[Bug]: 使用chatglm3 模型”THUDM/chatglm3-6b“进行python项目分析出现出现cuda问题

### Installation Method | 安装方法与平台

Docker-Compose（Linux）

### Version | 版本

Latest | 最新版

### OS | 操作系统

Docker

### Describe the bug | 简述

模型THUDM/chatglm3-6b
本地加载模型
能正常使用文本对话
但是上传tar包或者zip包进行python项目解析时候出现问题：

使用docker部署
后代报错内容：
gpt_academic_with_chatglm_1 | ../aten/src/ATen/native/cuda/IndexKernel.cu:92: operator(): block: [1103,0,0], thread: [26,0,0] Assertion `-sizes[i] <= index && index < sizes[i] && "index out of bounds"` failed.
该报错大量重复，无其他error报错

### Screen Shot | 有帮助的截图


### Terminal Traceback & Material to Help Reproduce Bugs | 终端traceback（如有） + 帮助我们复现的测试材料样本（如有）

_No response_

Start by reproducing the failure with Docker Compose on Linux using the local THUDM/chatglm3-6b model: text chat succeeds, but uploading a tar or zip Python project triggers the CUDA index-out-of-bounds assertion. Inspect the project-analysis entry point and its model-processing path; done means archive analysis completes without the reported CUDA error and the reproduction is covered by a test or documented verification.

Written by the indexing model from the issue text.