diff --git a/.gitattributes b/.gitattributes index 9bf5df35e8f5abe5bbafb15beec8889a982cc794..3bdc8c7a7192bbb4fb75bc8adacb2dcfdf3dbc06 100644 --- a/.gitattributes +++ b/.gitattributes @@ -14295,3 +14295,11 @@ results/re10k_short_reverse_dememwm_full_monitor8_train731edd42758e_eval940b6fb7 results/re10k_short_reverse_dememwm_full_monitor8_train731edd42758e_eval940b6fb72bb1_step70000_job81244/videos/test_loop_true_reverse_h100_f199_v1_step00000000/bcc76b58e2585320.mp4 filter=lfs diff=lfs merge=lfs -text results/re10k_short_reverse_dememwm_full_monitor8_train731edd42758e_eval940b6fb72bb1_step70000_job81244/videos/test_loop_true_reverse_h100_f199_v1_step00000000/c49a14de01d7bd35.mp4 filter=lfs diff=lfs merge=lfs -text results/re10k_short_reverse_dememwm_full_monitor8_train731edd42758e_eval940b6fb72bb1_step70000_job81244/videos/test_loop_true_reverse_h100_f199_v1_step00000000/f287c679c250a841.mp4 filter=lfs diff=lfs merge=lfs -text +results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/0ece034988793847.mp4 filter=lfs diff=lfs merge=lfs -text +results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/1735d8d1b4015669.mp4 filter=lfs diff=lfs merge=lfs -text +results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/1cdfd3abfcc3a64e.mp4 filter=lfs diff=lfs merge=lfs -text +results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/6afff9609b65a604.mp4 filter=lfs diff=lfs merge=lfs -text +results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/b009fbe3950ebeb2.mp4 filter=lfs diff=lfs merge=lfs -text +results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/bcc76b58e2585320.mp4 filter=lfs diff=lfs merge=lfs -text +results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/c49a14de01d7bd35.mp4 filter=lfs diff=lfs merge=lfs -text +results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/f287c679c250a841.mp4 filter=lfs diff=lfs merge=lfs -text diff --git a/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/metrics/test_history.csv b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/metrics/test_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..2075190a353a3b049b1979f9341a25dcf70a1a3a --- /dev/null +++ b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/metrics/test_history.csv @@ -0,0 +1,2 @@ +method,namespace,global_step,sampling_steps,selected_scenes,generated_frames,overall/mse,overall/psnr,overall/ssim,overall/lpips,overall/rfid,final/mse,final/psnr,final/ssim,final/lpips,frames_000_025/mse,frames_000_025/psnr,frames_000_025/ssim,frames_000_025/lpips,frames_000_025/rfid,frames_025_050/mse,frames_025_050/psnr,frames_025_050/ssim,frames_025_050/lpips,frames_025_050/rfid,frames_050_075/mse,frames_050_075/psnr,frames_050_075/ssim,frames_050_075/lpips,frames_050_075/rfid,frames_075_100/mse,frames_075_100/psnr,frames_075_100/ssim,frames_075_100/lpips,frames_075_100/rfid,frames_000_050/mse,frames_000_050/psnr,frames_000_050/ssim,frames_000_050/lpips,frames_000_050/rfid,frames_050_100/mse,frames_050_100/psnr,frames_050_100/ssim,frames_050_100/lpips,frames_050_100/rfid,frame_bin_size,frame_bin_interval +worldmem,test,0,50,8,800,0.08874138444662094,12.040312767028809,0.3002903461456299,0.6288661360740662,166.0,0.09402408450841904,10.716129302978516,0.30140751600265503,0.8330078125,0.03203858807682991,17.623973846435547,0.44365641474723816,0.275991827249527,76.0,0.09714064002037048,10.674241065979004,0.2576228082180023,0.6598437428474426,198.0,0.12046549469232559,9.604694366455078,0.22269946336746216,0.7605859637260437,249.0,0.10532081872224808,10.258338928222656,0.2771826982498169,0.819042980670929,272.0,0.06458961218595505,14.14910888671875,0.35063961148262024,0.467917799949646,123.0,0.11289317160844803,9.931517601013184,0.24994109570980072,0.7898144721984863,246.0,25,"[start,end) generated-frame offset" diff --git a/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/metrics/test_monitor8_loop_true_reverse_h100_f199_v1_step00000000_metrics.json b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/metrics/test_monitor8_loop_true_reverse_h100_f199_v1_step00000000_metrics.json new file mode 100644 index 0000000000000000000000000000000000000000..b6e5179b7c76b85104e13fd8faeb9d8d0343184e --- /dev/null +++ b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/metrics/test_monitor8_loop_true_reverse_h100_f199_v1_step00000000_metrics.json @@ -0,0 +1,55 @@ +{ + "artifact_type": "re10k_benchmark_metrics", + "benchmark": "loop_true_reverse_h100_f199_v1", + "case_count": 8, + "generated_frames": 800, + "global_step": 0, + "memory_mode": "correct", + "method": "worldmem", + "metrics": { + "final/lpips": 0.8330078125, + "final/mse": 0.09402408450841904, + "final/psnr": 10.716129302978516, + "final/ssim": 0.30140751600265503, + "frames_000_025/lpips": 0.275991827249527, + "frames_000_025/mse": 0.03203858807682991, + "frames_000_025/psnr": 17.623973846435547, + "frames_000_025/rfid": 76.0, + "frames_000_025/ssim": 0.44365641474723816, + "frames_000_050/lpips": 0.467917799949646, + "frames_000_050/mse": 0.06458961218595505, + "frames_000_050/psnr": 14.14910888671875, + "frames_000_050/rfid": 123.0, + "frames_000_050/ssim": 0.35063961148262024, + "frames_025_050/lpips": 0.6598437428474426, + "frames_025_050/mse": 0.09714064002037048, + "frames_025_050/psnr": 10.674241065979004, + "frames_025_050/rfid": 198.0, + "frames_025_050/ssim": 0.2576228082180023, + "frames_050_075/lpips": 0.7605859637260437, + "frames_050_075/mse": 0.12046549469232559, + "frames_050_075/psnr": 9.604694366455078, + "frames_050_075/rfid": 249.0, + "frames_050_075/ssim": 0.22269946336746216, + "frames_050_100/lpips": 0.7898144721984863, + "frames_050_100/mse": 0.11289317160844803, + "frames_050_100/psnr": 9.931517601013184, + "frames_050_100/rfid": 246.0, + "frames_050_100/ssim": 0.24994109570980072, + "frames_075_100/lpips": 0.819042980670929, + "frames_075_100/mse": 0.10532081872224808, + "frames_075_100/psnr": 10.258338928222656, + "frames_075_100/rfid": 272.0, + "frames_075_100/ssim": 0.2771826982498169, + "overall/lpips": 0.6288661360740662, + "overall/mse": 0.08874138444662094, + "overall/psnr": 12.040312767028809, + "overall/rfid": 166.0, + "overall/ssim": 0.3002903461456299 + }, + "namespace": "test", + "sampling_steps": 50, + "scene_count": 8, + "schema_version": 1, + "scope": "monitor8" +} diff --git a/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/metrics/test_step00000000.json b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/metrics/test_step00000000.json new file mode 100644 index 0000000000000000000000000000000000000000..c3aa9d2ff504cc1bba2444d12bd1babb79398c16 --- /dev/null +++ b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/metrics/test_step00000000.json @@ -0,0 +1,49 @@ +{ + "final/lpips": 0.8330078125, + "final/mse": 0.09402408450841904, + "final/psnr": 10.716129302978516, + "final/ssim": 0.30140751600265503, + "frame_bin_interval": "[start,end) generated-frame offset", + "frame_bin_size": 25, + "frames_000_025/lpips": 0.275991827249527, + "frames_000_025/mse": 0.03203858807682991, + "frames_000_025/psnr": 17.623973846435547, + "frames_000_025/rfid": 76.0, + "frames_000_025/ssim": 0.44365641474723816, + "frames_000_050/lpips": 0.467917799949646, + "frames_000_050/mse": 0.06458961218595505, + "frames_000_050/psnr": 14.14910888671875, + "frames_000_050/rfid": 123.0, + "frames_000_050/ssim": 0.35063961148262024, + "frames_025_050/lpips": 0.6598437428474426, + "frames_025_050/mse": 0.09714064002037048, + "frames_025_050/psnr": 10.674241065979004, + "frames_025_050/rfid": 198.0, + "frames_025_050/ssim": 0.2576228082180023, + "frames_050_075/lpips": 0.7605859637260437, + "frames_050_075/mse": 0.12046549469232559, + "frames_050_075/psnr": 9.604694366455078, + "frames_050_075/rfid": 249.0, + "frames_050_075/ssim": 0.22269946336746216, + "frames_050_100/lpips": 0.7898144721984863, + "frames_050_100/mse": 0.11289317160844803, + "frames_050_100/psnr": 9.931517601013184, + "frames_050_100/rfid": 246.0, + "frames_050_100/ssim": 0.24994109570980072, + "frames_075_100/lpips": 0.819042980670929, + "frames_075_100/mse": 0.10532081872224808, + "frames_075_100/psnr": 10.258338928222656, + "frames_075_100/rfid": 272.0, + "frames_075_100/ssim": 0.2771826982498169, + "generated_frames": 800, + "global_step": 0, + "method": "worldmem", + "namespace": "test", + "overall/lpips": 0.6288661360740662, + "overall/mse": 0.08874138444662094, + "overall/psnr": 12.040312767028809, + "overall/rfid": 166.0, + "overall/ssim": 0.3002903461456299, + "sampling_steps": 50, + "selected_scenes": 8 +} diff --git a/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/resolved_config.yaml b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/resolved_config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..670a3dfb499f501ffd3902a8e98082f056e835e0 --- /dev/null +++ b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/resolved_config.yaml @@ -0,0 +1,400 @@ +experiment: + debug: false + tasks: + - training + num_nodes: 1 + training: + precision: bf16-mixed + compile: false + lr: 4.0e-05 + batch_size: 4 + max_epochs: -1 + max_steps: 100000 + max_time: null + data: + num_workers: 4 + shuffle: true + optim: + accumulate_grad_batches: 1 + gradient_clip_val: 1.0 + checkpointing: + every_n_train_steps: 2500 + every_n_epochs: null + train_time_interval: null + enable_version_counter: false + validation: + precision: bf16-mixed + compile: false + batch_size: 1 + val_every_n_step: 2500 + val_every_n_epoch: null + num_sanity_val_steps: 0 + limit_batch: 2 + data: + num_workers: 4 + shuffle: false + inference_mode: false + test: + precision: 16-mixed + compile: false + batch_size: 1 + limit_batch: 1.0 + data: + num_workers: 4 + shuffle: false + inference_mode: true + find_unused_parameters: