I am currently running OpenP5 with t5 as the backbone.
However, the memory (RAM) is full after about 10% of the first epoch (Dataset is ML1M, batch size is 4) .
I have 64GB, enough to run t5-small.
I am using one GPU, a Geforce RTX 3090.
I would be grateful if you could give me a solution.
I am currently running OpenP5 with t5 as the backbone.
However, the memory (RAM) is full after about 10% of the first epoch (Dataset is ML1M, batch size is 4) .
I have 64GB, enough to run t5-small.
I am using one GPU, a Geforce RTX 3090.
I would be grateful if you could give me a solution.