Fixed bugs in main.ipynb & added a requirements.txt

Author: lianghsun

.gitignore

@@ -4,3 +4,4 @@ logs/
 utils/
 model/
 .ipynb_checkpoints/
+*.npy

README.md

@@ -1,19 +1,16 @@
-# 問題背景
-目前全球新藥開發領域正運用AI技術進行候選藥物篩選及結構設計，利用數據驅動（Data-Driven）的思維引導與輔助頂尖藥物設計科學家之藥物開發工作，將是前瞻性藥物開發新技術之重大突破與轉折契機。藥物發展極為複雜且極為耗時，包含：（1）標的選擇與確認（2）化合物篩選與主要結構優化（3）臨床前試驗（4）臨床試驗。傳統藥物發展程序為：當化學結構之作用確認後，輔以藥物專家或化學家定義的Rule-Based規則（in silico、QSAR、logP等）與細胞功能測試成為候選藥物，接續進行動物體內試驗與臨床試驗，用以驗證分子功效與生物毒性作用。但並非所有的藥物都吻合Rule-Based，且美國發展一種新藥平均需耗費26億美金與歷時12年。如何減少研發成本與縮短藥物開發時程，為產業最大挑戰與最緊急之問題，我們希望能透過 AI 模型去學習如何探索新藥開發
+# Prerequisites
+- Install the required packages by running the following command:
+  ```bash
+  pip install -r requirements.txt
+  ```
 
-## Prerequisites
-- Install RDKit from https://www.rdkit.org/
+# Dataset
+- Download the SMILES dataset from [Kaggle ZINC 250k](https://www.kaggle.com/datasets/lianghsunhuang/zinc-250k). Change the file extension to `.smi` and remove the header row.
 
-## Dataset
-- Download SMILES from https://zinc.docking.org/tranches/home/
+# Running the Main Script
+We have provided a token file for this project, so you can skip the `Load Training Data` section and proceed with running the rest of the code.
 
-## Run Main
-We have prepared token file for this project, so please skip `Load Training Data` section and run all of rest code.
-Due to VAE, sometimes it might not generate new compound (sampling problem), please run more times to get valid compound.
+ > Note: Due to the nature of Variational Autoencoders (VAE), there might be instances where new compounds are not generated (sampling problem). If this happens, please run the code multiple times to obtain a valid compound.
 
-## Oh, you have own dataset?
-1. Please create a folder named `dataset`, and place your `.smi` files in this folder.
-2. run `preprocess` to create new token file for your dataset, then run `main` to train your model
-
-## last..
-Enjoy your new compound!
+# License
+This project is licensed under the Apache License 2.0 - see the [LICENSE](LICENSE) file for details.