false + reload_dataloaders_every_n_epochs: 1 + ema: + enable: false + decay: 0.9999 + validate_original_weights: false + _name: video_generation +dataset: + debug: false + save_dir: /proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/re10k_dfot/real-estate-10k + data_mean: + - - - 0.577 + - - - 0.517 + - - - 0.461 + data_std: + - - - 0.249 + - - - 0.249 + - - - 0.268 + latent: + enable: false + type: pre_sample + suffix: null + downsampling_factor: + - 1 + - 8 + num_channels: 4 + resolution: 256 + observation_shape: + - 3 + - 256 + - 256 + max_frames: 12 + n_frames: 12 + context_length: 10 + frame_skip: 1 + filter_min_len: null + external_cond_dim: 16 + external_cond_stack: false + external_cond_processing: null + preload: false + subdataset_size: null + num_eval_videos: 8 + maximize_training_data: false + augmentation: + frame_skip_increase: 0 + horizontal_flip_prob: 0 + reverse_prob: 0 + back_and_forth_prob: 0 + local_window_frames: 8 + training_frame_skip: 1 + training_history_frames: 10 + eval_history_frames: 100 + eval_min_generated_frames: 100 + eval_max_generated_frames: 100 + video_fps: 10.0 + seed: 0 + memory_selection: + enabled: true + selector_version: re10k_faithful_trajectory_h100_f399_v1 + method: worldmem + training_selection_mode: schedule_cache_direct + selector_cache_dir: /proj/cvl/users/x_fahkh2/DeMemWM/.exp_artifact/re10k_faithful_trajectory_h100_f399_v1/cache_worldmem_clipped_731edd42758e + selector_cache_source_commit: 731edd42758e7aec3b0bfd03ed52b72d716b65df + causal: true + max_anchor_frames: 0 + max_dynamic_frames: 4 + max_revisit_frames: 0 + training_dropout: 0.0 + local_context_exclusion_frames: 8 + match_position_scale: 2.0 + match_view_angle_degrees: 110.0 + evaluation_manifest: /proj/cvl/users/x_fahkh2/DeMemWM/evaluation/manifests/re10k_real_trajectory_h100_f399_scopes_v1.csv + evaluation_manifest_scope: monitor8 + evaluation_benchmark: loop_true_reverse_h100_f199_v1 + evaluation_benchmarks: + - loop_true_reverse_h100_f199_v1 + evaluation_memory_mode: correct + evaluation_case_counts: + loop_true_reverse_h100_f199_v1: 8 + evaluation_case_order: scene_manifest_then_path_ordinal + evaluation_case_seed: 1000003*metadata_index+path_ordinal + evaluation_trajectory: G0..G99,G99..G0 + evaluation_output_frames: 200 + evaluation_observed_output_interval: + - 0 + - 100 + evaluation_generated_output_interval: + - 100 + - 200 + evaluation_generated_segments: + return: + - 0 + - 100 + return_chunks: + - - 0 + - 25 + - - 25 + - 50 + - - 50 + - 75 + - - 75 + - 100 + evaluation_frame_bin_size: 25 + evaluation_turnaround_duplication: true + _name: video_realestate10k +algorithm: + debug: false + lr: 4.0e-05 + backbone: + name: u_vit3d_pose + channels: + - 128 + - 256 + - 576 + - 1152 + emb_channels: 1024 + patch_size: 2 + block_types: + - ResBlock + - ResBlock + - TransformerBlock + - TransformerBlock + block_dropouts: + - 0.0 + - 0.0 + - 0.1 + - 0.1 + num_updown_blocks: + - 3 + - 3 + - 6 + num_mid_blocks: 20 + num_heads: 9 + pos_emb_type: rope + use_checkpointing: + - false + - false + - false + - false + conditioning: + dim: null + external_cond_dropout: 0.1 + use_fourier_noise_embedding: true + x_shape: + - 3 + - 256 + - 256 + max_frames: 12 + n_frames: 12 + frame_skip: 1 + context_frames: 10 + latent: + enable: false + type: pre_sample + suffix: null + downsampling_factor: + - 1 + - 8 + num_channels: 4 + data_mean: + - - - 0.577 + - - - 0.517 + - - - 0.461 + data_std: + - - - 0.249 + - - - 0.249 + - - - 0.268 + external_cond_dim: 16 + external_cond_stack: false + external_cond_processing: null + compile: false + weight_decay: 0.002 + optimizer_beta: + - 0.9 + - 0.99 + lr_scheduler: + name: constant_with_warmup + num_warmup_steps: 1000 + num_training_steps: 100000 + noise_level: random_independent + uniform_future: + enabled: false + fixed_context: + enabled: false + indices: null + dropout: 0 + variable_context: + enabled: false + prob: 0 + dropout: 0 + chunk_size: -1 + scheduling_matrix: full_sequence + replacement: noisy_scale + diffusion: + is_continuous: true + timesteps: 1000 + beta_schedule: cosine_simple_diffusion + schedule_fn_kwargs: + shift: 1.0 + shifted: 0.125 + interpolated: false + use_causal_mask: false + clip_noise: 20.0 + objective: pred_v + loss_weighting: + strategy: sigmoid + snr_clip: 5.0 + cum_snr_decay: 0.9 + sigmoid_bias: -1.0 + sampling_timesteps: 50 + ddim_sampling_eta: 0.0 + reconstruction_guidance: 0.0 + training_schedule: + name: cosine + shift: 0.125 + precond_scale: 0.125 + vae: + pretrained_path: null + pretrained_kwargs: {} + use_fp16: true + batch_size: 2 + checkpoint: + reset_optimizer: false + strict: true + tasks: + prediction: + enabled: false + history_guidance: + name: conditional + keyframe_density: null + sliding_context_len: null + interpolation: + enabled: false + history_guidance: + name: conditional + max_batch_size: null + logging: + deterministic: 0 + loss_freq: 1 + grad_norm_freq: 0 + max_num_videos: 0 + n_metrics_frames: null + metrics: [] + metrics_batch_size: 16 + sanity_generation: false + raw_dir: null + camera_pose_conditioning: + normalize_by: first + bound: null + type: ray_encoding + memory: + local_window_frames: 8 + memory_selection: + enabled: true + selector_version: re10k_faithful_trajectory_h100_f399_v1 + method: worldmem + training_selection_mode: schedule_cache_direct + selector_cache_dir: /proj/cvl/users/x_fahkh2/DeMemWM/.exp_artifact/re10k_faithful_trajectory_h100_f399_v1/cache_worldmem_clipped_731edd42758e + selector_cache_source_commit: 731edd42758e7aec3b0bfd03ed52b72d716b65df + causal: true + max_anchor_frames: 0 + max_dynamic_frames: 4 + max_revisit_frames: 0 + training_dropout: 0.0 + local_context_exclusion_frames: 8 + match_position_scale: 2.0 + match_view_angle_degrees: 110.0 + memory_projection_mode: worldmem_reference_v1 + memory_ffn_ratio: 2 + memory_conditioning_input_dim: 1024 + memory_conditioning_hidden_dim: 128 + memory_role_embedding: false + memory_source_embedding: false + memory_stream_count_correction: false + memory_lora_rank: 8 + memory_lora_alpha: 8 + memory_output_gate_init: 0.0 + memory_camera_hidden_dim: 128 + memory_anchor_zero_geometry: false + memory_age_conditioning: false + reference_timestamp_conditioning: true + memory_attention_key_only_geometry: false + memory_noise: + enabled: false + training_seed: 0 + training: + max_steps: 100000 + num_gpus: 8 + batch_size_per_gpu: 4 + accumulate_grad_batches: 1 + learning_rate: 4.0e-05 + memory_learning_rate_stage_a: 4.0e-05 + memory_learning_rate_stage_b: 2.0e-05 + lora_learning_rate_stage_b: 5.0e-06 + stage_b_start_step: 8000 + stage_c_start_step: 30000 + min_lr_ratio: 0.1 + optimizer_betas: + - 0.9 + - 0.99 + weight_decay: 0.002 + gradient_clip_val: 1.0 + warmup_steps: 1000 + ema_decay: 0.9999 + frame_skip: 1 + checkpoint_every_n_steps: 2500 + validation_every_n_steps: 2500 + monitoring_steps: + - 2500 + - 5000 + - 7500 + - 10000 + - 12500 + - 15000 + - 20000 + - 25000 + - 30000 + - 35000 + - 40000 + - 50000 + - 60000 + - 70000 + - 80000 + - 90000 + - 100000 + evaluation: + selector_mode: online + benchmarks: + - loop_true_reverse_h100_f199_v1 + benchmark: loop_true_reverse_h100_f199_v1 + history_frames: 100 + min_generated_frames: 100 + max_generated_frames: 100 + frame_skip: 1 + sampling_steps: 50 + deterministic_seed: 0 + history_guidance: + name: conditional + reverse_loop_probe: false + turnaround_duplication: true + frame_bin_size: 25 + num_videos: 8 + memory_mode: correct + manifest_scope: monitor8 + active_benchmarks: + - loop_true_reverse_h100_f199_v1 + _name: dfot_worldmem_realestate10k +debug: false +wandb: + entity: turlin + project: worldmem + mode: offline +resume: null +load: null +name: dfot_worldmem_realestate10k diff --git a/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/run_manifest.json b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/run_manifest.json new file mode 100644 index 0000000000000000000000000000000000000000..5f1188a1bb7d72ec7a369ef1cf2fa6ac2a63937f --- /dev/null +++ b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/run_manifest.json @@ -0,0 +1,185 @@ +{ + "method": "worldmem", + "dataset_root": "/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/re10k_dfot/real-estate-10k", + "dfot_root": "/proj/cvl/users/x_fahkh2/WorldMem_Repro/DFoT-code", + "dfot_commit": "530f8bf4db91a21964993f214fa4050dbd529591", + "checkpoint": "/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/DFoT_RE10K.ckpt", + "local_frames": 8, + "memory_frames": { + "anchor": 0, + "dynamic": 4, + "revisit": 0 + }, + "training_selector_version": "re10k_faithful_trajectory_h100_f399_v1", + "training_selector_method": "worldmem", + "training_selector_cache_source_commit": "731edd42758e7aec3b0bfd03ed52b72d716b65df", + "memory_output_gate_init": 0.0, + "batch_size_per_gpu": 4, + "accumulate_grad_batches": 1, + "effective_global_batch_size": 32, + "optimizer_lr_schedule": { + "name": "stagewise_warmup_stable_cosine_floor", + "warmup_steps": 1000, + "stage_b_start_step": 8000, + "decay_start_step": 30000, + "terminal_step": 100000, + "min_lr_ratio": 0.1 + }, + "runtime": { + "gradient_checkpointing": [ + false, + false, + false, + false + ], + "split_frozen_memory_stream": false, + "structural_memory_autograd": true + }, + "wandb": { + "mode": "offline", + "entity": "turlin", + "project": "worldmem", + "run_id": null + }, + "evaluation_history_frames": 100, + "evaluation_max_generated_frames": 100, + "evaluation_num_videos": 8, + "evaluation_sampling_steps": 50, + "evaluation_metrics": [ + "mse", + "psnr", + "ssim", + "lpips", + "rfid" + ], + "training_time_eval_batches_per_rank": 