deepfacelab on linux server with 2x rtx 3090 using vast.ai -- 2

I can't put the deepfacelab program on a server with two rtx 3090 cards and to work correctly without errors. when running scripts, such an error appears photo 1, I tried this command export openblas_NUM_threads=1 but the result is photo 2. I need a person who could help with this and record the process itself from beginning to end. When you run the program on the server using vast.ai, you have to send files which i will send to you to server and then run 5_xseg_train.sh , script 6_train_SAEHD.sh