diff --git a/lightrag/llm.py b/lightrag/llm.py
index eaaa2b75..bb0d6063 100644
--- a/lightrag/llm.py
+++ b/lightrag/llm.py
@@ -282,10 +282,11 @@ async def hf_model_if_cache(
     input_ids = hf_tokenizer(
         input_prompt, return_tensors="pt", padding=True, truncation=True
     ).to("cuda")
+    inputs = {k: v.to(hf_model.device) for k, v in input_ids.items()}
     output = hf_model.generate(
-        **input_ids, max_new_tokens=200, num_return_sequences=1, early_stopping=True
+        **input_ids, max_new_tokens=512, num_return_sequences=1, early_stopping=True
     )
-    response_text = hf_tokenizer.decode(output[0], skip_special_tokens=True)
+    response_text = hf_tokenizer.decode(output[0][len(inputs["input_ids"][0]):], skip_special_tokens=True)
     if hashing_kv is not None:
         await hashing_kv.upsert({args_hash: {"return": response_text, "model": model}})
     return response_text