MilaNLProc / prompting-italian-text-simplification

Code associated with the paper "Is It Really That Simple? Prompting Language Models for Automatic Text Simplification in Italian".

Is It Really That Simple?

Code and data associated with the paper "Is It Really That Simple? Prompting Language Models for Automatic Text Simplification in Italian".

Getting Started

Beware, the operation might break existing venv/conda environments. We recommend working on a separate environment. We conducted all our experiments with Python 3.10. To get started, install the requirements listed in requirements.txt.

pip install -r requirements.txt

Run the Simplification

Execute the following command from the root of this project to execute the simplification on Admin-It with Vicuna v1.3 13B.

Notes:

• edit the model name to use different models
• you will likely need to change some of the parameters to adapt to your setup

./bash/simplify_with_model.sh "lmsys/vicuna-13b-v1.3" "admin-It"

If you want to use LoRA weights, then use:

./bash/simplify_with_lora.sh "yahma/llama-7b-hf" "teelinsan/camoscio-7b-llama" "admin-It"

where the first argument is the base pretrained model, and the second is the name of the adapter weights.

Prompt Template

Both scripts will use the explicit prompt template by default. Edit them and pass --prompt_template="1" among the input arguments to use the implicit template.