Update app.py

app.py

@@ -34,26 +34,12 @@ def build_dataset():
 
     derma_examples = []
     for item in derma:
-        q = item.get('question') or item.get('Question') or item.get('Q')
-        a = item.get('answer') or item.get('Answer') or item.get('A')
+        q = item.get('prompt')  # Mreeb uses 'prompt' instead of 'question'
+        a = item.get('response')
         if q and a:
             derma_examples.append({"domain":"dermatology","context":q,"response":a})
 
-    # ----- MedQuAD subset
-    print("Loading MedQuAD subset...")
-    medquad = load_dataset("pythonafroz/MedQuAD")['train']
-    derma_keywords = ["skin", "eczema", "psoriasis", "dermatitis", "melanoma", "acne", "rash"]
-    medquad_derma = [{"domain":"dermatology",
-                      "context":item['question'],
-                      "response":item['answer']}
-                     for item in medquad if any(k in item['question'].lower() for k in derma_keywords)]
-    random.shuffle(medquad_derma)
-    medquad_derma = medquad_derma[:500]
-
-    dermatology_examples = derma_examples + medquad_derma
-    random.shuffle(dermatology_examples)
-
-    all_examples = general_examples + dermatology_examples
+    all_examples = general_examples + derma_examples
     random.shuffle(all_examples)
 
     # ذخیره به JSONL