https://github.com/timcsy/dsai-hw2
https://github.com/timcsy/dsai-hw2
Last synced: 2 months ago
JSON representation
- Host: GitHub
- URL: https://github.com/timcsy/dsai-hw2
- Owner: timcsy
- License: mit
- Created: 2022-04-26T15:51:19.000Z (about 3 years ago)
- Default Branch: master
- Last Pushed: 2022-05-16T03:42:23.000Z (about 3 years ago)
- Last Synced: 2025-01-18T23:15:02.711Z (4 months ago)
- Language: Jupyter Notebook
- Size: 3.58 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# DSAI-HW2
安裝
---
```
pip install -r requirements.txt
```
執行
---
```
python trader.py --training training_data.csv --testing testing_data.csv --output output.csv
```
心得
---
- 參考：[Stock predictions with state-of-the-art Transformer and Time Embeddings](https://towardsdatascience.com/stock-predictions-with-state-of-the-art-transformer-and-time-embeddings-3a4485237de6)
- 使用 Transformer 來預測收益率曲線
- 在訓練的時候常受到初始 weight 的影響，在一些時候訓練不太起來，但是在可以訓練起來的狀況下要到大約 30 epochs 才會開始收斂
- Better use GPU to run the model