CUDA out of memory.

outputs = model(input_ids) Even though I’m testing a small model, running model(input_ids) still causes an out-of-memory error, even when I test on Colab with batch_size = 1.