I have written a script using gradio and sometimes (I emphasize this- only sometimes) when I run it I get

huggingface/tokenizers: The current process just got forked, after parallelism has already been used. Disabling parallelism to avoid deadlocks...
To disable this warning, you can either:
    - Avoid using `tokenizers` before the fork if possible
    - Explicitly set the environment variable TOKENIZERS_PARALLELISM=(true | false)

The strange thing is that different to the answers I got searching the internet , I am not using tokenizers, or LLMs or accessing huggingface models or forking anything.

My code is simply a gradio script that uses dropdowns buttons and dataframes and that when clicked it sends a request to a remote machine. This remote machine is running an API that actually calls a LLava model, but that is process running on a remote machine.

The local machine- where the error is displayed- just sends a request and receives it

The only other thing that occurs to me is that when the process is finished I do

with mlflow.start_run() as run:
    mlflow.log_param("some_param",something)
    # ... some more param metrics and artifacts logging
    print(
                f"Logged data to MLFlow with run ID: {run.info.run_id} in experiment: {experiment_name}"
            )

The error message appears in this moment and before the final print

Any hint of why this could be happening?