2, + "post_training_evaluation": true, + "faithful_profile": { + "method": "worldmem", + "capacity": "clipped", + "selector_version": "re10k_faithful_trajectory_h100_f399_v1", + "lifecycle": "group1_59306102" + }, + "trainable_parameter_manifest": "/proj/cvl/users/x_fahkh2/DeMemWM/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/trainable_parameter_manifest.json", + "training_batch_composition": { + "global_natural": 8, + "global_loop": 24, + "rank_order": [ + "natural", + "loop", + "loop", + "loop" + ] + }, + "training_trajectory": { + "source_mapping": "physical_start+[0..199,199..0]", + "virtual_positions": 400, + "query_frames": 8, + "phase_bounds_half_open": [ + [ + 100, + 159 + ], + [ + 159, + 218 + ], + [ + 218, + 277 + ], + [ + 277, + 335 + ], + [ + 335, + 393 + ] + ] + }, + "evaluation_weight_source": "ema", + "post_training_checkpoint_policy": null, + "worldmem_git": { + "branch": "re10k-dfot-integration", + "head": "940b6fb72bb1543b1e5bfad72baf837140a603f0", + "status": [ + "?? results/" + ] + }, + "profile_stage_b": false, + "schedule_step_offset": 0, + "algorithm_config": "/proj/cvl/users/x_fahkh2/DeMemWM/configurations/algorithm/dfot_worldmem_faithful_f200b200_clipped_realestate10k.yaml", + "training_seed": 0, + "evaluation_seed": 0, + "training_seed_label": "seed0", + "checkpoint_kind": "ema-v2", + "evaluation_checkpoint": "/proj/cvl/users/x_fahkh2/DeMemWM/results/dfot_worldmem_clipped_f200b200_seed0_g32_100k_731edd42758e/checkpoints/step=30000.ckpt", + "evaluation_manifest": "/proj/cvl/users/x_fahkh2/DeMemWM/evaluation/manifests/re10k_real_trajectory_h100_f399_scopes_v1.csv", + "evaluation_manifest_scope": "monitor8", + "evaluation_benchmark": "loop_true_reverse_h100_f199_v1", + "evaluation_benchmarks": [ + "loop_true_reverse_h100_f199_v1" + ], + "evaluation_memory_mode": "correct", + "evaluation_case_counts": { + "loop_true_reverse_h100_f199_v1": 8 + }, + "evaluation_case_order": "scene_manifest_then_path_ordinal", + "evaluation_case_seed": "1000003*metadata_index+path_ordinal", + "evaluation_trajectory": "G0..G99,G99..G0", + "evaluation_output_frames": 200, + "evaluation_observed_output_interval": [ + 0, + 100 + ], + "evaluation_generated_output_interval": [ + 100, + 200 + ], + "evaluation_generated_segments": { + "return": [ + 0, + 100 + ], + "return_chunks": [ + [ + 0, + 25 + ], + [ + 25, + 50 + ], + [ + 50, + 75 + ], + [ + 75, + 100 + ] + ] + }, + "evaluation_frame_bin_size": 25, + "evaluation_turnaround_duplication": true, + "monitoring_steps": [ + 2500, + 5000, + 7500, + 10000, + 12500, + 15000, + 20000, + 25000, + 30000, + 35000, + 40000, + 50000, + 60000, + 70000, + 80000, + 90000, + 100000 + ] +} diff --git a/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/test_loop_true_reverse_h100_f199_v1_case_manifest.json b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/test_loop_true_reverse_h100_f199_v1_case_manifest.json new file mode 100644 index 0000000000000000000000000000000000000000..921b11ce76cb27a8bff91508ac5022f4cc370b97 --- /dev/null +++ b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/test_loop_true_reverse_h100_f199_v1_case_manifest.json @@ -0,0 +1,120 @@ +{ + "split": "test", + "evaluation_seed": 0, + "manifest": "/proj/cvl/users/x_fahkh2/DeMemWM/evaluation/manifests/re10k_real_trajectory_h100_f399_scopes_v1.csv", + "manifest_scope": "monitor8", + "history_frames": 100, + "benchmark": "loop_true_reverse_h100_f199_v1", + "scene_count": 8, + "scene_order": [ + { + "metadata_index": 4865, + "stem": "b009fbe3950ebeb2" + }, + { + "metadata_index": 418, + "stem": "0ece034988793847" + }, + { + "metadata_index": 5224, + "stem": "bcc76b58e2585320" + }, + { + "metadata_index": 640, + "stem": "1735d8d1b4015669" + }, + { + "metadata_index": 2952, + "stem": "6afff9609b65a604" + }, + { + "metadata_index": 5459, + "stem": "c49a14de01d7bd35" + }, + { + "metadata_index": 6743, + "stem": "f287c679c250a841" + }, + { + "metadata_index": 806, + "stem": "1cdfd3abfcc3a64e" + } + ], + "case_count": 8, + "case_order": "scene_manifest_then_path_ordinal", + "total_generated_frames": 800, + "cases": [ + { + "benchmark": "loop_true_reverse_h100_f199_v1", + "metadata_index": 4865, + "scene_id": "b009fbe3950ebeb2", + "path_name": "true_trajectory_g0_g99_g99_g0", + "path_ordinal": 8, + "generated_frames": 100, + "case_seed": 4865014603 + }, + { + "benchmark": "loop_true_reverse_h100_f199_v1", + "metadata_index": 418, + "scene_id": "0ece034988793847", + "path_name": "true_trajectory_g0_g99_g99_g0", + "path_ordinal": 8, + "generated_frames": 100, + "case_seed": 418001262 + }, + { + "benchmark": "loop_true_reverse_h100_f199_v1", + "metadata_index": 5224, + "scene_id": "bcc76b58e2585320", + "path_name": "true_trajectory_g0_g99_g99_g0", + "path_ordinal": 8, + "generated_frames": 100, + "case_seed": 5224015680 + }, + { + "benchmark": "loop_true_reverse_h100_f199_v1", + "metadata_index": 640, + "scene_id": "1735d8d1b4015669", + "path_name": "true_trajectory_g0_g99_g99_g0", + "path_ordinal": 8, + "generated_frames": 100, + "case_seed": 640001928 + }, + { + "benchmark": "loop_true_reverse_h100_f199_v1", + "metadata_index": 2952, + "scene_id": "6afff9609b65a604", + "path_name": "true_trajectory_g0_g99_g99_g0", + "path_ordinal": 8, + "generated_frames": 100, + "case_seed": 2952008864 + }, + { + "benchmark": "loop_true_reverse_h100_f199_v1", + "metadata_index": 5459, + "scene_id": "c49a14de01d7bd35", + "path_name": "true_trajectory_g0_g99_g99_g0", + "path_ordinal": 8, + "generated_frames": 100, + "case_seed": 5459016385 + }, + { + "benchmark": "loop_true_reverse_h100_f199_v1", + "metadata_index": 6743, + "scene_id": "f287c679c250a841", + "path_name": "true_trajectory_g0_g99_g99_g0", + "path_ordinal": 8, + "generated_frames": 100, + "case_seed": 6743020237 + }, + { + "benchmark": "loop_true_reverse_h100_f199_v1", + "metadata_index": 806, + "scene_id": "1cdfd3abfcc3a64e", + "path_name": "true_trajectory_g0_g99_g99_g0", + "path_ordinal": 8, + "generated_frames": 100, + "case_seed": 806002426 + } + ] +} diff --git a/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/test_scene_manifest.json b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/test_scene_manifest.json new file mode 100644 index 0000000000000000000000000000000000000000..bdfcdd818e360506525020d0ea948e2141904931 --- /dev/null +++ b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/test_scene_manifest.json @@ -0,0 +1,42 @@ +{ + "split": "test", + "evaluation_seed": 0, + "manifest": "/proj/cvl/users/x_fahkh2/DeMemWM/evaluation/manifests/re10k_real_trajectory_h100_f399_scopes_v1.csv", + "manifest_scope": "monitor8", + "history_frames": 100, + "scene_count": 8, + "scenes": [ + { + "metadata_index": 4865, + "stem": "b009fbe3950ebeb2" + }, + { + "metadata_index": 418, + "stem": "0ece034988793847" + }, + { + "metadata_index": 5224, + "stem": "bcc76b58e2585320" + }, + { + "metadata_index": 640, + "stem": "1735d8d1b4015669" + }, + { + "metadata_index": 2952, + "stem": "6afff9609b65a604" + }, + { + "metadata_index": 5459, + "stem": "c49a14de01d7bd35" + }, + { + "metadata_index": 6743, + "stem": "f287c679c250a841" + }, + { + "metadata_index": 806, + "stem": "1cdfd3abfcc3a64e" + } + ] +} diff --git a/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/trainable_parameter_manifest.json b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/trainable_parameter_manifest.json new file mode 100644 index 0000000000000000000000000000000000000000..5512b64fd775153a9cce5c815400ad4450ad567d --- /dev/null +++ b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/trainable_parameter_manifest.json @@ -0,0 +1,2049 @@ +{ + "base_lora_token_scope": "all_tokens", + "base_names": [ + "diffusion_model.model.noise_level_pos_embedding.embedding.linear_1.weight", + "diffusion_model.model.noise_level_pos_embedding.embedding.linear_1.bias", + "diffusion_model.model.noise_level_pos_embedding.embedding.linear_2.weight", + "diffusion_model.model.noise_level_pos_embedding.embedding.linear_2.bias", + "diffusion_model.model.external_cond_embedding.patch_embedder.proj.weight", + "diffusion_model.model.external_cond_embedding.patch_embedder.proj.bias", + "diffusion_model.model.embed_input.proj.weight", + "diffusion_model.model.embed_input.proj.bias", + "diffusion_model.model.project_output.proj.weight", + "diffusion_model.model.project_output.proj.bias", + "diffusion_model.model.down_blocks.0.0.emb_layer.weight", + "diffusion_model.model.down_blocks.0.0.emb_layer.bias", + "diffusion_model.model.down_blocks.0.0.in_layers.0.weight", + "diffusion_model.model.down_blocks.0.0.in_layers.0.bias", + "diffusion_model.model.down_blocks.0.0.in_layers.2.weight", + "diffusion_model.model.down_blocks.0.0.in_layers.2.bias", + "diffusion_model.model.down_blocks.0.0.out_norm.weight", + "diffusion_model.model.down_blocks.0.0.out_norm.bias", + "diffusion_model.model.down_blocks.0.0.out_rest.1.weight", + "diffusion_model.model.down_blocks.0.0.out_rest.1.bias", + "diffusion_model.model.down_blocks.0.1.emb_layer.weight", + "diffusion_model.model.down_blocks.0.1.emb_layer.bias", + "diffusion_model.model.down_blocks.0.1.in_layers.0.weight", + "diffusion_model.model.down_blocks.0.1.in_layers.0.bias", + "diffusion_model.model.down_blocks.0.1.in_layers.2.weight", + "diffusion_model.model.down_blocks.0.1.in_layers.2.bias", + "diffusion_model.model.down_blocks.0.1.out_norm.weight", + "diffusion_model.model.down_blocks.0.1.out_norm.bias", + "diffusion_model.model.down_blocks.0.1.out_rest.1.weight", + "diffusion_model.model.down_blocks.0.1.out_rest.1.bias", + "diffusion_model.model.down_blocks.0.2.emb_layer.weight", + "diffusion_model.model.down_blocks.0.2.emb_layer.bias", + "diffusion_model.model.down_blocks.0.2.in_layers.0.weight", + "diffusion_model.model.down_blocks.0.2.in_layers.0.bias", + "diffusion_model.model.down_blocks.0.2.in_layers.2.weight", + "diffusion_model.model.down_blocks.0.2.in_layers.2.bias", + "diffusion_model.model.down_blocks.0.2.out_norm.weight", + "diffusion_model.model.down_blocks.0.2.out_norm.bias", + "diffusion_model.model.down_blocks.0.2.out_rest.1.weight", + "diffusion_model.model.down_blocks.0.2.out_rest.1.bias", + "diffusion_model.model.down_blocks.0.3.conv.weight", + "diffusion_model.model.down_blocks.0.3.conv.bias", + "diffusion_model.model.down_blocks.1.0.emb_layer.weight", + "diffusion_model.model.down_blocks.1.0.emb_layer.bias", + "diffusion_model.model.down_blocks.1.0.in_layers.0.weight", + "diffusion_model.model.down_blocks.1.0.in_layers.0.bias", + "diffusion_model.model.down_blocks.1.0.in_layers.2.weight", + "diffusion_model.model.down_blocks.1.0.in_layers.2.bias", + "diffusion_model.model.down_blocks.1.0.out_norm.weight", + "diffusion_model.model.down_blocks.1.0.out_norm.bias", + "diffusion_model.model.down_blocks.1.0.out_rest.1.weight", + "diffusion_model.model.down_blocks.1.0.out_rest.1.bias", + "diffusion_model.model.down_blocks.1.1.emb_layer.weight", + "diffusion_model.model.down_blocks.1.1.emb_layer.bias", + "diffusion_model.model.down_blocks.1.1.in_layers.0.weight", + "diffusion_model.model.down_blocks.1.1.in_layers.0.bias", + "diffusion_model.model.down_blocks.1.1.in_layers.2.weight", + "diffusion_model.model.down_blocks.1.1.in_layers.2.bias", + "diffusion_model.model.down_blocks.1.1.out_norm.weight", + "diffusion_model.model.down_blocks.1.1.out_norm.bias", + "diffusion_model.model.down_blocks.1.1.out_rest.1.weight", + "diffusion_model.model.down_blocks.1.1.out_rest.1.bias", + "diffusion_model.model.down_blocks.1.2.emb_layer.weight", + "diffusion_model.model.down_blocks.1.2.emb_layer.bias", + "diffusion_model.model.down_blocks.1.2.in_layers.0.weight", + "diffusion_model.model.down_blocks.1.2.in_layers.0.bias", + "diffusion_model.model.down_blocks.1.2.in_layers.2.weight", + "diffusion_model.model.down_blocks.1.2.in_layers.2.bias", + "diffusion_model.model.down_blocks.1.2.out_norm.weight", + "diffusion_model.model.down_blocks.1.2.out_norm.bias", + "diffusion_model.model.down_blocks.1.2.out_rest.1.weight", + "diffusion_model.model.down_blocks.1.2.out_rest.1.bias", + "diffusion_model.model.down_blocks.1.3.conv.weight", + "diffusion_model.model.down_blocks.1.3.conv.bias", + "diffusion_model.model.down_blocks.2.0.norm.emb_layer.weight", + "diffusion_model.model.down_blocks.2.0.norm.emb_layer.bias", + "diffusion_model.model.down_blocks.2.0.norm.norm.weight", + "diffusion_model.model.down_blocks.2.0.fused_attn_mlp_proj.weight", + "diffusion_model.model.down_blocks.2.0.fused_attn_mlp_proj.bias", + "diffusion_model.model.down_blocks.2.0.q_norm.weight", + "diffusion_model.model.down_blocks.2.0.k_norm.weight", + "diffusion_model.model.down_blocks.2.0.attn_out.weight", + "diffusion_model.model.down_blocks.2.0.attn_out.bias", + "diffusion_model.model.down_blocks.2.0.mlp_out.2.weight", + "diffusion_model.model.down_blocks.2.0.mlp_out.2.bias", + "diffusion_model.model.down_blocks.2.1.norm.emb_layer.weight", + "diffusion_model.model.down_blocks.2.1.norm.emb_layer.bias", + "diffusion_model.model.down_blocks.2.1.norm.norm.weight", + "diffusion_model.model.down_blocks.2.1.fused_attn_mlp_proj.weight", + "diffusion_model.model.down_blocks.2.1.fused_attn_mlp_proj.bias", + "diffusion_model.model.down_blocks.2.1.q_norm.weight", + "diffusion_model.model.down_blocks.2.1.k_norm.weight", + "diffusion_model.model.down_blocks.2.1.attn_out.weight", + "diffusion_model.model.down_blocks.2.1.attn_out.bias", + "diffusion_model.model.down_blocks.2.1.mlp_out.2.weight", + "diffusion_model.model.down_blocks.2.1.mlp_out.2.bias", + "diffusion_model.model.down_blocks.2.2.norm.emb_layer.weight", + "diffusion_model.model.down_blocks.2.2.norm.emb_layer.bias", + "diffusion_model.model.down_blocks.2.2.norm.norm.weight", + "diffusion_model.model.down_blocks.2.2.fused_attn_mlp_proj.weight", + "diffusion_model.model.down_blocks.2.2.fused_attn_mlp_proj.bias", + "diffusion_model.model.down_blocks.2.2.q_norm.weight", + "diffusion_model.model.down_blocks.2.2.k_norm.weight", + "diffusion_model.model.down_blocks.2.2.attn_out.weight", + "diffusion_model.model.down_blocks.2.2.attn_out.bias", + "diffusion_model.model.down_blocks.2.2.mlp_out.2.weight", + "diffusion_model.model.down_blocks.2.2.mlp_out.2.bias", + "diffusion_model.model.down_blocks.2.3.norm.emb_layer.weight", + "diffusion_model.model.down_blocks.2.3.norm.emb_layer.bias", + "diffusion_model.model.down_blocks.2.3.norm.norm.weight", + "diffusion_model.model.down_blocks.2.3.fused_attn_mlp_proj.weight", + "diffusion_model.model.down_blocks.2.3.fused_attn_mlp_proj.bias", + "diffusion_model.model.down_blocks.2.3.q_norm.weight", + "diffusion_model.model.down_blocks.2.3.k_norm.weight", + "diffusion_model.model.down_blocks.2.3.attn_out.weight", + "diffusion_model.model.down_blocks.2.3.attn_out.bias", + "diffusion_model.model.down_blocks.2.3.mlp_out.2.weight", + "diffusion_model.model.down_blocks.2.3.mlp_out.2.bias", + "diffusion_model.model.down_blocks.2.4.norm.emb_layer.weight", + "diffusion_model.model.down_blocks.2.4.norm.emb_layer.bias", + "diffusion_model.model.down_blocks.2.4.norm.norm.weight", + "diffusion_model.model.down_blocks.2.4.fused_attn_mlp_proj.weight", + "diffusion_model.model.down_blocks.2.4.fused_attn_mlp_proj.bias", + "diffusion_model.model.down_blocks.2.4.q_norm.weight", + "diffusion_model.model.down_blocks.2.4.k_norm.weight", + "diffusion_model.model.down_blocks.2.4.attn_out.weight", + "diffusion_model.model.down_blocks.2.4.attn_out.bias", + "diffusion_model.model.down_blocks.2.4.mlp_out.2.weight", + "diffusion_model.model.down_blocks.2.4.mlp_out.2.bias", + "diffusion_model.model.down_blocks.2.5.norm.emb_layer.weight", + "diffusion_model.model.down_blocks.2.5.norm.emb_layer.bias", + "diffusion_model.model.down_blocks.2.5.norm.norm.weight", + "diffusion_model.model.down_blocks.2.5.fused_attn_mlp_proj.weight", + "diffusion_model.model.down_blocks.2.5.fused_attn_mlp_proj.bias", + "diffusion_model.model.down_blocks.2.5.q_norm.weight", + "diffusion_model.model.down_blocks.2.5.k_norm.weight", + "diffusion_model.model.down_blocks.2.5.attn_out.weight", + "diffusion_model.model.down_blocks.2.5.attn_out.bias", + "diffusion_model.model.down_blocks.2.5.mlp_out.2.weight", + "diffusion_model.model.down_blocks.2.5.mlp_out.2.bias", + "diffusion_model.model.down_blocks.2.6.conv.weight", + "diffusion_model.model.down_blocks.2.6.conv.bias", + "diffusion_model.model.up_blocks.0.0.conv.weight", + "diffusion_model.model.up_blocks.0.0.conv.bias", + "diffusion_model.model.up_blocks.0.1.norm.emb_layer.weight", + "diffusion_model.model.up_blocks.0.1.norm.emb_layer.bias", + "diffusion_model.model.up_blocks.0.1.norm.norm.weight", + "diffusion_model.model.up_blocks.0.1.fused_attn_mlp_proj.weight", + "diffusion_model.model.up_blocks.0.1.fused_attn_mlp_proj.bias", + "diffusion_model.model.up_blocks.0.1.q_norm.weight", + "diffusion_model.model.up_blocks.0.1.k_norm.weight", + "diffusion_model.model.up_blocks.0.1.attn_out.weight", + "diffusion_model.model.up_blocks.0.1.attn_out.bias", + "diffusion_model.model.up_blocks.0.1.mlp_out.2.weight", + "diffusion_model.model.up_blocks.0.1.mlp_out.2.bias", + "diffusion_model.model.up_blocks.0.2.norm.emb_layer.weight", + "diffusion_model.model.up_blocks.0.2.norm.emb_layer.bias", + "diffusion_model.model.up_blocks.0.2.norm.norm.weight", + "diffusion_model.model.up_blocks.0.2.fused_attn_mlp_proj.weight", + "diffusion_model.model.up_blocks.0.2.fused_attn_mlp_proj.bias", + "diffusion_model.model.up_blocks.0.2.q_norm.weight", + "diffusion_model.model.up_blocks.0.2.k_norm.weight", + "diffusion_model.model.up_blocks.0.2.attn_out.weight", + "diffusion_model.model.up_blocks.0.2.attn_out.bias", + "diffusion_model.model.up_blocks.0.2.mlp_out.2.weight", + "diffusion_model.model.up_blocks.0.2.mlp_out.2.bias", + "diffusion_model.model.up_blocks.0.3.norm.emb_layer.weight", + "diffusion_model.model.up_blocks.0.3.norm.emb_layer.bias", + "diffusion_model.model.up_blocks.0.3.norm.norm.weight", + "diffusion_model.model.up_blocks.0.3.fused_attn_mlp_proj.weight", + "diffusion_model.model.up_blocks.0.3.fused_attn_mlp_proj.bias", + "diffusion_model.model.up_blocks.0.3.q_norm.weight", + "diffusion_model.model.up_blocks.0.3.k_norm.weight", + "diffusion_model.model.up_blocks.0.3.attn_out.weight", + "diffusion_model.model.up_blocks.0.3.attn_out.bias", + "diffusion_model.model.up_blocks.0.3.mlp_out.2.weight", + "diffusion_model.model.up_blocks.0.3.mlp_out.2.bias", + "diffusion_model.model.up_blocks.0.4.norm.emb_layer.weight", + "diffusion_model.model.up_blocks.0.4.norm.emb_layer.bias", + "diffusion_model.model.up_blocks.0.4.norm.norm.weight", + "diffusion_model.model.up_blocks.0.4.fused_attn_mlp_proj.weight", + "diffusion_model.model.up_blocks.0.4.fused_attn_mlp_proj.bias", + "diffusion_model.model.up_blocks.0.4.q_norm.weight", + "diffusion_model.model.up_blocks.0.4.k_norm.weight", + "diffusion_model.model.up_blocks.0.4.attn_out.weight", + "diffusion_model.model.up_blocks.0.4.attn_out.bias", + "diffusion_model.model.up_blocks.0.4.mlp_out.2.weight", + "diffusion_model.model.up_blocks.0.4.mlp_out.2.bias", + "diffusion_model.model.up_blocks.0.5.norm.emb_layer.weight", + "diffusion_model.model.up_blocks.0.5.norm.emb_layer.bias", + "diffusion_model.model.up_blocks.0.5.norm.norm.weight", + "diffusion_model.model.up_blocks.0.5.fused_attn_mlp_proj.weight", + "diffusion_model.model.up_blocks.0.5.fused_attn_mlp_proj.bias", + "diffusion_model.model.up_blocks.0.5.q_norm.weight", + "diffusion_model.model.up_blocks.0.5.k_norm.weight", + "diffusion_model.model.up_blocks.0.5.attn_out.weight", + "diffusion_model.model.up_blocks.0.5.attn_out.bias", + "diffusion_model.model.up_blocks.0.5