)
+                ds_count += 1
+                if ds_count >= 20:
+                    break
+        
+        print(f"✅ Loaded {len(samples)} samples: GSM8K ({50}), Fineweb-edu ({fw_count}), Ultrachat ({ds_count})")
         math_samples = samples
         return samples
+        
     except Exception as e:
         print(f"⚠️ Dataset error: {e}, using fallback")
         math_samples = [
@@ -31,7 +55,9 @@ def load_datasets_lazy():
             "A triangle has sides of length 5, 12, and 13. What is its area?",
             "If log₂(x) + log₂(x+6) = 4, find the value of x.",
             "Find the limit: lim(x->0) (sin(x)/x)",
-            "Solve the system: x + 2y = 7, 3x - y = 4"
+            "Solve the system: x + 2y = 7, 3x - y = 4",
+            "Calculate the integral of sin(x) from 0 to pi.",
+            "What is the probability of rolling a 6 on a die 3 times in a row?"
         ]
         return math_samples