| model_id_with_origin_paths: Wan-AI/Wan2.1-T2V-1.3B:diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.1-T2V-1.3B:Wan2.1_VAE.pth |
| use_helios_dit: true |
| helios_repo_root: /mnt/hdd/wjy/czy/DVD/Helios-main |
| helios_transformer_path: BestWishYsh/Helios-Distilled |
| helios_transformer_subfolder: transformer |
| helios_attention_backend: null |
| use_helios_history_memory: true |
| helios_latent_window_size: 9 |
| helios_history_sizes: |
| - 16 |
| - 2 |
| - 1 |
| use_helios_section_rollout: true |
| helios_section_rollout_mode: depth_history |
| helios_depth_history_train_mode: burn_in_predicted |
| helios_depth_history_burn_in_chunks: 2 |
| helios_section_stride_latent: 9 |
| helios_zero_history_timestep: true |
| helios_train_history_patches: true |
| trainable_models: dit |
| remove_prefix_in_ckpt: pipe.dit |
| mode: regression |
| denoise_step: 0.5 |
| training_target: x |
| lora_base_model: dit |
| lora_rank: 512 |
| grad_loss: true |
| grad_co: 0.5 |
| use_latent_flow: true |
| resume: true |
| training_state_dir: /mnt/hdd/wjy/czy/DVD/output/checkpoint-step-4000/ |
| load_optimizer: false |
| output_path: /mnt/hdd/wjy/czy/DVD/output_self_forcing/ |
| global_step: 4000 |
| num_epochs: 1000000 |
| gradient_accumulation_steps: 16 |
| validate_step: 100 |
| init_validate: true |
| clip_grad_norm: true |
| learning_rate: 2.0e-05 |
| use_gradient_checkpointing: true |
| log_step: 10 |
| debug_hang: true |
| debug_hang_start_step: 2300 |
| debug_hang_end_step: 2400 |
| debug_hang_every: 1 |
| debug_hang_tensor_stats: false |
| debug_dataset1_io: true |
| debug_dataset2_io: true |
| no_extra_frame: true |
| warmup_steps: 0 |
| mixed_precision: bf16 |
| train_all: false |
| denoise_strength: 1 |
| validation_scheduler_timesteps: 1 |
| dataloader_num_workers: 2 |
| train_ratio: 1 |
| train_microsteps_per_epoch: 100000 |
| video_dataset_repeat_safety: 1.25 |
| prob: |
| - 9 |
| - 1 |
| - 40 |
| - 10 |
| batch_size: 1 |
| train_data_dir_hypersim: /mnt/hdd/wjy/datasets/hypersim/hypersim_filtered/ |
| resolution_hypersim: |
| - 480 |
| - 640 |
| random_flip: true |
| norm_type: trunc_disparity |
| truncnorm_min: 0.02 |
| align_cam_normal: false |
| train_data_dir_vkitti: /mnt/hdd/wjy/datasets/vkitti/ |
| max_num_frame: 69 |
| min_num_frame: 69 |
| max_sample_stride: 1 |
| min_sample_stride: 1 |
| train_data_dir_ttr_vid: /mnt/hdd/wjy/datasets/ttr/ |
| train_data_dir_vkitti_vid: /mnt/hdd/wjy/datasets/vkitti/ |
| test_batch_size: 1 |
| test_max_num_frame: 69 |
| test_min_num_frame: 69 |
| test_max_sample_stride: 1 |
| test_min_sample_stride: 1 |
| kitti_vid_test_data_root: /mnt/hdd/wjy/datasets/kitti_depth/ |
| scannet_vid_test_data_root: /mnt/hdd/wjy/datasets/scannet/scannetoutput/ |
| scannet_split_ls: configs/scannetv1_test.txt |
| nyuv2_test_data_root: /mnt/hdd/wjy/datasets/nyuv2/ |
| scheduler_training_weight_type: equal |
| training_schedule_mode: default |
| training_scheduler_timesteps: 1 |
|
|
