SVFT-main/vision_experiments/run_vit_large.sh

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 1  --output_dir ./results/vit-large/cifar100/lora_1/seed_42 --model_name vit-large --finetuning_method lora --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 8  --output_dir ./results/vit-large/cifar100/lora_8/seed_42 --model_name vit-large --finetuning_method lora --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 1 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/cifar100/lora_1_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method lora --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/cifar100/lora_8_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method lora --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 1  --output_dir ./results/vit-large/cifar100/dora_1/seed_42 --model_name vit-large --finetuning_method dora --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 8  --output_dir ./results/vit-large/cifar100/dora_8/seed_42 --model_name vit-large --finetuning_method dora --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 1 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/cifar100/dora_1_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method dora --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/cifar100/dora_8_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method dora --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --vera_rank 256  --output_dir ./results/vit-large/cifar100/vera_256/seed_42 --model_name vit-large --finetuning_method vera --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 2e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --boft_block_size 2 --boft_n_butterfly_factor 2  --output_dir ./results/vit-large/cifar100/boft_2_2/seed_42 --model_name vit-large --finetuning_method boft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --boft_block_size 4 --boft_n_butterfly_factor 4  --output_dir ./results/vit-large/cifar100/boft_4_4/seed_42 --model_name vit-large --finetuning_method boft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10   --output_dir ./results/vit-large/cifar100/svft_/seed_42 --model_name vit-large --finetuning_method svft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --svft_rank 2  --output_dir ./results/vit-large/cifar100/svft_2/seed_42 --model_name vit-large --finetuning_method svft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --svft_rank 4  --output_dir ./results/vit-large/cifar100/svft_4/seed_42 --model_name vit-large --finetuning_method svft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --svft_rank 8  --output_dir ./results/vit-large/cifar100/svft_8/seed_42 --model_name vit-large --finetuning_method svft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10  --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/cifar100/svft_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method svft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --svft_rank 2 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/cifar100/svft_2_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method svft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --svft_rank 4 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/cifar100/svft_4_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method svft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --svft_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/cifar100/svft_8_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method svft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10   --output_dir ./results/vit-large/cifar100/full_/seed_42 --model_name vit-large --finetuning_method full --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-5 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10   --output_dir ./results/vit-large/cifar100/head_/seed_42 --model_name vit-large --finetuning_method head --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-5 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 1  --output_dir ./results/vit-large/cifar100/lora_1/seed_123 --model_name vit-large --finetuning_method lora --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 8  --output_dir ./results/vit-large/cifar100/lora_8/seed_123 --model_name vit-large --finetuning_method lora --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 1 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/cifar100/lora_1_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method lora --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/cifar100/lora_8_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method lora --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 1  --output_dir ./results/vit-large/cifar100/dora_1/seed_123 --model_name vit-large --finetuning_method dora --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 8  --output_dir ./results/vit-large/cifar100/dora_8/seed_123 --model_name vit-large --finetuning_method dora --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 1 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/cifar100/dora_1_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method dora --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/cifar100/dora_8_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method dora --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --vera_rank 256  --output_dir ./results/vit-large/cifar100/vera_256/seed_123 --model_name vit-large --finetuning_method vera --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 2e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --boft_block_size 2 --boft_n_butterfly_factor 2  --output_dir ./results/vit-large/cifar100/boft_2_2/seed_123 --model_name vit-large --finetuning_method boft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --boft_block_size 4 --boft_n_butterfly_factor 4  --output_dir ./results/vit-large/cifar100/boft_4_4/seed_123 --model_name vit-large --finetuning_method boft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10   --output_dir ./results/vit-large/cifar100/svft_/seed_123 --model_name vit-large --finetuning_method svft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --svft_rank 2  --output_dir ./results/vit-large/cifar100/svft_2/seed_123 --model_name vit-large --finetuning_method svft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --svft_rank 4  --output_dir ./results/vit-large/cifar100/svft_4/seed_123 --model_name vit-large --finetuning_method svft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --svft_rank 8  --output_dir ./results/vit-large/cifar100/svft_8/seed_123 --model_name vit-large --finetuning_method svft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10  --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/cifar100/svft_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method svft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --svft_rank 2 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/cifar100/svft_2_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method svft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --svft_rank 4 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/cifar100/svft_4_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method svft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --svft_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/cifar100/svft_8_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method svft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10   --output_dir ./results/vit-large/cifar100/full_/seed_123 --model_name vit-large --finetuning_method full --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-5 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10   --output_dir ./results/vit-large/cifar100/head_/seed_123 --model_name vit-large --finetuning_method head --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-5 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 1  --output_dir ./results/vit-large/cifar100/lora_1/seed_3076 --model_name vit-large --finetuning_method lora --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 8  --output_dir ./results/vit-large/cifar100/lora_8/seed_3076 --model_name vit-large --finetuning_method lora --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 1 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/cifar100/lora_1_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method lora --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/cifar100/lora_8_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method lora --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 1  --output_dir ./results/vit-large/cifar100/dora_1/seed_3076 --model_name vit-large --finetuning_method dora --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 8  --output_dir ./results/vit-large/cifar100/dora_8/seed_3076 --model_name vit-large --finetuning_method dora --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 1 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/cifar100/dora_1_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method dora --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/cifar100/dora_8_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method dora --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --vera_rank 256  --output_dir ./results/vit-large/cifar100/vera_256/seed_3076 --model_name vit-large --finetuning_method vera --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 2e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --boft_block_size 2 --boft_n_butterfly_factor 2  --output_dir ./results/vit-large/cifar100/boft_2_2/seed_3076 --model_name vit-large --finetuning_method boft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --boft_block_size 4 --boft_n_butterfly_factor 4  --output_dir ./results/vit-large/cifar100/boft_4_4/seed_3076 --model_name vit-large --finetuning_method boft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10   --output_dir ./results/vit-large/cifar100/svft_/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --svft_rank 2  --output_dir ./results/vit-large/cifar100/svft_2/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --svft_rank 4  --output_dir ./results/vit-large/cifar100/svft_4/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --svft_rank 8  --output_dir ./results/vit-large/cifar100/svft_8/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10  --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/cifar100/svft_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --svft_rank 2 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/cifar100/svft_2_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --svft_rank 4 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/cifar100/svft_4_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --svft_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/cifar100/svft_8_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10   --output_dir ./results/vit-large/cifar100/full_/seed_3076 --model_name vit-large --finetuning_method full --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-5 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10   --output_dir ./results/vit-large/cifar100/head_/seed_3076 --model_name vit-large --finetuning_method head --dataset_name cifar100 --clf_learning_rate 4e-3 --other_learning_rate 4e-5 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 1  --output_dir ./results/vit-large/food101/lora_1/seed_42 --model_name vit-large --finetuning_method lora --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 8  --output_dir ./results/vit-large/food101/lora_8/seed_42 --model_name vit-large --finetuning_method lora --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 1 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/food101/lora_1_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method lora --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/food101/lora_8_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method lora --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 1  --output_dir ./results/vit-large/food101/dora_1/seed_42 --model_name vit-large --finetuning_method dora --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 8  --output_dir ./results/vit-large/food101/dora_8/seed_42 --model_name vit-large --finetuning_method dora --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 1 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/food101/dora_1_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method dora --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/food101/dora_8_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method dora --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --vera_rank 256  --output_dir ./results/vit-large/food101/vera_256/seed_42 --model_name vit-large --finetuning_method vera --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --boft_block_size 2 --boft_n_butterfly_factor 2  --output_dir ./results/vit-large/food101/boft_2_2/seed_42 --model_name vit-large --finetuning_method boft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --boft_block_size 4 --boft_n_butterfly_factor 