.mlp_out.2.weight", + "diffusion_model.model.up_blocks.0.5.mlp_out.2.bias", + "diffusion_model.model.up_blocks.0.6.norm.emb_layer.weight", + "diffusion_model.model.up_blocks.0.6.norm.emb_layer.bias", + "diffusion_model.model.up_blocks.0.6.norm.norm.weight", + "diffusion_model.model.up_blocks.0.6.fused_attn_mlp_proj.weight", + "diffusion_model.model.up_blocks.0.6.fused_attn_mlp_proj.bias", + "diffusion_model.model.up_blocks.0.6.q_norm.weight", + "diffusion_model.model.up_blocks.0.6.k_norm.weight", + "diffusion_model.model.up_blocks.0.6.attn_out.weight", + "diffusion_model.model.up_blocks.0.6.attn_out.bias", + "diffusion_model.model.up_blocks.0.6.mlp_out.2.weight", + "diffusion_model.model.up_blocks.0.6.mlp_out.2.bias", + "diffusion_model.model.up_blocks.1.0.conv.weight", + "diffusion_model.model.up_blocks.1.0.conv.bias", + "diffusion_model.model.up_blocks.1.1.emb_layer.weight", + "diffusion_model.model.up_blocks.1.1.emb_layer.bias", + "diffusion_model.model.up_blocks.1.1.in_layers.0.weight", + "diffusion_model.model.up_blocks.1.1.in_layers.0.bias", + "diffusion_model.model.up_blocks.1.1.in_layers.2.weight", + "diffusion_model.model.up_blocks.1.1.in_layers.2.bias", + "diffusion_model.model.up_blocks.1.1.out_norm.weight", + "diffusion_model.model.up_blocks.1.1.out_norm.bias", + "diffusion_model.model.up_blocks.1.1.out_rest.1.weight", + "diffusion_model.model.up_blocks.1.1.out_rest.1.bias", + "diffusion_model.model.up_blocks.1.2.emb_layer.weight", + "diffusion_model.model.up_blocks.1.2.emb_layer.bias", + "diffusion_model.model.up_blocks.1.2.in_layers.0.weight", + "diffusion_model.model.up_blocks.1.2.in_layers.0.bias", + "diffusion_model.model.up_blocks.1.2.in_layers.2.weight", + "diffusion_model.model.up_blocks.1.2.in_layers.2.bias", + "diffusion_model.model.up_blocks.1.2.out_norm.weight", + "diffusion_model.model.up_blocks.1.2.out_norm.bias", + "diffusion_model.model.up_blocks.1.2.out_rest.1.weight", + "diffusion_model.model.up_blocks.1.2.out_rest.1.bias", + "diffusion_model.model.up_blocks.1.3.emb_layer.weight", + "diffusion_model.model.up_blocks.1.3.emb_layer.bias", + "diffusion_model.model.up_blocks.1.3.in_layers.0.weight", + "diffusion_model.model.up_blocks.1.3.in_layers.0.bias", + "diffusion_model.model.up_blocks.1.3.in_layers.2.weight", + "diffusion_model.model.up_blocks.1.3.in_layers.2.bias", + "diffusion_model.model.up_blocks.1.3.out_norm.weight", + "diffusion_model.model.up_blocks.1.3.out_norm.bias", + "diffusion_model.model.up_blocks.1.3.out_rest.1.weight", + "diffusion_model.model.up_blocks.1.3.out_rest.1.bias", + "diffusion_model.model.up_blocks.2.0.conv.weight", + "diffusion_model.model.up_blocks.2.0.conv.bias", + "diffusion_model.model.up_blocks.2.1.emb_layer.weight", + "diffusion_model.model.up_blocks.2.1.emb_layer.bias", + "diffusion_model.model.up_blocks.2.1.in_layers.0.weight", + "diffusion_model.model.up_blocks.2.1.in_layers.0.bias", + "diffusion_model.model.up_blocks.2.1.in_layers.2.weight", + "diffusion_model.model.up_blocks.2.1.in_layers.2.bias", + "diffusion_model.model.up_blocks.2.1.out_norm.weight", + "diffusion_model.model.up_blocks.2.1.out_norm.bias", + "diffusion_model.model.up_blocks.2.1.out_rest.1.weight", + "diffusion_model.model.up_blocks.2.1.out_rest.1.bias", + "diffusion_model.model.up_blocks.2.2.emb_layer.weight", + "diffusion_model.model.up_blocks.2.2.emb_layer.bias", + "diffusion_model.model.up_blocks.2.2.in_layers.0.weight", + "diffusion_model.model.up_blocks.2.2.in_layers.0.bias", + "diffusion_model.model.up_blocks.2.2.in_layers.2.weight", + "diffusion_model.model.up_blocks.2.2.in_layers.2.bias", + "diffusion_model.model.up_blocks.2.2.out_norm.weight", + "diffusion_model.model.up_blocks.2.2.out_norm.bias", + "diffusion_model.model.up_blocks.2.2.out_rest.1.weight", + "diffusion_model.model.up_blocks.2.2.out_rest.1.bias", + "diffusion_model.model.up_blocks.2.3.emb_layer.weight", + "diffusion_model.model.up_blocks.2.3.emb_layer.bias", + "diffusion_model.model.up_blocks.2.3.in_layers.0.weight", + "diffusion_model.model.up_blocks.2.3.in_layers.0.bias", + "diffusion_model.model.up_blocks.2.3.in_layers.2.weight", + "diffusion_model.model.up_blocks.2.3.in_layers.2.bias", + "diffusion_model.model.up_blocks.2.3.out_norm.weight", + "diffusion_model.model.up_blocks.2.3.out_norm.bias", + "diffusion_model.model.up_blocks.2.3.out_rest.1.weight", + "diffusion_model.model.up_blocks.2.3.out_rest.1.bias", + "diffusion_model.model.mid_blocks.0.norm.emb_layer.weight", + "diffusion_model.model.mid_blocks.0.norm.emb_layer.bias", + "diffusion_model.model.mid_blocks.0.norm.norm.weight", + "diffusion_model.model.mid_blocks.0.fused_attn_mlp_proj.weight", + "diffusion_model.model.mid_blocks.0.fused_attn_mlp_proj.bias", + "diffusion_model.model.mid_blocks.0.q_norm.weight", + "diffusion_model.model.mid_blocks.0.k_norm.weight", + "diffusion_model.model.mid_blocks.0.attn_out.weight", + "diffusion_model.model.mid_blocks.0.attn_out.bias", + "diffusion_model.model.mid_blocks.0.mlp_out.2.weight", + "diffusion_model.model.mid_blocks.0.mlp_out.2.bias", + "diffusion_model.model.mid_blocks.1.norm.emb_layer.weight", + "diffusion_model.model.mid_blocks.1.norm.emb_layer.bias", + "diffusion_model.model.mid_blocks.1.norm.norm.weight", + "diffusion_model.model.mid_blocks.1.fused_attn_mlp_proj.weight", + "diffusion_model.model.mid_blocks.1.fused_attn_mlp_proj.bias", + "diffusion_model.model.mid_blocks.1.q_norm.weight", + "diffusion_model.model.mid_blocks.1.k_norm.weight", + "diffusion_model.model.mid_blocks.1.attn_out.weight", + "diffusion_model.model.mid_blocks.1.attn_out.bias", + "diffusion_model.model.mid_blocks.1.mlp_out.2.weight", + "diffusion_model.model.mid_blocks.1.mlp_out.2.bias", + "diffusion_model.model.mid_blocks.2.norm.emb_layer.weight", + "diffusion_model.model.mid_blocks.2.norm.emb_layer.bias", + "diffusion_model.model.mid_blocks.2.norm.norm.weight", + "diffusion_model.model.mid_blocks.2.fused_attn_mlp_proj.weight", + "diffusion_model.model.mid_blocks.2.fused_attn_mlp_proj.bias", + "diffusion_model.model.mid_blocks.2.q_norm.weight", + "diffusion_model.model.mid_blocks.2.k_norm.weight", + "diffusion_model.model.mid_blocks.2.attn_out.weight", + "diffusion_model.model.mid_blocks.2.attn_out.bias", + "diffusion_model.model.mid_blocks.2.mlp_out.2.weight", + "diffusion_model.model.mid_blocks.2.mlp_out.2.bias", + "diffusion_model.model.mid_blocks.3.norm.emb_layer.weight", + "diffusion_model.model.mid_blocks.3.norm.emb_layer.bias", + "diffusion_model.model.mid_blocks.3.norm.norm.weight", + "diffusion_model.model.mid_blocks.3.fused_attn_mlp_proj.weight", + "diffusion_model.model.mid_blocks.3.fused_attn_mlp_proj.bias", + "diffusion_model.model.mid_blocks.3.q_norm.weight", + "diffusion_model.model.mid_blocks.3.k_norm.weight", + "diffusion_model.model.mid_blocks.3.attn_out.weight", + "diffusion_model.model.mid_blocks.3.attn_out.bias", + "diffusion_model.model.mid_blocks.3.mlp_out.2.weight", + "diffusion_model.model.mid_blocks.3.mlp_out.2.bias", + "diffusion_model.model.mid_blocks.4.norm.emb_layer.weight", + "diffusion_model.model.mid_blocks.4.norm.emb_layer.bias", + "diffusion_model.model.mid_blocks.4.norm.norm.weight", + "diffusion_model.model.mid_blocks.4.fused_attn_mlp_proj.weight", + "diffusion_model.model.mid_blocks.4.fused_attn_mlp_proj.bias", + "diffusion_model.model.mid_blocks.4.q_norm.weight", + "diffusion_model.model.mid_blocks.4.k_norm.weight", + "diffusion_model.model.mid_blocks.4.attn_out.weight", + "diffusion_model.model.mid_blocks.4.attn_out.bias", + "diffusion_model.model.mid_blocks.4.mlp_out.2.weight", + "diffusion_model.model.mid_blocks.4.mlp_out.2.bias", + "diffusion_model.model.mid_blocks.5.norm.emb_layer.weight", + "diffusion_model.model.mid_blocks.5.norm.emb_layer.bias", + "diffusion_model.model.mid_blocks.5.norm.norm.weight", + "diffusion_model.model.mid_blocks.5.fused_attn_mlp_proj.weight", + "diffusion_model.model.mid_blocks.5.fused_attn_mlp_proj.bias", + "diffusion_model.model.mid_blocks.5.q_norm.weight", + "diffusion_model.model.mid_blocks.5.k_norm.weight", + "diffusion_model.model.mid_blocks.5.attn_out.weight", + "diffusion_model.model.mid_blocks.5.attn_out.bias", + "diffusion_model.model.mid_blocks.5.mlp_out.2.weight", + "diffusion_model.model.mid_blocks.5.mlp_out.2.bias", + "diffusion_model.model.mid_blocks.6.norm.emb_layer.weight", + "diffusion_model.model.mid_blocks.6.norm.emb_layer.bias", + "diffusion_model.model.mid_blocks.6.norm.norm.weight", + "diffusion_model.model.mid_blocks.6.fused_attn_mlp_proj.weight", + "diffusion_model.model.mid_blocks.6.fused_attn_mlp_proj.bias", + "diffusion_model.model.mid_blocks.6.q_norm.weight", + "diffusion_model.model.mid_blocks.6.k_norm.weight", + "diffusion_model.model.mid_blocks.6.attn_out.weight", + "diffusion_model.model.mid_blocks.6.attn_out.bias", + "diffusion_model.model.mid_blocks.6.mlp_out.2.weight", + "diffusion_model.model.mid_blocks.6.mlp_out.2.bias", + "diffusion_model.model.mid_blocks.7.norm.emb_layer.weight", + "diffusion_model.model.mid_blocks.7.norm.emb_layer.bias", + "diffusion_model.model.mid_blocks.7.norm.norm.weight", + "diffusion_model.model.mid_blocks.7.fused_attn_mlp_proj.weight", + "diffusion_model.model.mid_blocks.7.fused_attn_mlp_proj.bias", + "diffusion_model.model.mid_blocks.7.q_norm.weight", + "diffusion_model.model.mid_blocks.7.k_norm.weight", + "diffusion_model.model.mid_blocks.7.attn_out.weight", + "diffusion_model.model.mid_blocks.7.attn_out.bias", + "diffusion_model.model.mid_blocks.7.mlp_out.2.weight", + "diffusion_model.model.mid_blocks.7.mlp_out.2.bias", + "diffusion_model.model.mid_blocks.8.norm.emb_layer.weight", + "diffusion_model.model.mid_blocks.8.norm.emb_layer.bias", + "diffusion_model.model.mid_blocks.8.norm.norm.weight", + "diffusion_model.model.mid_blocks