FRAMEWORK=transformers
GPU_TYPE=Iluvatar_bi-150
TASK_TYPE=vision_classification