Update app.py

app.py

@@ -66,7 +66,8 @@ def generate(
     generation_kwargs = dict(input_ids=torch.tensor([input_ids]).to(llm.device), do_sample=True,
                              max_new_tokens=512, temperature=0.5, top_p=0.85, top_k=50, repetition_penalty=1.05)
     llm_result = llm.generate(**generation_kwargs)
-    llm_result = BOT_PREFIX + tokenizer.decode(llm_result.cpu()[0], skip_special_tokens=True)
+    llm_result = llm_result.cpu()[0][len(input_ids):]
+    llm_result = BOT_PREFIX + tokenizer.decode(llm_result, skip_special_tokens=True)
     print(llm_result)
     expanded_description = json.loads(llm_result)["expanded_description"]
     print(expanded_description)