SimpleText@CLEF-2022 Tools
- JURASSIC: AI21 Studio provides access to Jurassic-1 suite of language models. Jurassic-1 is large AI model able to transform an existing text. One common transformation is summarization. Summarization task can be fine-tuned via the AI21 API or in the GUI https://www.ai21.com/blog/ai21-studio-use-cases
- GPT-2 is a large-scale unsupervised language model which performs machine translation, question answering, and summarization without task-specific training https://github.com/openai/gpt-2
- Multilingual T5 (mT5) is a large multilingual pretrained text-to-text transformer model developped by Google covering 101 languages. mT5 can be fine-tuned for any text-to-text generation. It is possible to fine-tune the mT5 model by using the SimpleT5 library. https://github.com/google-research/multilingual-t5
Task 3: Text simplification tools
- JURASSIC: AI21 Studio provides access to Jurassic-1 suite of language models. Jurassic-1 Jumbo is the largest model publicly available with no waitlist. The AI21 studio’s playground provides ready-to-use promts for text simplification (see De-Jargonizer) https://www.ai21.com/studio
- GPT-2 is a large-scale unsupervised language model which performs machine translation, question answering, and summarization without task-specific training https://github.com/openai/gpt-2
- Multilingual T5 (mT5) transformer model can be fine-tuned for text simplification e.g. by using SimpleT5 library. https://github.com/google-research/multilingual-t5