| { |
| "adam_weight_decay": 0.0, |
| "alltracker_checkpoint": null, |
| "alltracker_device": null, |
| "alltracker_inference_iters": 4, |
| "alltracker_query_interval": 4, |
| "alltracker_root": null, |
| "alltracker_tiny": false, |
| "alltracker_window_len": 16, |
| "appearance_mask_rho": 0.2, |
| "batch_size": 1, |
| "cfg_scale": 5.0, |
| "connector_learning_rate": 0.0001, |
| "connector_update_steps": 1, |
| "connector_use_opd_loss": false, |
| "connector_warmup_steps": 50, |
| "deterministic_trajectory": false, |
| "device": "cuda:0", |
| "dist_backend": "nccl", |
| "dist_timeout_seconds": 7200, |
| "dit_learning_rate": 2e-06, |
| "dit_lr_warmup_steps": 50, |
| "dit_update_steps": 16, |
| "dtype": "bf16", |
| "ema_decay": 0.99, |
| "ema_dtype": "fp32", |
| "eval_at_start": false, |
| "eval_every": 0, |
| "eval_fps": 15, |
| "eval_height": 480, |
| "eval_num_frames": 81, |
| "eval_num_inference_steps": 50, |
| "eval_prompt": null, |
| "eval_prompts_json": null, |
| "eval_quality": 5, |
| "eval_seed": 0, |
| "eval_tile_size": [ |
| 30, |
| 52 |
| ], |
| "eval_tile_stride": [ |
| 15, |
| 26 |
| ], |
| "eval_tiled": true, |
| "eval_width": 832, |
| "fsdp": true, |
| "fsdp_sharding": "none", |
| "gradient_checkpointing": true, |
| "height": 480, |
| "image_connector_allow_resize": false, |
| "image_opd_timestep_scale": 1000.0, |
| "image_student_frame_channels": 16, |
| "image_teacher_latent_channels": 16, |
| "image_teacher_parallel_workers": 3, |
| "image_teacher_vae_scale_factor": 8, |
| "kl_dim_reduction": "mean", |
| "log_every": 1, |
| "lr_min_ratio": 0.1, |
| "lr_scheduler": "cosine", |
| "lr_warmup_ratio": 0.05, |
| "max_grad_norm_connector": 0.1, |
| "max_grad_norm_dit": 1.0, |
| "max_records": null, |
| "max_train_steps": 500, |
| "motion_anchor_teacher": "wan_base_motion_teacher", |
| "motion_correction_ratio_max": 0.1, |
| "motion_discriminator_device": "cuda:4", |
| "motion_gradient_score_ratio": 1.0, |
| "motion_guidance_scale": 0.05, |
| "motion_mask_blur_ks": 5, |
| "motion_mask_dilate_ks": 5, |
| "motion_mask_p_high": 95.0, |
| "motion_mask_p_low": 20.0, |
| "motion_mask_provider": "none", |
| "motion_mask_softness": 0.1, |
| "motion_mask_tau": 0.3, |
| "motion_preservation_weight": 1.0, |
| "motion_refinement": true, |
| "motion_refinement_dry_run": false, |
| "motion_refinement_dry_run_report": null, |
| "motion_refinement_timesteps_per_step": 1, |
| "motion_reward_checkpoint": "best", |
| "motion_reward_config": "/mnt/geminisgceph1/geminicephfs/mmsearch-ai-research/users/mandajiang/image-diffusion-opd-cross-arch/configs/motion_refinement/motion_d_full.json", |
| "motion_reward_device": "cuda:4", |
| "motion_reward_kind": "logit", |
| "motion_reward_pair_batch_size": 4, |
| "motion_reward_vae_device": "cuda:4", |
| "motion_reward_vae_gradient_checkpointing": true, |
| "motion_reward_vae_tiled": true, |
| "motion_t_max": 0.6, |
| "motion_t_min": 0.2, |
| "motion_t_ratio_clip": 4.0, |
