问题描述 / Problem Description
初始化数据库阶段，报出prompt length 过长错误。

复现问题的步骤 / Steps to Reproduce

1. 执行
   cd libs/chatchat-server
   python chatchat/cli.py kb --recreate-vs
2. 滚动到
   正在将 samples/大模型指令对齐训练原理.md 添加到向量库，共包含3条文档
   Retrying chatchat.server.localai_embeddings.embed_with_retry.._embed_with_retry in 4.0 seconds as it raised InternalServerError: Error code: 500 - {'detail': '[address=0.0.0.0:44205, pid=3320924] The decoder prompt (length 515) is longer than the maximum model length of 512. Make sure that max_model_len is no smaller than the number of text tokens.'}.
   Retrying chatchat.server.localai_embeddings.embed_with_retry.._embed_with_retry in 4.0 seconds as it raised InternalServerError: Error code: 500 - {'detail': '[address=0.0.0.0:44205, pid=3320924] The decoder prompt (length 515) is longer than the maximum model length of 512. Make sure that max_model_len is no smaller than the number of text tokens.'}.
3. 问题出现
   2025-12-28 05:03:42.407 | ERROR | chatchat.server.utils:run_in_thread_pool:812 - error in sub thread: Error code: 500 - {'detail': '[address=0.0.0.0:44205, pid=3320924] The decoder prompt (length 515) is longer than the maximum model length of 512. Make sure that max_model_len is no smaller than the number of text tokens.'}
   2025-12-28 05:03:42.409 | ERROR | chatchat.init_database:worker:61 - texts and metadatas expected to be equal length but len(texts)=2 and len(metadatas)=3

环境信息 / Environment Information

• Langchain-Chatchat 版本 / commit 号：0.3.1.3
• 部署方式（pypi 安装 / 源码部署 / docker 部署）：源码部署
• 使用的模型推理框架（Xinference / Ollama / OpenAI API 等）：Xinference
• 使用的 LLM 模型（GLM-4-9B / Qwen2-7B-Instruct 等）：qwen3
• 使用的 Embedding 模型（bge-large-zh-v1.5 / m3e-base 等）：bge-large-zh-v1.5
• 使用的向量库类型 (faiss / milvus / pg_vector 等)： faiss
• 操作系统及版本 / Operating system and version: Ubuntu 22.04
• Python 版本 / Python version: 3.10
• 推理使用的硬件（GPU / CPU / MPS / NPU 等） : GPU