Metadata-Version: 2.1
Name: dlk
Version: 0.0.10
Summary: dlk: Deep Learning Kit
Home-page: https://github.com/cstsunfu/dlk
Author: cstsunfu
Author-email: cstsunfu@gmail.com
License: Apache Software License
Platform: UNKNOWN
Requires-Python: >=3.8
Description-Content-Type: text/markdown
License-File: LICENSE


# A Deep Learning Kit

This project is WIP, only provide some basic method, and is not been tested.

```
WARNING: The documents in `read the docs` is not complete( I do not know what I miss.). You can build the docs locally by read the docs/readme.md.
```


## Install

```
pip install dlk

or 
git clong this repo and do

python setup.py install

```
## What is this project do?

* Provide a templete for deep learning (especially for nlp) training and deploy.
* Provide parameters search.
* Provide basic architecture search.
* Provide some basic modules and models.
* Provide basic deploy method.

## More Feature is Comming

- [ ] Add more documents.
- [ ] Distill structure.
- [ ] Adv training.
- [ ] Add disable tokenizer post process.

- [ ] Predict

    - [X] Complete the main predict code.
    - [ ] Test.
    - [ ] Convert to TorchScript.
    - [ ] Convert to ONNXRT

* [X] When the high config change the _name of the basic config, the base config should be coverd.

- [ ] One `optimizer` different para groups use different `scheduler`s.
  
    - [ ] Ref [diff_schedule](https://pytorch.org/docs/stable/generated/torch.optim.lr_scheduler.CyclicLR.html#torch.optim.lr_scheduler.CyclicLR)
    - [ ] Add loss schedule, get best checkpoint by loss.

- [ ] Support LightGBM, it's maybe not necessary? Will split to another package.
* [ ] Make CRF and more modules which uses the op like `for` and `if` to be scriptable（`for`，`if`）
* [ ] Validating LSTM module is scriptable or not?
* [ ] Add UnitTest


