Update app.py

app.py

@@ -72,28 +72,42 @@ def summarize_text(text: str):
     output = summarizer(prompt, max_length=256, do_sample=False)
     return output[0]["generated_text"].strip()
 
+async def generate_random_query():
+    """Use your main model to invent a search query."""
+    prompt = (
+        "Generate one short, interesting, real-world question "
+        "about technology, startups, AI, or science that someone might Google. "
+        "Output only the query, nothing else."
+    )
+    output = generator(prompt, max_new_tokens=20, do_sample=True, temperature=0.9, truncation=True)
+    query = output[0]["generated_text"].strip()
+    # Safety: strip weird trailing punctuation
+    return query.split("\n")[0].strip("?!. ")
+
 async def update_knowledge_graph_periodically():
-    queries = [
-        "latest tech startup news",
-        "AI breakthroughs 2025",
-        "funding trends in tech startups",
-        "popular programming languages 2025",
-        "open source AI models"
-    ]
     while True:
-        query = random.choice(queries)
-        print(f"[KG Updater] Searching DuckDuckGo for query: {query}")
         try:
+            query = await generate_random_query()
+            print(f"[KG Updater] Searching DuckDuckGo for query: {query}")
+
             ddg_data = await fetch_ddg_search(query)
             cleaned = clean_ddg_text(ddg_data)
-            if not cleaned:
-                cleaned = "No useful info found."
-            print(f"[KG Updater] DuckDuckGo cleaned text length: {len(cleaned)}")
 
-            # Summarize using your instruct model in a thread (blocking)
+            if not cleaned or len(cleaned) < 50:
+                print("[KG Updater] Too little info found, retrying with new query...")
+                continue  # Skip saving, go to next loop
+
+            # Summarize with instruct model
             loop = asyncio.get_event_loop()
-            summary = await loop.run_in_executor(None, summarize_text, cleaned)
-            print(f"[KG Updater] Summary length: {len(summary)}")
+            summary = await loop.run_in_executor(
+                None,
+                lambda: summarizer(
+                    f"Summarize this info concisely:\n{cleaned}",
+                    max_length=256,
+                    truncation=True,
+                    do_sample=False
+                )[0]["generated_text"].strip()
+            )
 
             knowledge_graph[query] = {
                 "raw_text": cleaned,
@@ -105,7 +119,8 @@ async def update_knowledge_graph_periodically():
         except Exception as e:
             print(f"[KG Updater] Error: {e}")
 
-        await asyncio.sleep(60)
+        await asyncio.sleep(60)  # Run forever, every 60s
+
 
 @app.on_event("startup")
 async def startup_event():