.8.fused_attn_mlp_proj.weight", + "diffusion_model.model.mid_blocks.8.fused_attn_mlp_proj.bias", + "diffusion_model.model.mid_blocks.8.q_norm.weight", + "diffusion_model.model.mid_blocks.8.k_norm.weight", + "diffusion_model.model.mid_blocks.8.attn_out.weight", + "diffusion_model.model.mid_blocks.8.attn_out.bias", + "diffusion_model.model.mid_blocks.8.mlp_out.2.weight", + "diffusion_model.model.mid_blocks.8.mlp_out.2.bias", + "diffusion_model.model.mid_blocks.9.norm.emb_layer.weight", + "diffusion_model.model.mid_blocks.9.norm.emb_layer.bias", + "diffusion_model.model.mid_blocks.9.norm.norm.weight", + "diffusion_model.model.mid_blocks.9.fused_attn_mlp_proj.weight", + "diffusion_model.model.mid_blocks.9.fused_attn_mlp_proj.bias", + "diffusion_model.model.mid_blocks.9.q_norm.weight", + "diffusion_model.model.mid_blocks.9.k_norm.weight", + "diffusion_model.model.mid_blocks.9.attn_out.weight", + "diffusion_model.model.mid_blocks.9.attn_out.bias", + "diffusion_model.model.mid_blocks.9.mlp_out.2.weight", + "diffusion_model.model.mid_blocks.9.mlp_out.2.bias", + "diffusion_model.model.mid_blocks.10.norm.emb_layer.weight", + "diffusion_model.model.mid_blocks.10.norm.emb_layer.bias", + "diffusion_model.model.mid_blocks.10.norm.norm.weight", + "diffusion_model.model.mid_blocks.10.fused_attn_mlp_proj.weight", + "diffusion_model.model.mid_blocks.10.fused_attn_mlp_proj.bias", + "diffusion_model.model.mid_blocks.10.q_norm.weight", + "diffusion_model.model.mid_blocks.10.k_norm.weight", + "diffusion_model.model.mid_blocks.10.attn_out.weight", + "diffusion_model.model.mid_blocks.10.attn_out.bias", + "diffusion_model.model.mid_blocks.10.mlp_out.2.weight", + "diffusion_model.model.mid_blocks.10.mlp_out.2.bias", + "diffusion_model.model.mid_blocks.11.norm.emb_layer.weight", + "diffusion_model.model.mid_blocks.11.norm.emb_layer.bias", + "diffusion_model.model.mid_blocks.11.norm.norm.weight", + "diffusion_model.model.mid_blocks.11.fused_attn_mlp_proj.weight", + "diffusion_model.model.mid_blocks.11.fused_attn_mlp_proj.bias", + "diffusion_model.model.mid_blocks.11.q_norm.weight", + "diffusion_model.model.mid_blocks.11.k_norm.weight", + "diffusion_model.model.mid_blocks.11.attn_out.weight", + "diffusion_model.model.mid_blocks.11.attn_out.bias", + "diffusion_model.model.mid_blocks.11.mlp_out.2.weight", + "diffusion_model.model.mid_blocks.11.mlp_out.2.bias", + "diffusion_model.model.mid_blocks.12.norm.emb_layer.weight", + "diffusion_model.model.mid_blocks.12.norm.emb_layer.bias", + "diffusion_model.model.mid_blocks.12.norm.norm.weight", + "diffusion_model.model.mid_blocks.12.fused_attn_mlp_proj.weight", + "diffusion_model.model.mid_blocks.12.fused_attn_mlp_proj.bias", + "diffusion_model.model.mid_blocks.12.q_norm.weight", + "diffusion_model.model.mid_blocks.12.k_norm.weight", + "diffusion_model.model.mid_blocks.12.attn_out.weight", + "diffusion_model.model.mid_blocks.12.attn_out.bias", + "diffusion_model.model.mid_blocks.12.mlp_out.2.weight", + "diffusion_model.model.mid_blocks.12.mlp_out.2.bias", + "diffusion_model.model.mid_blocks.13.norm.emb_layer.weight", + "diffusion_model.model.mid_blocks.13.norm.emb_layer.bias", + "diffusion_model.model.mid_blocks.13.norm.norm.weight", + "diffusion_model.model.mid_blocks.13.fused_attn_mlp_proj.weight", + "diffusion_model.model.mid_blocks.13.fused_attn_mlp_proj.bias", + "diffusion_model.model.mid_blocks.13.q_norm.weight", + "diffusion_model.model.mid_blocks.13.k_norm.weight", + "diffusion_model.model.mid_blocks.13.attn_out.weight", + "diffusion_model.model.mid_blocks.13.attn_out.bias", + "diffusion_model.model.mid_blocks.13.mlp_out.2.weight", + "diffusion_model.model.mid_blocks.13.mlp_out.2.bias", + "diffusion_model.model.mid_blocks.14.norm.emb_layer.weight", + "diffusion_model.model.mid_blocks.14.norm.emb_layer.bias", + "diffusion_model.model.mid_blocks.14.norm.norm.weight", + "diffusion_model.model.mid_blocks.14.fused_attn_mlp_proj.weight", + "diffusion_model.model.mid_blocks.14.fused_attn_mlp_proj.bias", + "diffusion_model.model.mid_blocks.14.q_norm.weight", + "diffusion_model.model.mid_blocks.14.k_norm.weight", + "diffusion_model.model.mid_blocks.14.attn_out.weight", + "diffusion_model.model.mid_blocks.14.attn_out.bias", + "diffusion_model.model.mid_blocks.14.mlp_out.2.weight", + "diffusion_model.model.mid_blocks.14.mlp_out.2.bias", + "diffusion_model.model.mid_blocks.15.norm.emb_layer.weight", + "diffusion_model.model.mid_blocks.15.norm.emb_layer.bias", + "diffusion_model.model.mid_blocks.15.norm.norm.weight", + "diffusion_model.model.mid_blocks.15.fused_attn_mlp_proj.weight", + "diffusion_model.model.mid_blocks.15.fused_attn_mlp_proj.bias", + "diffusion_model.model.mid_blocks.15.q_norm.weight", + "diffusion_model.model.mid_blocks.15.k_norm.weight", + "diffusion_model.model.mid_blocks.15.attn_out.weight", + "diffusion_model.model.mid_blocks.15.attn_out.bias", + "diffusion_model.model.mid_blocks.15.mlp_out.2.weight", + "diffusion_model.model.mid_blocks.15.mlp_out.2.bias", + "diffusion_model.model.mid_blocks.16.norm.emb_layer.weight", + "diffusion_model.model.mid_blocks.16.norm.emb_layer.bias", + "diffusion_model.model.mid_blocks.16.norm.norm.weight", + "diffusion_model.model.mid_blocks.16.fused_attn_mlp_proj.weight", + "diffusion_model.model.mid_blocks.16.fused_attn_mlp_proj.bias", + "diffusion_model.model.mid_blocks.16.q_norm.weight", + "diffusion_model.model.mid_blocks.16.k_norm.weight", + "diffusion_model.model.mid_blocks.16.attn_out.weight", + "diffusion_model.model.mid_blocks.16.attn_out.bias", + "diffusion_model.model.mid_blocks.16.mlp_out.2.weight", + "diffusion_model.model.mid_blocks.16.mlp_out.2.bias", + "diffusion_model.model.mid_blocks.17.norm.emb_layer.weight", + "diffusion_model.model.mid_blocks.17.norm.emb_layer.bias", + "diffusion_model.model.mid_blocks.17.norm.norm.weight", + "diffusion_model.model.mid_blocks.17.fused_attn_mlp_proj.weight", + "diffusion_model.model.mid_blocks.17.fused_attn_mlp_proj.bias", + "diffusion_model.model.mid_blocks.17.q_norm.weight", + "diffusion_model.model.mid_blocks.17.k_norm.weight", + "diffusion_model.model.mid_blocks.17.attn_out.weight", + "diffusion_model.model.mid_blocks.17.attn_out.bias", + "diffusion_model.model.mid_blocks.17.mlp_out.2.weight", + "diffusion_model.model.mid_blocks.17.mlp_out.2.bias", + "diffusion_model.model.mid_blocks.18.norm.emb_layer.weight", + "diffusion_model.model.mid_blocks.18.norm.emb_layer.bias", + "diffusion_model.model.mid_blocks.18.norm.norm.weight", + "diffusion_model.model.mid_blocks.18.fused_attn_mlp_proj.weight", + "diffusion_model.model.mid_blocks.18.fused_attn_mlp_proj.bias", + "diffusion_model.model.mid_blocks.18.q_norm.weight", + "diffusion_model.model.mid_blocks.18.k_norm.weight", + "diffusion_model.model.mid_blocks.18.attn_out.weight", + "diffusion_model.model.mid_blocks.18.attn_out.bias", + "diffusion_model.model.mid_blocks.18.mlp_out.2.weight", + "diffusion_model.model.mid_blocks.18.mlp_out.2.bias", + "diffusion_model.model.mid_blocks.19.norm.emb_layer.weight", + "diffusion_model.model.mid_blocks.19.norm.emb_layer.bias", + "diffusion_model.model.mid_blocks.19.norm.norm.weight", + "diffusion_model.model.mid_blocks.19.fused_attn_mlp_proj.weight", + "diffusion_model.model.mid_blocks.19.fused_attn_mlp_proj.bias", + "diffusion_model.model.mid_blocks.19.q_norm.weight", + "diffusion_model.model.mid_blocks.19.k_norm.weight", + "diffusion_model.model.mid_blocks.19.attn_out.weight", + "diffusion_model.model.mid_blocks.19.attn_out.bias", + "diffusion_model.model.mid_blocks.19.mlp_out.2.weight", + "diffusion_model.model.mid_blocks.19.mlp_out.2.bias" + ], + "base_numel": 458818051, + "base_tensor_count": 494, + "dfot_update": "frozen_lora_stage8k", + "dynamic_selection_enabled": true, + "evaluation_weight_source": "ema", + "families": { + "ffn_input": 61106688, + "ffn_output": 61076736, + "pose_cond_mlp": 30553344, + "pose_encoder": 224704, + "reference_adaln": 23182848, + "reference_output": 30553344, + "reference_qkv": 91570176, + "timestamp_encoder": 2104704, + "timestep_encoder": 131200 + }, + "frozen_numel": 458818051, + "full_model_numel": 762916035, + "layers": [ + { + "block": "down_blocks.2.0", + "dim": 576, + "ffn_input": 664704, + "ffn_output": 664128, + "pose_cond_mlp": 332352, + "reference_adaln": 445824, + "reference_output": 332352, + "reference_qkv": 995328, + "tensor_count": 11, + "total": 3434688 + }, + { + "block": "down_blocks.2.1", + "dim": 576, + "ffn_input": 664704, + "ffn_output": 664128, + "pose_cond_mlp": 332352, + "reference_adaln": 445824, + "reference_output": 332352, + "reference_qkv": 995328, + "tensor_count": 11, + "total": 3434688 + }, + { + "block": "down_blocks.2.2", + "dim": 576, + "ffn_input": 664704, + "ffn_output": 664128, + "pose_cond_mlp": 332352, + "reference_adaln": 445824, + "reference_output": 332352, + "reference_qkv": 995328, + "tensor_count": 11, + "total": 3434688 + }, + { + "block": "down_blocks.2.3", + "dim": 576, + "ffn_input": 664704, + "ffn_output": 664128, + "pose_cond_mlp": 332352, + "reference_adaln": 445824, + "reference_output": 332352, + "reference_qkv": 995328, + "tensor_count": 11, + "total": 3434688 + }, + { + "block": "down_blocks.2.4", + "dim": 576, + "ffn_input": 664704, + "ffn_output": 664128, + "pose_cond_mlp": 332352, + "reference_adaln": 445824, + "reference_output": 332352, + "reference_qkv": 995328, + "tensor_count": 11, + "total": 3434688 + }, + { + "block": "down_blocks.2.5", + "dim": 576, + "ffn_input": 664704, + "ffn_output": 664128, + "pose_cond_mlp": 332352, + "reference_adaln": 445824, + "reference_output": 332352, + "reference_qkv": 995328, + "tensor_count": 11, + "total": 3434688 + }, + { + "block": "up_blocks.0.1", + "dim": 576, + "ffn_input": 664704, + "ffn_output": 664128, + "pose_cond_mlp": 