Deberta V3 Base Absa V1.1

🚀 DeBERTa for Aspect - Based Sentiment Analysis

Powered by PyABSA, an open - source tool for aspect - based sentiment analysis, this model provides reliable solutions for sentiment analysis tasks.

🚀 Quick Start

This model is trained with over 30k ABSA samples from ABSADatasets. Note that the test sets are not included in pre - training. So, you can use this model for training and benchmarking on common ABSA datasets, such as Laptop14 and Rest14 datasets (except for the Rest15 dataset!).

✨ Features

• Rich Training Data: Trained with 30k+ ABSA samples.
• Versatile for Benchmarking: Can be used for training and benchmarking on common ABSA datasets.

📦 Installation

No specific installation steps are provided in the original README. So, this section is skipped.

💻 Usage Examples

Basic Usage

from transformers import AutoTokenizer, AutoModelForSequenceClassification

# Load the ABSA model and tokenizer
model_name = "yangheng/deberta-v3-base-absa-v1.1"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSequenceClassification.from_pretrained(model_name)

classifier = pipeline("text-classification", model=model, tokenizer=tokenizer)

for aspect in ['camera', 'phone']:
   print(aspect, classifier('The camera quality of this phone is amazing.',  text_pair=aspect))

📚 Documentation

Model Details

The deberta - v3 - base - absa model is designed for aspect - based sentiment analysis and is trained with English datasets from ABSADatasets.

Training Model

This model is trained based on the FAST - LCF - BERT model with microsoft/deberta - v3 - base, which comes from PyABSA. To keep track of state - of - the - art models, please refer to PyASBA.

Example in PyASBA

You can find an example for using FAST - LCF - BERT in PyASBA datasets.

Datasets

This model is fine - tuned with 180k examples for the ABSA dataset (including augmented data). The training dataset files are as follows:

loading: integrated_datasets/apc_datasets/SemEval/laptop14/Laptops_Train.xml.seg
loading: integrated_datasets/apc_datasets/SemEval/restaurant14/Restaurants_Train.xml.seg
loading: integrated_datasets/apc_datasets/SemEval/restaurant16/restaurant_train.raw
loading: integrated_datasets/apc_datasets/ACL_Twitter/acl-14-short-data/train.raw
loading: integrated_datasets/apc_datasets/MAMS/train.xml.dat
loading: integrated_datasets/apc_datasets/Television/Television_Train.xml.seg
loading: integrated_datasets/apc_datasets/TShirt/Menstshirt_Train.xml.seg
loading: integrated_datasets/apc_datasets/Yelp/yelp.train.txt

Citation

If you use this model in your research, please cite the following papers:

@inproceedings{DBLP:conf/cikm/0008ZL23,
  author       = {Heng Yang and
                  Chen Zhang and
                  Ke Li},
  editor       = {Ingo Frommholz and
                  Frank Hopfgartner and
                  Mark Lee and
                  Michael Oakes and
                  Mounia Lalmas and
                  Min Zhang and
                  Rodrygo L. T. Santos},
  title        = {PyABSA: {A} Modularized Framework for Reproducible Aspect-based Sentiment
                  Analysis},
  booktitle    = {Proceedings of the 32nd {ACM} International Conference on Information
                  and Knowledge Management, {CIKM} 2023, Birmingham, United Kingdom,
                  October 21-25, 2023},
  pages        = {5117--5122},
  publisher    = {{ACM}},
  year         = {2023},
  url          = {https://doi.org/10.1145/3583780.3614752},
  doi          = {10.1145/3583780.3614752},
  timestamp    = {Thu, 23 Nov 2023 13:25:05 +0100},
  biburl       = {https://dblp.org/rec/conf/cikm/0008ZL23.bib},
  bibsource    = {dblp computer science bibliography, https://dblp.org}
}
@article{YangZMT21,
  author    = {Heng Yang and
               Biqing Zeng and
               Mayi Xu and
               Tianxing Wang},
  title     = {Back to Reality: Leveraging Pattern-driven Modeling to Enable Affordable
               Sentiment Dependency Learning},
  journal   = {CoRR},
  volume    = {abs/2110.08604},
  year      = {2021},
  url       = {https://arxiv.org/abs/2110.08604},
  eprinttype = {arXiv},
  eprint    = {2110.08604},
  timestamp = {Fri, 22 Oct 2021 13:33:09 +0200},
  biburl    = {https://dblp.org/rec/journals/corr/abs-2110-08604.bib},
  bibsource = {dblp computer science bibliography, https://dblp.org}
}

📄 License

This model is released under the MIT license.

Additional Information