Resize context window to 640, and add provided datas for precise result.

app.py

@@ -15,7 +15,7 @@ headers = {
 }
 
 hf_hub_download(repo_id="LLukas22/gpt4all-lora-quantized-ggjt", filename="ggjt-model.bin", local_dir=".")
-llm = Llama(model_path="./ggjt-model.bin", n_ctx=768)
+llm = Llama(model_path="./ggjt-model.bin", n_ctx=640)
 
 ins = '''### Instruction:
 {question}
@@ -47,8 +47,8 @@ def search_ddg(question: str):
 		else:
 			continue
 
-	for element in results:
-		output_string += element + '\n\n'
+	for step in range(len(results)):
+		output_string += f"{step}. {results[step]} \n"
 
 	return output_string