Update app.py

app.py

@@ -7,18 +7,16 @@ from transformers import AutoTokenizer, AutoModelForCausalLM
 
 torch.set_num_threads(1)
 
-# ─────────────────────
+# ─────────────────────────
 # MODEL
-# ─────────────────────
+# ─────────────────────────
 BASE_MODEL = "TinyLlama/TinyLlama-1.1B-Chat-v1.0"
 
 print("Loading model...")
 
 model = AutoModelForCausalLM.from_pretrained(
     BASE_MODEL,
-    device_map="cpu",
-    torch_dtype=torch.float32,
-    low_cpu_mem_usage=True,
+    torch_dtype=torch.float32
 )
 
 tokenizer = AutoTokenizer.from_pretrained(BASE_MODEL)
@@ -26,16 +24,16 @@ tokenizer = AutoTokenizer.from_pretrained(BASE_MODEL)
 model.eval()
 print("Model ready")
 
-# ─────────────────────
+# ─────────────────────────
 # GENERATION
-# ─────────────────────
+# ─────────────────────────
 def generate_sql(question):
     if not question.strip():
         return "Enter SQL question."
 
     prompt = f"""
 You are a SQL expert.
-Convert the user request into SQL query only.
+Return ONLY SQL query.
 
 User: {question}
 SQL:
@@ -56,15 +54,15 @@ SQL:
 
     return text.split("SQL:")[-1].strip()
 
-# ─────────────────────
+# ─────────────────────────
 # UI
-# ─────────────────────
+# ─────────────────────────
 demo = gr.Interface(
     fn=generate_sql,
     inputs=gr.Textbox(lines=3, label="SQL Question"),
     outputs=gr.Textbox(lines=8, label="Generated SQL"),
     title="SQL Generator (Portfolio Demo)",
-    description="Fast CPU model for portfolio demo.",
+    description="Fast CPU AI SQL generator.",
     examples=[
         ["Find duplicate emails in users table"],
         ["Top 5 highest paid employees"],