4  --output_dir ./results/vit-large/food101/boft_4_4/seed_42 --model_name vit-large --finetuning_method boft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10   --output_dir ./results/vit-large/food101/svft_/seed_42 --model_name vit-large --finetuning_method svft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --svft_rank 2  --output_dir ./results/vit-large/food101/svft_2/seed_42 --model_name vit-large --finetuning_method svft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --svft_rank 4  --output_dir ./results/vit-large/food101/svft_4/seed_42 --model_name vit-large --finetuning_method svft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --svft_rank 8  --output_dir ./results/vit-large/food101/svft_8/seed_42 --model_name vit-large --finetuning_method svft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10  --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/food101/svft_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method svft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --svft_rank 2 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/food101/svft_2_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method svft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --svft_rank 4 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/food101/svft_4_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method svft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --svft_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/food101/svft_8_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method svft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10   --output_dir ./results/vit-large/food101/full_/seed_42 --model_name vit-large --finetuning_method full --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-5 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10   --output_dir ./results/vit-large/food101/head_/seed_42 --model_name vit-large --finetuning_method head --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-5 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 1  --output_dir ./results/vit-large/food101/lora_1/seed_123 --model_name vit-large --finetuning_method lora --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 8  --output_dir ./results/vit-large/food101/lora_8/seed_123 --model_name vit-large --finetuning_method lora --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 1 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/food101/lora_1_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method lora --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/food101/lora_8_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method lora --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 1  --output_dir ./results/vit-large/food101/dora_1/seed_123 --model_name vit-large --finetuning_method dora --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 8  --output_dir ./results/vit-large/food101/dora_8/seed_123 --model_name vit-large --finetuning_method dora --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 1 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/food101/dora_1_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method dora --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/food101/dora_8_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method dora --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --vera_rank 256  --output_dir ./results/vit-large/food101/vera_256/seed_123 --model_name vit-large --finetuning_method vera --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --boft_block_size 2 --boft_n_butterfly_factor 2  --output_dir ./results/vit-large/food101/boft_2_2/seed_123 --model_name vit-large --finetuning_method boft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --boft_block_size 4 --boft_n_butterfly_factor 4  --output_dir ./results/vit-large/food101/boft_4_4/seed_123 --model_name vit-large --finetuning_method boft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10   --output_dir ./results/vit-large/food101/svft_/seed_123 --model_name vit-large --finetuning_method svft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --svft_rank 2  --output_dir ./results/vit-large/food101/svft_2/seed_123 --model_name vit-large --finetuning_method svft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --svft_rank 4  --output_dir ./results/vit-large/food101/svft_4/seed_123 --model_name vit-large --finetuning_method svft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --svft_rank 8  --output_dir ./results/vit-large/food101/svft_8/seed_123 --model_name vit-large --finetuning_method svft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10  --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/food101/svft_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method svft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --svft_rank 2 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/food101/svft_2_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method svft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --svft_rank 4 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/food101/svft_4_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method svft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --svft_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/food101/svft_8_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method svft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10   --output_dir ./results/vit-large/food101/full_/seed_123 --model_name vit-large --finetuning_method full --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-5 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10   --output_dir ./results/vit-large/food101/head_/seed_123 --model_name vit-large --finetuning_method head --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-5 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 1  --output_dir ./results/vit-large/food101/lora_1/seed_3076 --model_name vit-large --finetuning_method lora --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 8  --output_dir ./results/vit-large/food101/lora_8/seed_3076 --model_name vit-large --finetuning_method lora --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 1 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/food101/lora_1_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method lora --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/food101/lora_8_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method lora --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 