Hi, I just want to know how to use multi-gpu when fine-tuning the commonsense reasoning task? I tried running the script as below since the last argument determines the GPU to use but it doesn't work.
sh llama_7B_Dora.sh 32 64 ./finetuned_result/dora_r32 0,1