OUT OF MEMory on 4 a100 gpus

Dear authors ,
I put each and every thig right as per the repo ...NO i wanted to do the sft for that I put model _path as
--model_name_or_path Qwen/Qwen2.5-VL-7B-Instruct\ in run_sft_video.sh But why i am getting Out of Memory . I Have a question did u use LORA PEFT while doing this SFT if yes what was the lora confg
