# AdaSeq: An All-in-One Library for Developing State-of-the-Art Sequence Understanding Models
<div align="center">
[](./LICENSE)
[](https://modelscope.cn/)
[](https://github.com/modelscope/AdaSeq/issues)
[](https://github.com/modelscope/AdaSeq/stargazers)
[](https://pypi.org/project/adaseq)
[](./CONTRIBUTING.md)
</div>
<div align="center">
English | [简体中文](./README_zh.md)
</div>
## Introduction
***AdaSeq*** (**A**libaba **D**amo **A**cademy **Seq**uence Understanding Toolkit) is an easy-to-use all-in-one library, built on [ModelScope](https://modelscope.cn/home), that allows researchers and developers to train custom models for sequence understanding tasks, including part-of-speech tagging (POS Tagging), chunking, named entity recognition (NER), entity typing, relation extraction (RE), etc.
<details open>
<summary>🌟 <b>Features:</b></summary>
- **Plentiful Models**:
AdaSeq provide plenty of cutting-edge models, training methods and useful toolkits for sequence understanding tasks.
- **State-of-the-Art**:
Our aim to develop the best implementation, which can beat many off-the-shelf frameworks on performance.
- **Easy-to-Use**:
One line of command is all you need to obtain the best model.
- **Extensible**:
It's easy to register a module, or build a customized sequence understanding model by assembling the predefined modules.
</details>
⚠️**Notice:** This project is under quick development. This means some interfaces could be changed in the future.
## 📢 What's New
- 2022-07: [SemEval 2023] Our U-RaNER paper won [Best Paper Award](https://semeval.github.io/SemEval2023/awards)!
- 2022-03: [SemEval 2023] Our U-RaNER won ***1st place in 9 tracks*** at [SemEval 2023 Task2](https://multiconer.github.io/results): Multilingual Complex Named Entity Recognition! [Model introduction and source code can be found here](./examples/U-RaNER).
- 2022-12: [[EMNLP 2022] Retrieval-augmented Multimodal Entity Understanding Model (MoRe)](./examples/MoRe)
- 2022-11: [[EMNLP 2022] Ultra-Fine Entity Typing Model (NPCRF)](./examples/NPCRF)
- 2022-11: [[EMNLP 2022] Unsupervised Boundary-Aware Language Model (BABERT)](./examples/babert)
## ⚡ Quick Experience
You can try out our models via online demos built on ModelScope:
[[English NER]](https://modelscope.cn/models/damo/nlp_raner_named-entity-recognition_english-large-news/summary)
[[Chinese NER]](https://modelscope.cn/models/damo/nlp_raner_named-entity-recognition_chinese-base-news/summary)
[[CWS]](https://modelscope.cn/models/damo/nlp_structbert_word-segmentation_chinese-base/summary)
More tasks, more languages, more domains: All modelcards we released can be found in this page [Modelcards](./docs/modelcards.md).
## 🛠️ Model Zoo
<details open>
<summary><b>Supported models:</b></summary>
- [Transformer-based CRF](./examples/bert_crf)
- [Partial CRF](./examples/partial_bert_crf)
- [Retrieval Augmented NER](./examples/RaNER)
- [Biaffine NER](./examples/biaffine_ner)
- [Global-Pointer](./examples/global_pointer)
- [Multi-label Entity Typing](./examples/entity_typing)
- ...
</details>
## 💾 Dataset Zoo
We collected many datasets for sequence understanding tasks. All can be found in this page [Datasets](./docs/datasets.md).
## 📦 Installation
AdaSeq project is based on `Python >= 3.7`, `PyTorch >= 1.8` and `ModelScope >= 1.4`. We assure that AdaSeq can run smoothly when `ModelScope == 1.9.5`.
- installation via pip：
```
pip install adaseq
```
- installation from source：
```
git clone https://github.com/modelscope/adaseq.git
cd adaseq
pip install -r requirements.txt -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo.html
```
### Verify the Installation
To verify whether AdaSeq is installed properly, we provide a demo config for training a model (the demo config will be automatically downloaded).
```
adaseq train -c demo.yaml
```
You will see the training logs on your terminal. Once the training is done, the results on test set will be printed: `test: {"precision": xxx, "recall": xxx, "f1": xxx}`. A folder `experiments/toy_msra/` will be generated to save all experimental results and model checkpoints.
## 📖 Tutorials
- [Quick Start](./docs/tutorials/quick_start.md)
- Basics
- [Learning about Configs](./docs/tutorials/learning_about_configs.md)
- [Customizing Dataset](./docs/tutorials/customizing_dataset.md)
- [TODO] Common Architectures
- [TODO] Useful Hooks
- [Hyperparameter Optimization](./docs/tutorials/hyperparameter_optimization.md)
- [Training with Multiple GPUs](./docs/tutorials/training_with_multiple_gpus.md)
- Best Practice
- [Training a Model with Custom Dataset](./docs/tutorials/training_a_model.md)
- [Reproducing Results in Published Papers](./docs/tutorials/reproducing_papers.md)
- [TODO] Uploading Saved Model to ModelScope
- [TODO] Customizing your Model
- [TODO] Serving with AdaLA
## 📝 Contributing
All contributions are welcome to improve AdaSeq. Please refer to [CONTRIBUTING.md](./CONTRIBUTING.md) for the contributing guideline.