1  --output_dir ./results/vit-large/food101/dora_1/seed_3076 --model_name vit-large --finetuning_method dora --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 8  --output_dir ./results/vit-large/food101/dora_8/seed_3076 --model_name vit-large --finetuning_method dora --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 1 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/food101/dora_1_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method dora --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/food101/dora_8_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method dora --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --vera_rank 256  --output_dir ./results/vit-large/food101/vera_256/seed_3076 --model_name vit-large --finetuning_method vera --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --boft_block_size 2 --boft_n_butterfly_factor 2  --output_dir ./results/vit-large/food101/boft_2_2/seed_3076 --model_name vit-large --finetuning_method boft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --boft_block_size 4 --boft_n_butterfly_factor 4  --output_dir ./results/vit-large/food101/boft_4_4/seed_3076 --model_name vit-large --finetuning_method boft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10   --output_dir ./results/vit-large/food101/svft_/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --svft_rank 2  --output_dir ./results/vit-large/food101/svft_2/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --svft_rank 4  --output_dir ./results/vit-large/food101/svft_4/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --svft_rank 8  --output_dir ./results/vit-large/food101/svft_8/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10  --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/food101/svft_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --svft_rank 2 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/food101/svft_2_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --svft_rank 4 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/food101/svft_4_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --svft_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/food101/svft_8_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10   --output_dir ./results/vit-large/food101/full_/seed_3076 --model_name vit-large --finetuning_method full --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-5 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10   --output_dir ./results/vit-large/food101/head_/seed_3076 --model_name vit-large --finetuning_method head --dataset_name food101 --clf_learning_rate 4e-3 --other_learning_rate 4e-5 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 1  --output_dir ./results/vit-large/flowers102/lora_1/seed_42 --model_name vit-large --finetuning_method lora --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 8  --output_dir ./results/vit-large/flowers102/lora_8/seed_42 --model_name vit-large --finetuning_method lora --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 1 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/flowers102/lora_1_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method lora --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/flowers102/lora_8_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method lora --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 1  --output_dir ./results/vit-large/flowers102/dora_1/seed_42 --model_name vit-large --finetuning_method dora --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 8  --output_dir ./results/vit-large/flowers102/dora_8/seed_42 --model_name vit-large --finetuning_method dora --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 1 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/flowers102/dora_1_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method dora --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/flowers102/dora_8_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method dora --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --vera_rank 256  --output_dir ./results/vit-large/flowers102/vera_256/seed_42 --model_name vit-large --finetuning_method vera --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --boft_block_size 2 --boft_n_butterfly_factor 2  --output_dir ./results/vit-large/flowers102/boft_2_2/seed_42 --model_name vit-large --finetuning_method boft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --boft_block_size 4 --boft_n_butterfly_factor 4  --output_dir ./results/vit-large/flowers102/boft_4_4/seed_42 --model_name vit-large --finetuning_method boft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10   --output_dir ./results/vit-large/flowers102/svft_/seed_42 --model_name vit-large --finetuning_method svft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --svft_rank 2  --output_dir ./results/vit-large/flowers102/svft_2/seed_42 --model_name vit-large --finetuning_method svft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --svft_rank 4  --output_dir ./results/vit-large/flowers102/svft_4/seed_42 --model_name vit-large --finetuning_method svft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --svft_rank 8  --output_dir ./results/vit-large/flowers102/svft_8/seed_42 --model_name vit-large --finetuning_method svft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10  --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/flowers102/svft_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method svft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --svft_rank 2 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/flowers102/svft_2_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method svft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --svft_rank 4 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/flowers102/svft_4_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method svft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --svft_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/flowers102/svft_8_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method svft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10   --output_dir ./results/vit-large/flowers102/full_/seed_42 --model_name vit-large --finetuning_method full --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-5 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10   --output_dir ./results/vit-large/flowers102/head_/seed_42 --model_name vit-large --finetuning_method head --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-5 