simplify_dutch

Model Details

simplify_dutch

This is the source code for my thesis on "Controllable Sentence Simplification in Dutch" in the Masters of AI at KU Leuven. The full code can be found at: https://github.com/tsei902/simplify_dutch

Data

The origin of the datasets in resources/datasets is:

1. Wikilarge, available under: https://github.com/XingxingZhang/dress The wikilarge data is limited the first 10000 rows.

2. ASSET, available under: https://github.com/facebookresearch Which both have been translated to Dutch.

Model

The Dutch T5 model t5-base-dutch from Hugging Face has been adopted and trained on the task of sentence simplification. The folder /saved model contains the final trained model on 10000 rows of data, as stated in the Thesis.

Sequence:

1. TRAINING DATA needs preprocessing with preprocessor.py
2. Generation can be done with generate_on_pretrained.py with a prior adjustment of
3. The generation parameters in model.simplify() where the decoding method needs to be chosen (Greedy decoding, Top-p & top-k, or Beam search)
4. Manual scoring of a generated text is possible with evaluate.py

Further remarks:

1. The folder resources/processed data contains the training set with the prepended control tokens
2. The folder resources/DUMPS contains the Word embeddings from Fares et al. (2017) have been used. The data is available under: http://vectors.nlpl.eu/repository. (Fares, M., Kutuzov, A., Oepen, S., & Velldal, E. (2017). Word vectors, reuse, and replicability: Towards a community repository of large-text resources. Proceedings of the 21st Nordic Conference on Computational Linguistics, Gothenburg, Sweden.)
3. The folder resources/outputs/final_decoder_outputs contains the final generated text per decoding strategy (Greedy decoding, Top-p & top-k, or Beam search) for both the full test set and the sample dataset
4. The folder translations contains sampled text (106 and 84 rows) from the original English datasets (WIKILarge and ASSET), a machine-translated version as well as the human translated references.

Credits

The preprocessor.py and the utils.py contain code that has been adapted from https://github.com/KimChengSHEANG/TS_T5 (Sheang, K. C., & Saggion, H. (2021). Controllable Sentence Simplification with a Unified Text-to-Text Transfer Transformer.INLG 2021 International Conference on Natural Language Generation, Aberdeen, Scotland, UK.) The preprocessor.py has been adapted to the usage of Dutch.

Model Description

<!-- Provide a longer summary of what this model is. -->

• Developed by: Theresa Seidl
• Funded by [optional]: [More Information Needed]
• Shared by [optional]: [More Information Needed]
• Model type: [More Information Needed]
• Language(s) (NLP): Dutsch
• License: [More Information Needed]
• Finetuned from model [optional]: https://huggingface.co/yhavinga/t5-base-dutch

Model Sources [optional]

<!-- Provide the basic links for the model. -->

• Repository: https://github.com/tsei902/simplify_dutch
• Paper [optional]: [More Information Needed]
• Demo [optional]: [More Information Needed]

Uses

<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->

Direct Use

<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->

[More Information Needed]

Downstream Use [optional]

<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->

[More Information Needed]

Out-of-Scope Use

<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->

[More Information Needed]

Bias, Risks, and Limitations

<!-- This section is meant to convey both technical and sociotechnical limitations. -->

[More Information Needed]

Recommendations

<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->

Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.

How to Get Started with the Model

Use the code below to get started with the model.

[More Information Needed]

Training Details

Training Data

<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->

[More Information Needed]

Training Procedure

<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->

Preprocessing [optional]

[More Information Needed]

Training Hyperparameters

• Training regime: [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->

Speeds, Sizes, Times [optional]

<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->

[More Information Needed]

Evaluation

<!-- This section describes the evaluation protocols and provides the results. -->

Testing Data, Factors & Metrics

Testing Data

<!-- This should link to a Dataset Card if possible. -->

[More Information Needed]

Metrics

<!-- These are the evaluation metrics being used, ideally with a description of why. -->

[More Information Needed]

Results

[More Information Needed]

Summary

Model Architecture and Objective

[More Information Needed]

Compute Infrastructure

[More Information Needed]

Hardware

[More Information Needed]

Citation [optional]

<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->

BibTeX:

[More Information Needed]

APA:

[More Information Needed]