## 📄 License
This project is licensed under the Apache License (Version 2.0).
Raw data
{
"_id": null,
"home_page": "https://github.com/modelscope/adaseq",
"name": "adaseq",
"maintainer": "",
"docs_url": null,
"requires_python": ">=3.7.0",
"maintainer_email": "",
"keywords": "",
"author": "Alibaba Damo Academy NLP foundation team",
"author_email": "",
"download_url": "https://files.pythonhosted.org/packages/20/04/00c44c007d14027a586954e016dc340b3315b726cf14906c494a1a94cb97/adaseq-0.6.6.tar.gz",
"platform": "any",
"description": "# AdaSeq: An All-in-One Library for Developing State-of-the-Art Sequence Understanding Models\n\n<div align=\"center\">\n\n[](./LICENSE)\n[](https://modelscope.cn/)\n\n[](https://github.com/modelscope/AdaSeq/issues)\n[](https://github.com/modelscope/AdaSeq/stargazers)\n[](https://pypi.org/project/adaseq)\n[](./CONTRIBUTING.md)\n\n</div>\n\n<div align=\"center\">\n\nEnglish | [\u7b80\u4f53\u4e2d\u6587](./README_zh.md)\n\n</div>\n\n## Introduction\n***AdaSeq*** (**A**libaba **D**amo **A**cademy **Seq**uence Understanding Toolkit) is an easy-to-use all-in-one library, built on [ModelScope](https://modelscope.cn/home), that allows researchers and developers to train custom models for sequence understanding tasks, including part-of-speech tagging (POS Tagging), chunking, named entity recognition (NER), entity typing, relation extraction (RE), etc.\n\n\n\n<details open>\n<summary>\ud83c\udf1f <b>Features:</b></summary>\n\n- **Plentiful Models**:\n\n AdaSeq provide plenty of cutting-edge models, training methods and useful toolkits for sequence understanding tasks.\n\n- **State-of-the-Art**:\n\n Our aim to develop the best implementation, which can beat many off-the-shelf frameworks on performance.\n\n- **Easy-to-Use**:\n\n One line of command is all you need to obtain the best model.\n\n- **Extensible**:\n\n It's easy to register a module, or build a customized sequence understanding model by assembling the predefined modules.\n\n</details>\n\n\u26a0\ufe0f**Notice:** This project is under quick development. This means some interfaces could be changed in the future.\n\n## \ud83d\udce2 What's New\n- 2022-07: [SemEval 2023] Our U-RaNER paper won [Best Paper Award](https://semeval.github.io/SemEval2023/awards)!\n- 2022-03: [SemEval 2023] Our U-RaNER won ***1st place in 9 tracks*** at [SemEval 2023 Task2](https://multiconer.github.io/results): Multilingual Complex Named Entity Recognition! [Model introduction and source code can be found here](./examples/U-RaNER).\n- 2022-12: [[EMNLP 2022] Retrieval-augmented Multimodal Entity Understanding Model (MoRe)](./examples/MoRe)\n- 2022-11: [[EMNLP 2022] Ultra-Fine Entity Typing Model (NPCRF)](./examples/NPCRF)\n- 2022-11: [[EMNLP 2022] Unsupervised Boundary-Aware Language Model (BABERT)](./examples/babert)\n\n## \u26a1 Quick Experience\nYou can try out our models via online demos built on ModelScope:\n[[English NER]](https://modelscope.cn/models/damo/nlp_raner_named-entity-recognition_english-large-news/summary)\n[[Chinese NER]](https://modelscope.cn/models/damo/nlp_raner_named-entity-recognition_chinese-base-news/summary)\n[[CWS]](https://modelscope.cn/models/damo/nlp_structbert_word-segmentation_chinese-base/summary)\n\nMore tasks, more languages, more domains: All modelcards we released can be found in this page [Modelcards](./docs/modelcards.md).\n\n## \ud83d\udee0\ufe0f Model Zoo\n<details open>\n<summary><b>Supported models:</b></summary>\n\n- [Transformer-based CRF](./examples/bert_crf)\n- [Partial CRF](./examples/partial_bert_crf)\n- [Retrieval Augmented NER](./examples/RaNER)\n- [Biaffine NER](./examples/biaffine_ner)\n- [Global-Pointer](./examples/global_pointer)\n- [Multi-label Entity Typing](./examples/entity_typing)\n- ...\n</details>\n\n## \ud83d\udcbe Dataset Zoo\nWe collected many datasets for sequence understanding tasks. All can be found in this page [Datasets](./docs/datasets.md).\n\n## \ud83d\udce6 Installation\nAdaSeq project is based on `Python >= 3.7`, `PyTorch >= 1.8` and `ModelScope >= 1.4`. We assure that AdaSeq can run smoothly when `ModelScope == 1.9.5`.