optim_type: adamw
lr_choice: layerwise_decay
lr_schedule: cosine_decay
lr: 0.0001
lr_decay: 0.9
end_lr: 0
lr_mult: 1
weight_decay: 0.001
warmup_steps: 0.1
validation_metric_name: accuracy
peft: null
mixup_off_epoch: 5
skip_final_val: false
track_grad_norm: -1
cross_modal_align: null
cross_modal_align_weight: 0
automatic_optimization: true
accumulate_grad_batches: 8
gradient_clip_val: 1
gradient_clip_algorithm: norm
use_aug_optim: false
aug_lr: 0.0001
aug_weight_decay: 1.0e-05
aug_optim_type: adamw