OOM when allocating tensor with shape[270000,2515] and type float on

Question

I am getting this error while trying to train a model, what can i do to remove this? Also this code only runs on dedicated Vega 8 GPU and not on my dedicated Notebook GPU (1