- optim.name=adafactor
- optim.lr_scheduler=legacy
- model.name=/home/jovyan/bert-train/nanot5/templates/base_sklarge_120k
- data.train_path=/home/jovyan/data/sklarge-shards
- optim.grad_acc=8
- 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