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 1  --output_dir ./results/vit-large/flowers102/lora_1/seed_123 --model_name vit-large --finetuning_method lora --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 8  --output_dir ./results/vit-large/flowers102/lora_8/seed_123 --model_name vit-large --finetuning_method lora --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 1 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/flowers102/lora_1_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method lora --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/flowers102/lora_8_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method lora --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 1  --output_dir ./results/vit-large/flowers102/dora_1/seed_123 --model_name vit-large --finetuning_method dora --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 8  --output_dir ./results/vit-large/flowers102/dora_8/seed_123 --model_name vit-large --finetuning_method dora --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 1 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/flowers102/dora_1_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method dora --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/flowers102/dora_8_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method dora --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --vera_rank 256  --output_dir ./results/vit-large/flowers102/vera_256/seed_123 --model_name vit-large --finetuning_method vera --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --boft_block_size 2 --boft_n_butterfly_factor 2  --output_dir ./results/vit-large/flowers102/boft_2_2/seed_123 --model_name vit-large --finetuning_method boft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --boft_block_size 4 --boft_n_butterfly_factor 4  --output_dir ./results/vit-large/flowers102/boft_4_4/seed_123 --model_name vit-large --finetuning_method boft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10   --output_dir ./results/vit-large/flowers102/svft_/seed_123 --model_name vit-large --finetuning_method svft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --svft_rank 2  --output_dir ./results/vit-large/flowers102/svft_2/seed_123 --model_name vit-large --finetuning_method svft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --svft_rank 4  --output_dir ./results/vit-large/flowers102/svft_4/seed_123 --model_name vit-large --finetuning_method svft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --svft_rank 8  --output_dir ./results/vit-large/flowers102/svft_8/seed_123 --model_name vit-large --finetuning_method svft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10  --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/flowers102/svft_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method svft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --svft_rank 2 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/flowers102/svft_2_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method svft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --svft_rank 4 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/flowers102/svft_4_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method svft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --svft_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/flowers102/svft_8_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method svft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10   --output_dir ./results/vit-large/flowers102/full_/seed_123 --model_name vit-large --finetuning_method full --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-5 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10   --output_dir ./results/vit-large/flowers102/head_/seed_123 --model_name vit-large --finetuning_method head --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-5 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 1  --output_dir ./results/vit-large/flowers102/lora_1/seed_3076 --model_name vit-large --finetuning_method lora --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 8  --output_dir ./results/vit-large/flowers102/lora_8/seed_3076 --model_name vit-large --finetuning_method lora --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 1 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/flowers102/lora_1_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method lora --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/flowers102/lora_8_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method lora --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 1  --output_dir ./results/vit-large/flowers102/dora_1/seed_3076 --model_name vit-large --finetuning_method dora --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 8  --output_dir ./results/vit-large/flowers102/dora_8/seed_3076 --model_name vit-large --finetuning_method dora --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 1 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/flowers102/dora_1_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method dora --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/flowers102/dora_8_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method dora --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --vera_rank 256  --output_dir ./results/vit-large/flowers102/vera_256/seed_3076 --model_name vit-large --finetuning_method vera --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --boft_block_size 2 --boft_n_butterfly_factor 2  --output_dir ./results/vit-large/flowers102/boft_2_2/seed_3076 --model_name vit-large --finetuning_method boft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --boft_block_size 4 --boft_n_butterfly_factor 4  --output_dir ./results/vit-large/flowers102/boft_4_4/seed_3076 --model_name vit-large --finetuning_method boft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10   --output_dir ./results/vit-large/flowers102/svft_/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --svft_rank 2  --output_dir ./results/vit-large/flowers102/svft_2/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --svft_rank 4  --output_dir ./results/vit-large/flowers102/svft_4/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --svft_rank 8  --output_dir ./results/vit-large/flowers102/svft_8/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10  --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/flowers102/svft_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --svft_rank 2 