| "negative_prompt": "\u8272\u8c03\u8273\u4e3d\uff0c\u8fc7\u66dd\uff0c\u9759\u6001\uff0c\u7ec6\u8282\u6a21\u7cca\u4e0d\u6e05\uff0c\u5b57\u5e55\uff0c\u98ce\u683c\uff0c\u4f5c\u54c1\uff0c\u753b\u4f5c\uff0c\u753b\u9762\uff0c\u9759\u6b62\uff0c\u6574\u4f53\u53d1\u7070\uff0c\u6700\u5dee\u8d28\u91cf\uff0c\u4f4e\u8d28\u91cf\uff0cJPEG\u538b\u7f29\u6b8b\u7559\uff0c\u4e11\u964b\u7684\uff0c\u6b8b\u7f3a\u7684\uff0c\u591a\u4f59\u7684\u624b\u6307\uff0c\u753b\u5f97\u4e0d\u597d\u7684\u624b\u90e8\uff0c\u753b\u5f97\u4e0d\u597d\u7684\u8138\u90e8\uff0c\u7578\u5f62\u7684\uff0c\u6bc1\u5bb9\u7684\uff0c\u5f62\u6001\u7578\u5f62\u7684\u80a2\u4f53\uff0c\u624b\u6307\u878d\u5408\uff0c\u9759\u6b62\u4e0d\u52a8\u7684\u753b\u9762\uff0c\u6742\u4e71\u7684\u80cc\u666f\uff0c\u4e09\u6761\u817f\uff0c\u80cc\u666f\u4eba\u5f88\u591a\uff0c\u5012\u7740\u8d70", |
| "num_frames": 81, |
| "num_inference_steps": 40, |
| "num_train_timesteps_per_step": 4, |
| "output_dir": "/mnt/geminisgceph1/geminicephfs/mmsearch-ai-research/users/mandajiang/image-diffusion-opd-cross-arch/outputs/motion_refined_multi_teacher_opd_81frame_8interval", |
| "prompt_field": "prompt", |
| "require_all_visible_gpus_used": true, |
| "resume_from_checkpoint": "latest", |
| "run_label": "multi_teacher_opd", |
| "save_every": 250, |
| "sde_noise_level": 0.7, |
| "sde_noise_schedule": "paper", |
| "seed": 42, |
| "sigma_shift": 5.0, |
| "single_copy_image_teacher": true, |
| "student_dit": [ |
| "/root/.cache/huggingface/hub/models--Wan-AI--Wan2.1-T2V-1.3B/snapshots/37ec512624d61f7aa208f7ea8140a131f93afc9a/diffusion_pytorch_model.safetensors" |
| ], |
| "student_dit2": null, |
| "student_t5": "/root/.cache/huggingface/hub/models--Wan-AI--Wan2.1-T2V-1.3B/snapshots/37ec512624d61f7aa208f7ea8140a131f93afc9a/models_t5_umt5-xxl-enc-bf16.pth", |
| "student_tokenizer_path": "/root/.cache/huggingface/hub/models--Wan-AI--Wan2.1-T2V-1.3B/snapshots/37ec512624d61f7aa208f7ea8140a131f93afc9a/google/umt5-xxl", |
| "student_use_usp": true, |
| "student_usp_attn_backend": "allgather_sdpa", |
| "student_vae": "/root/.cache/huggingface/hub/models--Wan-AI--Wan2.1-T2V-1.3B/snapshots/37ec512624d61f7aa208f7ea8140a131f93afc9a/Wan2.1_VAE.pth", |
| "switch_dit_boundary": 0.875, |
| "teacher_x0_height": null, |
| "teacher_x0_num_frames": null, |
| "teacher_x0_num_inference_steps": null, |
| "teacher_x0_width": null, |
| "teachers_config": "/mnt/geminisgceph1/geminicephfs/mmsearch-ai-research/users/mandajiang/image-diffusion-opd-cross-arch/multi_teacher_opd/teachers.example.json", |
| "train_jsonl": "/mnt/geminisgceph1/geminicephfs/mmsearch-ai-research/users/mandajiang/diffusion-opd-cross-arch/data/openvid_lvd_mixed_10k.jsonl", |
| "train_timestep_max_index": 37, |
| "train_timestep_min_index": 2, |
| "train_timestep_sampling": "random", |
| "use_ema": true, |
| "wandb_dir": null, |
| "wandb_entity": null, |
| "wandb_mode": null, |
| "wandb_project": null, |
| "wandb_run_name": null, |
| "width": 832 |
| } |
