python -m nanoT5.main   optim.name=adafactor   optim.lr_scheduler=legacy model.name=/home/jovyan/bert-train/nanot5/templates/base_slavic_120k_sub optim.grad_acc=8 model.overwrite.dropout_rate=0.1 model.klass=hf_t5 eval.every_steps=5000 optim.total_steps=120000 model.add_config.is_bf16=True checkpoint.every_steps=10000 model.compile=False