Is this the same model that failed to detect the attack from their environment against Hugging Face?