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/flowers102/svft_2_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --svft_rank 4 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/flowers102/svft_4_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --svft_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/flowers102/svft_8_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10   --output_dir ./results/vit-large/flowers102/full_/seed_3076 --model_name vit-large --finetuning_method full --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-5 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10   --output_dir ./results/vit-large/flowers102/head_/seed_3076 --model_name vit-large --finetuning_method head --dataset_name flowers102 --clf_learning_rate 4e-3 --other_learning_rate 4e-5 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 1  --output_dir ./results/vit-large/resisc45/lora_1/seed_42 --model_name vit-large --finetuning_method lora --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 8  --output_dir ./results/vit-large/resisc45/lora_8/seed_42 --model_name vit-large --finetuning_method lora --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 1 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/resisc45/lora_1_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method lora --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/resisc45/lora_8_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method lora --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 1  --output_dir ./results/vit-large/resisc45/dora_1/seed_42 --model_name vit-large --finetuning_method dora --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 8  --output_dir ./results/vit-large/resisc45/dora_8/seed_42 --model_name vit-large --finetuning_method dora --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 1 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/resisc45/dora_1_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method dora --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --lora_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/resisc45/dora_8_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method dora --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --vera_rank 256  --output_dir ./results/vit-large/resisc45/vera_256/seed_42 --model_name vit-large --finetuning_method vera --dataset_name resisc45 --clf_learning_rate 3e-3 --other_learning_rate 7e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --boft_block_size 2 --boft_n_butterfly_factor 2  --output_dir ./results/vit-large/resisc45/boft_2_2/seed_42 --model_name vit-large --finetuning_method boft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --boft_block_size 4 --boft_n_butterfly_factor 4  --output_dir ./results/vit-large/resisc45/boft_4_4/seed_42 --model_name vit-large --finetuning_method boft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10   --output_dir ./results/vit-large/resisc45/svft_/seed_42 --model_name vit-large --finetuning_method svft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --svft_rank 2  --output_dir ./results/vit-large/resisc45/svft_2/seed_42 --model_name vit-large --finetuning_method svft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --svft_rank 4  --output_dir ./results/vit-large/resisc45/svft_4/seed_42 --model_name vit-large --finetuning_method svft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --svft_rank 8  --output_dir ./results/vit-large/resisc45/svft_8/seed_42 --model_name vit-large --finetuning_method svft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10  --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/resisc45/svft_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method svft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --svft_rank 2 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/resisc45/svft_2_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method svft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --svft_rank 4 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/resisc45/svft_4_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method svft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10 --svft_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/resisc45/svft_8_query_key_value_intermediate.dense_output.dense/seed_42 --model_name vit-large --finetuning_method svft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10   --output_dir ./results/vit-large/resisc45/full_/seed_42 --model_name vit-large --finetuning_method full --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-5 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 42 --num_train_epochs 10   --output_dir ./results/vit-large/resisc45/head_/seed_42 --model_name vit-large --finetuning_method head --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-5 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 1  --output_dir ./results/vit-large/resisc45/lora_1/seed_123 --model_name vit-large --finetuning_method lora --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 8  --output_dir ./results/vit-large/resisc45/lora_8/seed_123 --model_name vit-large --finetuning_method lora --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 1 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/resisc45/lora_1_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method lora --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/resisc45/lora_8_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method lora --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 1  --output_dir ./results/vit-large/resisc45/dora_1/seed_123 --model_name vit-large --finetuning_method dora --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 8  --output_dir ./results/vit-large/resisc45/dora_8/seed_123 --model_name vit-large --finetuning_method dora --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 1 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/resisc45/dora_1_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method dora --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --lora_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/resisc45/dora_8_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method dora --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --vera_rank 256  --output_dir ./results/vit-large/resisc45/vera_256/seed_123 --model_name vit-large --finetuning_method vera --dataset_name resisc45 --clf_learning_rate 3e-3 --other_learning_rate 7e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --boft_block_size 