\n\n- installation via pip\uff1a\n```\npip install adaseq\n```\n\n- installation from source\uff1a\n```\ngit clone https://github.com/modelscope/adaseq.git\ncd adaseq\npip install -r requirements.txt -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo.html\n```\n\n### Verify the Installation\nTo verify whether AdaSeq is installed properly, we provide a demo config for training a model (the demo config will be automatically downloaded).\n```\nadaseq train -c demo.yaml\n```\nYou will see the training logs on your terminal. Once the training is done, the results on test set will be printed: `test: {\"precision\": xxx, \"recall\": xxx, \"f1\": xxx}`. A folder `experiments/toy_msra/` will be generated to save all experimental results and model checkpoints.\n\n## \ud83d\udcd6 Tutorials\n- [Quick Start](./docs/tutorials/quick_start.md)\n- Basics\n - [Learning about Configs](./docs/tutorials/learning_about_configs.md)\n - [Customizing Dataset](./docs/tutorials/customizing_dataset.md)\n - [TODO] Common Architectures\n - [TODO] Useful Hooks\n - [Hyperparameter Optimization](./docs/tutorials/hyperparameter_optimization.md)\n - [Training with Multiple GPUs](./docs/tutorials/training_with_multiple_gpus.md)\n- Best Practice\n - [Training a Model with Custom Dataset](./docs/tutorials/training_a_model.md)\n - [Reproducing Results in Published Papers](./docs/tutorials/reproducing_papers.md)\n - [TODO] Uploading Saved Model to ModelScope\n - [TODO] Customizing your Model\n - [TODO] Serving with AdaLA\n\n## \ud83d\udcdd Contributing\nAll contributions are welcome to improve AdaSeq. Please refer to [CONTRIBUTING.md](./CONTRIBUTING.md) for the contributing guideline.\n\n## \ud83d\udcc4 License\nThis project is licensed under the Apache License (Version 2.0).\n\n\n",
"bugtrack_url": null,
"license": "Apache License 2.0",
"summary": "AdaSeq: An All-in-One Library for Developing State-of-the-Art Sequence Understanding Models",
"version": "0.6.6",
"project_urls": {
"Homepage": "https://github.com/modelscope/adaseq"
},
"split_keywords": [],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "4947ddf684253dbb4c3e0716fcda67094aa3c407237d5eb8930ede0a91b9feb8",
"md5": "3590769c827849e06546b94707cba15d",
"sha256": "22b692b89bcee6e7ec77e6fe36501732036fe4cc0db256a6cb94648c0b3251b3"
},
"downloads": -1,
"filename": "adaseq-0.6.6-py3-none-any.whl",
"has_sig": false,
"md5_digest": "3590769c827849e06546b94707cba15d",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.7.0",
"size": 149508,
"upload_time": "2023-11-15T12:06:59",
"upload_time_iso_8601": "2023-11-15T12:06:59.032527Z",
"url": "https://files.pythonhosted.org/packages/49/47/ddf684253dbb4c3e0716fcda67094aa3c407237d5eb8930ede0a91b9feb8/adaseq-0.6.6-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "200400c44c007d14027a586954e016dc340b3315b726cf14906c494a1a94cb97",
"md5": "9ecbbf474188029060fd3da25e09be58",
"sha256": "395c66a77e3e2128b79811f9845416f64f619649601c5a1f94e37e6ca590212d"
},
"downloads": -1,
"filename": "adaseq-0.6.6.tar.gz",
"has_sig": false,
"md5_digest": "9ecbbf474188029060fd3da25e09be58",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.7.0",
"size": 97557,
"upload_time": "2023-11-15T12:07:01",
"upload_time_iso_8601": "2023-11-15T12:07:01.943186Z",
"url": "https://files.pythonhosted.org/packages/20/04/00c44c007d14027a586954e016dc340b3315b726cf14906c494a1a94cb97/adaseq-0.6.6.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2023-11-15 12:07:01",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "modelscope",
"github_project": "adaseq",
"travis_ci": false,
"coveralls": false,
"github_actions": false,
"requirements": [
{
"name": "addict",
"specs": []
},
{
"name": "datasets",
"specs": []
},
{
"name": "modelscope",
"specs": [
[
">=",
"1.4.0"
]
]
},
{
"name": "seqeval",
"specs": []
},
{
"name": "torch",
"specs": [
[
">=",
"1.11.0"
]
]
},
{
"name": "tqdm",
"specs": [
[
">=",
"4.64.0"
]
]
},
{
"name": "transformers",
"specs": [
[
">=",
"4.21.0"
]
]
},
{
"name": "urllib3",
"specs": [
[
">=",
"1.26.0"
]
]
}
],
"lcname": "adaseq"
}
JSON