332352, + "reference_adaln": 445824, + "reference_output": 332352, + "reference_qkv": 995328, + "tensor_count": 11, + "total": 3434688 + }, + { + "block": "up_blocks.0.2", + "dim": 576, + "ffn_input": 664704, + "ffn_output": 664128, + "pose_cond_mlp": 332352, + "reference_adaln": 445824, + "reference_output": 332352, + "reference_qkv": 995328, + "tensor_count": 11, + "total": 3434688 + }, + { + "block": "up_blocks.0.3", + "dim": 576, + "ffn_input": 664704, + "ffn_output": 664128, + "pose_cond_mlp": 332352, + "reference_adaln": 445824, + "reference_output": 332352, + "reference_qkv": 995328, + "tensor_count": 11, + "total": 3434688 + }, + { + "block": "up_blocks.0.4", + "dim": 576, + "ffn_input": 664704, + "ffn_output": 664128, + "pose_cond_mlp": 332352, + "reference_adaln": 445824, + "reference_output": 332352, + "reference_qkv": 995328, + "tensor_count": 11, + "total": 3434688 + }, + { + "block": "up_blocks.0.5", + "dim": 576, + "ffn_input": 664704, + "ffn_output": 664128, + "pose_cond_mlp": 332352, + "reference_adaln": 445824, + "reference_output": 332352, + "reference_qkv": 995328, + "tensor_count": 11, + "total": 3434688 + }, + { + "block": "up_blocks.0.6", + "dim": 576, + "ffn_input": 664704, + "ffn_output": 664128, + "pose_cond_mlp": 332352, + "reference_adaln": 445824, + "reference_output": 332352, + "reference_qkv": 995328, + "tensor_count": 11, + "total": 3434688 + }, + { + "block": "mid_blocks.0", + "dim": 1152, + "ffn_input": 2656512, + "ffn_output": 2655360, + "pose_cond_mlp": 1328256, + "reference_adaln": 891648, + "reference_output": 1328256, + "reference_qkv": 3981312, + "tensor_count": 11, + "total": 12841344 + }, + { + "block": "mid_blocks.1", + "dim": 1152, + "ffn_input": 2656512, + "ffn_output": 2655360, + "pose_cond_mlp": 1328256, + "reference_adaln": 891648, + "reference_output": 1328256, + "reference_qkv": 3981312, + "tensor_count": 11, + "total": 12841344 + }, + { + "block": "mid_blocks.2", + "dim": 1152, + "ffn_input": 2656512, + "ffn_output": 2655360, + "pose_cond_mlp": 1328256, + "reference_adaln": 891648, + "reference_output": 1328256, + "reference_qkv": 3981312, + "tensor_count": 11, + "total": 12841344 + }, + { + "block": "mid_blocks.3", + "dim": 1152, + "ffn_input": 2656512, + "ffn_output": 2655360, + "pose_cond_mlp": 1328256, + "reference_adaln": 891648, + "reference_output": 1328256, + "reference_qkv": 3981312, + "tensor_count": 11, + "total": 12841344 + }, + { + "block": "mid_blocks.4", + "dim": 1152, + "ffn_input": 2656512, + "ffn_output": 2655360, + "pose_cond_mlp": 1328256, + "reference_adaln": 891648, + "reference_output": 1328256, + "reference_qkv": 3981312, + "tensor_count": 11, + "total": 12841344 + }, + { + "block": "mid_blocks.5", + "dim": 1152, + "ffn_input": 2656512, + "ffn_output": 2655360, + "pose_cond_mlp": 1328256, + "reference_adaln": 891648, + "reference_output": 1328256, + "reference_qkv": 3981312, + "tensor_count": 11, + "total": 12841344 + }, + { + "block": "mid_blocks.6", + "dim": 1152, + "ffn_input": 2656512, + "ffn_output": 2655360, + "pose_cond_mlp": 1328256, + "reference_adaln": 891648, + "reference_output": 1328256, + "reference_qkv": 3981312, + "tensor_count": 11, + "total": 12841344 + }, + { + "block": "mid_blocks.7", + "dim": 1152, + "ffn_input": 2656512, + "ffn_output": 2655360, + "pose_cond_mlp": 1328256, + "reference_adaln": 891648, + "reference_output": 1328256, + "reference_qkv": 3981312, + "tensor_count": 11, + "total": 12841344 + }, + { + "block": "mid_blocks.8", + "dim": 1152, + "ffn_input": 2656512, + "ffn_output": 2655360, + "pose_cond_mlp": 1328256, + "reference_adaln": 891648, + "reference_output": 1328256, + "reference_qkv": 3981312, + "tensor_count": 11, + "total": 12841344 + }, + { + "block": "mid_blocks.9", + "dim": 1152, + "ffn_input": 2656512, + "ffn_output": 2655360, + "pose_cond_mlp": 1328256, + "reference_adaln": 891648, + "reference_output": 1328256, + "reference_qkv": 3981312, + "tensor_count": 11, + "total": 12841344 + }, + { + "block": "mid_blocks.10", + "dim": 1152, + "ffn_input": 2656512, + "ffn_output": 2655360, + "pose_cond_mlp": 1328256, + "reference_adaln": 891648, + "reference_output": 1328256, + "reference_qkv": 3981312, + "tensor_count": 11, + "total": 12841344 + }, + { + "block": "mid_blocks.11", + "dim": 1152, + "ffn_input": 2656512, + "ffn_output": 2655360, + "pose_cond_mlp": 1328256, + "reference_adaln": 891648, + "reference_output": 1328256, + "reference_qkv": 3981312, + "tensor_count": 11, + "total": 12841344 + }, + { + "block": "mid_blocks.12", + "dim": 1152, + "ffn_input": 2656512, + "ffn_output": 2655360, + "pose_cond_mlp": 1328256, + "reference_adaln": 891648, + "reference_output": 1328256, + "reference_qkv": 3981312, + "tensor_count": 11, + "total": 12841344 + }, + { + "block": "mid_blocks.13", + "dim": 1152, + "ffn_input": 2656512, + "ffn_output": 2655360, + "pose_cond_mlp": 1328256, + "reference_adaln": 891648, + "reference_output": 1328256, + "reference_qkv": 3981312, + "tensor_count": 11, + "total": 12841344 + }, + { + "block": "mid_blocks.14", + "dim": 1152, + "ffn_input": 2656512, + "ffn_output": 2655360, + "pose_cond_mlp": 1328256, + "reference_adaln": 891648, + "reference_output": 1328256, + "reference_qkv": 3981312, + "tensor_count": 11, + "total": 12841344 + }, + { + "block": "mid_blocks.15", + "dim": 1152, + "ffn_input": 2656512, + "ffn_output": 2655360, + "pose_cond_mlp": 1328256, + "reference_adaln": 891648, + "reference_output": 1328256, + "reference_qkv": 3981312, + "tensor_count": 11, + "total": 12841344 + }, + { + "block": "mid_blocks.16", + "dim": 1152, + "ffn_input": 2656512, + "ffn_output": 2655360, + "pose_cond_mlp": 1328256, + "reference_adaln": 891648, + "reference_output": 1328256, + "reference_qkv": 3981312, + "tensor_count": 11, + "total": 12841344 + }, + { + "block": "mid_blocks.17", + "dim": 1152, + "ffn_input": 2656512, + "ffn_output": 2655360, + "pose_cond_mlp": 1328256, + "reference_adaln": 891648, + "reference_output": 1328256, + "reference_qkv": 3981312, + "tensor_count": 11, + "total": 12841344 + }, + { + "block": "mid_blocks.18", + "dim": 1152, + "ffn_input": 2656512, + "ffn_output": 2655360, + "pose_cond_mlp": 1328256, + "reference_adaln": 891648, + "reference_output": 1328256, + "reference_qkv": 3981312, + "tensor_count": 11, + "total": 12841344 + }, + { + "block": "mid_blocks.19", + "dim": 1152, + "ffn_input": 2656512, + "ffn_output": 2655360, + "pose_cond_mlp": 1328256, + "reference_adaln": 891648, + "reference_output": 1328256, + "reference_qkv": 3981312, + "tensor_count": 11, + "total": 12841344 + } + ], + "lifecycle": "group1_59306102", + "lora_names": [ + "diffusion_model.model.down_blocks.2.0.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.down_blocks.2.0.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.down_blocks.2.0.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.down_blocks.2.0.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.down_blocks.2.0.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.down_blocks.2.0.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.down_blocks.2.1.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.down_blocks.2.1.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.down_blocks.2.1.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.down_blocks.2.1.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.down_blocks.2.1.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.down_blocks.2.1.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.down_blocks.2.2.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.down_blocks.2.2.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.down_blocks.2.2.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.down_blocks.2.2.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.down_blocks.2.2.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.down_blocks.2.2.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.down_blocks.2.3.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.down_blocks.2.3.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.down_blocks.2.3.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.down_blocks.2.3.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.down_blocks.2.3.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.down_blocks.2.3.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.down_blocks.2.4.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.down_blocks.2.4.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.down_blocks.2.4.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.down_blocks.2.4.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.down_blocks.2.4.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.down_blocks.2.4.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.down_blocks.2.5.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.down_blocks.2.5.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.down_blocks.2.5.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.down_blocks.2.5.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.down_blocks.2.5.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.down_blocks.2.5.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.up_blocks.0.1.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.up_blocks.0.1.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.up_blocks.0.1.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.up_blocks.0.1.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.up_blocks.0.1.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.up_blocks.0.1.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.up_blocks.0.2.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.up_blocks.0.2.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.up_blocks.0.2.