2 --boft_n_butterfly_factor 2  --output_dir ./results/vit-large/resisc45/boft_2_2/seed_123 --model_name vit-large --finetuning_method boft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --boft_block_size 4 --boft_n_butterfly_factor 4  --output_dir ./results/vit-large/resisc45/boft_4_4/seed_123 --model_name vit-large --finetuning_method boft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10   --output_dir ./results/vit-large/resisc45/svft_/seed_123 --model_name vit-large --finetuning_method svft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --svft_rank 2  --output_dir ./results/vit-large/resisc45/svft_2/seed_123 --model_name vit-large --finetuning_method svft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --svft_rank 4  --output_dir ./results/vit-large/resisc45/svft_4/seed_123 --model_name vit-large --finetuning_method svft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --svft_rank 8  --output_dir ./results/vit-large/resisc45/svft_8/seed_123 --model_name vit-large --finetuning_method svft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10  --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/resisc45/svft_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method svft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --svft_rank 2 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/resisc45/svft_2_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method svft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --svft_rank 4 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/resisc45/svft_4_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method svft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10 --svft_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/resisc45/svft_8_query_key_value_intermediate.dense_output.dense/seed_123 --model_name vit-large --finetuning_method svft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10   --output_dir ./results/vit-large/resisc45/full_/seed_123 --model_name vit-large --finetuning_method full --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-5 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 123 --num_train_epochs 10   --output_dir ./results/vit-large/resisc45/head_/seed_123 --model_name vit-large --finetuning_method head --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-5 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 1  --output_dir ./results/vit-large/resisc45/lora_1/seed_3076 --model_name vit-large --finetuning_method lora --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 8  --output_dir ./results/vit-large/resisc45/lora_8/seed_3076 --model_name vit-large --finetuning_method lora --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 1 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/resisc45/lora_1_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method lora --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/resisc45/lora_8_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method lora --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 1  --output_dir ./results/vit-large/resisc45/dora_1/seed_3076 --model_name vit-large --finetuning_method dora --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 8  --output_dir ./results/vit-large/resisc45/dora_8/seed_3076 --model_name vit-large --finetuning_method dora --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 1 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/resisc45/dora_1_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method dora --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --lora_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/resisc45/dora_8_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method dora --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --vera_rank 256  --output_dir ./results/vit-large/resisc45/vera_256/seed_3076 --model_name vit-large --finetuning_method vera --dataset_name resisc45 --clf_learning_rate 3e-3 --other_learning_rate 7e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --boft_block_size 2 --boft_n_butterfly_factor 2  --output_dir ./results/vit-large/resisc45/boft_2_2/seed_3076 --model_name vit-large --finetuning_method boft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --boft_block_size 4 --boft_n_butterfly_factor 4  --output_dir ./results/vit-large/resisc45/boft_4_4/seed_3076 --model_name vit-large --finetuning_method boft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10   --output_dir ./results/vit-large/resisc45/svft_/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --svft_rank 2  --output_dir ./results/vit-large/resisc45/svft_2/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --svft_rank 4  --output_dir ./results/vit-large/resisc45/svft_4/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --svft_rank 8  --output_dir ./results/vit-large/resisc45/svft_8/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10  --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/resisc45/svft_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --svft_rank 2 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/resisc45/svft_2_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 5e-2 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --svft_rank 4 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/resisc45/svft_4_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10 --svft_rank 8 --target_modules query key value intermediate.dense output.dense --output_dir ./results/vit-large/resisc45/svft_8_query_key_value_intermediate.dense_output.dense/seed_3076 --model_name vit-large --finetuning_method svft --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 5e-3 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10   --output_dir ./results/vit-large/resisc45/full_/seed_3076 --model_name vit-large --finetuning_method full --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-5 --warmup_ratio 0.1 --weight_decay 0.01

python finetuning_setup.py --evaluation_strategy epoch --save_strategy epoch --gradient_accumulation_steps 1 --logging_steps 10 --load_best_model_at_end True --save_total_limit 2 --metric_for_best_model eval_accuracy --label_names labels --remove_unused_columns False --per_device_train_batch_size 64 --per_device_eval_batch_size 256 --seed 3076 --num_train_epochs 10   --output_dir ./results/vit-large/resisc45/head_/seed_3076 --model_name vit-large --finetuning_method head --dataset_name resisc45 --clf_learning_rate 4e-3 --other_learning_rate 4e-5 --warmup_ratio 0.1 --weight_decay 0.01