German Qg T5 Quad

🚀 german-qg-t5-quad

This model is fine - tuned for question generation in German. It requires the expected answer to be highlighted with a token, which helps in generating relevant questions from German text.

🚀 Quick Start

This model is designed to generate questions from German text. To use it, simply provide input text with the expected answer highlighted by tokens, and the model will generate corresponding questions.

✨ Features

• German Question Generation: Specifically fine - tuned for generating questions in German.
• Answer Highlighting: Relies on tokens to identify the expected answer for question generation.

📦 Installation

No installation steps are provided in the original document, so this section is skipped.

💻 Usage Examples

Basic Usage

Input

generate question: Obwohl die Vereinigten Staaten wie auch viele Staaten des Commonwealth Erben des <hl> britischen Common Laws <hl> sind, setzt sich das amerikanische Recht bedeutend davon ab. Dies rührt größtenteils von dem langen Zeitraum her, [...]

Expected output

Von welchem Gesetzt stammt das Amerikanische ab? 

📚 Documentation

Model description

This model is a fine - tuned version of valhalla/t5-base-qg-hl on the GermanQUAD dataset.

Training and evaluation data

The training script can be accessed here.

Evaluation

The model achieves a BLEU - 4 score of 11.30 on the GermanQuAD test set (n = 2204).

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0001
• train_batch_size: 2
• eval_batch_size: 2
• seed: 100
• gradient_accumulation_steps: 8
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon = 1e - 08
• lr_scheduler_type: linear
• num_epochs: 10

Framework versions

• Transformers 4.13.0.dev0
• Pytorch 1.10.0+cu102
• Datasets 1.16.1
• Tokenizers 0.10.3

🔧 Technical Details

The model's fine - tuning on the GermanQUAD dataset and the specific hyperparameters used during training contribute to its performance in German question generation. The use of the token for answer highlighting is a key technical aspect that guides the question - generation process.

📄 License

The model is released under the MIT license.