This is an environment record I created during my thesis writing process to reproduce other researchers' work. Whether it can be successfully reproduced in other experiments is not guaranteed, but this environment has been confirmed to work for the following two papers:
Since TensorFlow has not been updated for a long time, deep learning papers using TF should theoretically all run in this environment.
I mainly use PyTorch and am not familiar with TF. If you run into issues, using an LLM to debug should do the trick. The full environment has been uploaded to the cloud as a backup: 📦 Google Drive Download
After downloading, you can place it directly into your conda env
For any other issues, I'm afraid I cannot help. Hope this mess of a record is useful to you — expect some detours along the way, good luck XD
| Item | Description |
|---|---|
| OS | Windows |
| Package Manager | Conda (required) |
| Python Framework | TensorFlow 2.10.0 |
| GPU | 5070 Ti |
| GPU2 | 3090 |
| GPU3 | 5090 |
這是我在撰寫論文過程中，為了復現其他學者研究所建立的環境紀錄。 能否在其他實驗上順利複現並不保證，但此環境已確認可在以下兩篇論文的實驗中執行：
因 TensorFlow 已久未更新，理論上使用 TF 的深度學習論文應皆可在此環境中執行。
本人主用 PyTorch，對 TF 不熟，若遇到問題，建議透過 LLM 輔助排查，基本上應能迎刃而解。為此已將整包環境上傳至雲端備用：📦 Google Drive 下載
下載後可直接放入 conda env 環境
如有其他問題，恕無法協助，希望這坨紀錄對你有所幫助，歪路可能還是少不了，祝順利 XD
| 項目 | 說明 |
|---|---|
| 作業系統 | Windows |
| 套件管理 | Conda（必須） |
| Python 框架 | TensorFlow 2.10.0 |
| GPU | 5070 Ti |
| GPU2 | 3090 |
| GPU3 | 5090 |
