modality: instance
noise_type: None
lr_exp_warmup_steps: 100
lr_total_iter_length: 40000
pretrained_model_name_or_path: facebook/vit-mae-huge
revision: None
variant: None
output_dir: model-finetuned/mae_full_e2e_ft_mean_mixed_sqrtsep_meansep_norm
seed: 500
train_batch_size: 2
num_train_epochs: 15
max_train_steps: 30000
gradient_accumulation_steps: 4
gradient_checkpointing: True
learning_rate: 6e-05
dataloader_num_workers: 0
adam_beta1: 0.9
adam_beta2: 0.999
adam_weight_decay: 0.01
adam_epsilon: 1e-08
max_grad_norm: 1.0
logging_dir: logs
mixed_precision: no
report_to: tensorboard
local_rank: 0
checkpointing_steps: 10000
checkpoints_total_limit: None
resume_from_checkpoint: None
enable_xformers_memory_efficient_attention: True
tracker_project_name: e2e-ft-diffusion
random_state_file: /nfs_share3/om/diffusion-e2e-ft/model-finetuned/stable_diffusion_e2e_ft_instance_10k/checkpoint-10000/random_states_0.pkl