Increase context window to 2048

app.py

@@ -30,7 +30,7 @@ pipe = pipeline(
     "text-generation",
     model=model,
     tokenizer=tokenizer,
-    max_length=512,
+    max_length=2048,  # LLaMA default context window
     temperature=0.7,
     top_p=0.95,
     repetition_penalty=1.15