WCGA AI validator

• Install the required dependencies

pip install -r requirements.txt

.env variable

• mllm_end_point_openai=‘https://hiis-accessibility-fonderia.cognitiveservices.azure.com/openai/deployments/gpt-4o/chat/completions?api-version=2025-01-01-preview’

• mllm_api_key_openai=

• mllm_model_id_openai=‘gpt-4o’

• mllm_end_point_local=‘https://vgpu.hiis.cloud.isti.cnr.it/api/chat’

• mllm_api_key_local=

• #mllm_model_id_local=‘gemma3:12b’

• mllm_model_id_local=‘gemma3:4b’

use_openai_model=‘False’ # use ‘True’ if openai else False

For the CLI version use:

python wcag_validator.py

For the RESTservice use:

python wcag_validator_RESTserver.py

For UI use:

python wcag_validator_ui.py

Docker (file docker placed at path /LLM_accessibility_validator)

Rest server

docker build -t wcag_resr_server . docker run –env-file .env -p 8000:8000 –name wcag_rest_server -d wcag_rest_server ### UI docker build -t wcag_ui . docker run –env-file UI/.env -p 7860:7860 –name wcag_ui -d wcag_ui

The scripts folder contains some elaboration scripts. They require a dedicated requirements file