https://github.com/hscspring/hnlp
Humanly DeepLearning for NLP, on top of Transformers.
https://github.com/hscspring/hnlp
Last synced: about 1 year ago
JSON representation
Humanly DeepLearning for NLP, on top of Transformers.
- Host: GitHub
- URL: https://github.com/hscspring/hnlp
- Owner: hscspring
- License: mit
- Created: 2020-04-26T01:50:08.000Z (about 6 years ago)
- Default Branch: master
- Last Pushed: 2022-10-13T09:23:18.000Z (over 3 years ago)
- Last Synced: 2025-02-16T19:48:37.905Z (over 1 year ago)
- Language: Python
- Homepage:
- Size: 11.6 MB
- Stars: 2
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# hnlp
Humanly DeepLearning NLP, quickly demo and easily play.
- Quickly setup a production, from 0 to 90%+, gain the maximum marginal income.
- Quickly evaluate a new idea based on the existing datasets and models.
```python
data = Corpus() >> Preprocessor() >> Tokenizer() >> DataManager()
# x_train, y_train = data.run("/path/to/file")
pipe = (
Corpus() >>
Preprocessor() >>
Tokenizer() >>
DataManager() >>
Pretrained() >>
Model(config) >>
Task(optimizer, loss_fn, metric)
)
pipe.train("/path/to/train_file")
pipe.test("/path/to/test_file")
```
## Test
```bash
$ python -m pytest
```
- [x] 自动寻找 LearningRate