From 66a551217eac4f888d5f6ad2f53d92a72cbfebbc Mon Sep 17 00:00:00 2001 From: ViperEkura <3081035982@qq.com> Date: Fri, 7 Nov 2025 07:24:00 +0800 Subject: [PATCH] =?UTF-8?q?refactor(generator):=20=20=E4=BC=98=E5=8C=96?= =?UTF-8?q?=E7=94=9F=E6=88=90=E9=80=BB=E8=BE=91?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- khaosz/inference/generator.py | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-) diff --git a/khaosz/inference/generator.py b/khaosz/inference/generator.py index 8e19c81..1402a46 100644 --- a/khaosz/inference/generator.py +++ b/khaosz/inference/generator.py @@ -118,7 +118,6 @@ class ChatGenerator(GeneratorCore): ids = self.tokenizer.encode(build_prompt(query, history)) input_ids = torch.tensor([ids], device=device, dtype=torch.long) - cpy_history = history.copy() start_cache_pos = len(ids) cur_cache_pos = 0 @@ -132,9 +131,8 @@ class ChatGenerator(GeneratorCore): ) response = self.tokenizer.decode(ids[start_cache_pos:]) - cpy_history.append((query, response)) - return response, cpy_history + return response class StreamGenerator(GeneratorCore): @@ -278,7 +276,7 @@ class RetrievalGenerator(GeneratorCore): history = [] retrieved = "\n".join([f"{idx + 1}. {key}" for idx, key in enumerate(retrieved)]) if retrieved else "" - retrieved_query = f"{retrieved}\n\n根据以上内容回答: {query}" if retrieved else query + retrieved_query = f"{retrieved}\n\n{query}" if retrieved else query parameter = ModelParameter(self.model, self.tokenizer, self.config) return ChatGenerator(parameter).generate(