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.up_blocks.0.2.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.up_blocks.0.2.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.up_blocks.0.2.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.up_blocks.0.3.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.up_blocks.0.3.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.up_blocks.0.3.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.up_blocks.0.3.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.up_blocks.0.3.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.up_blocks.0.3.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.up_blocks.0.4.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.up_blocks.0.4.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.up_blocks.0.4.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.up_blocks.0.4.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.up_blocks.0.4.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.up_blocks.0.4.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.up_blocks.0.5.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.up_blocks.0.5.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.up_blocks.0.5.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.up_blocks.0.5.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.up_blocks.0.5.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.up_blocks.0.5.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.up_blocks.0.6.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.up_blocks.0.6.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.up_blocks.0.6.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.up_blocks.0.6.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.up_blocks.0.6.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.up_blocks.0.6.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.0.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.0.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.0.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.0.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.0.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.0.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.1.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.1.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.1.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.1.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.1.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.1.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.2.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.2.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.2.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.2.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.2.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.2.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.3.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.3.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.3.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.3.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.3.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.3.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.4.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.4.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.4.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.4.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.4.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.4.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.5.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.5.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.5.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.5.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.5.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.5.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.6.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.6.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.6.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.6.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.6.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.6.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.7.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.7.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.7.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.7.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.7.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.7.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.8.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.8.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.8.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.8.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.8.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.8.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.9.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.9.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.9.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.9.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.9.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.9.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.10.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.10.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.10.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.10.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.10.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.10.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.11.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.11.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.11.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.11.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.11.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.11.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.12.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.12.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.12.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.12.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.12.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.12.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.13.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.13.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.13.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.13.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.13.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.13.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.14.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.14.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.14.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.14.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.14.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.14.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.15.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.15.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.15.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.15.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.15.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.15.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.16.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.16.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.16.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.16.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.16.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.16.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.17.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.17.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.17.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.17.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.17.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.17.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.18.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.18.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.18.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.18.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.18.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.18.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.19.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.19.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.19.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.19.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.19.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.19.memory_block.base_mlp_output_lora.up.weight" + ], + "lora_numel": 3594240, + "lora_tensor_count": 192, + "memory_age_conditioning": false, + "memory_capacity": "clipped", + "method": "worldmem", + "profile_variant": "faithful_f200b200", + "reference_names": [ + "diffusion_model.model.down_blocks.2.0.memory_block.reference_qkv.weight", + "diffusion_model.model.down_blocks.2.0.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.down_blocks.2.0.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.down_blocks.2.0.memory_block.reference_output.weight", + "diffusion_model.model.down_blocks.2.0.memory_block.reference_output.bias", + "diffusion_model.model.down_blocks.2.0.memory_block.reference_ffn.0.weight", + "diffusion_model.model.down_blocks.2.0.memory_block.reference_ffn.0.bias", + "diffusion_model.model.down_blocks.2.0.memory_block.reference_ffn.2.weight", + "diffusion_model.model.down_blocks.2.0.memory_block.reference_ffn.2.bias", + "diffusion_model.model.down_blocks.2.0.memory_block.reference_adaln.1.weight", + "diffusion_model.model.down_blocks.2.0.memory_block.reference_adaln.1.bias", + "diffusion_model.model.down_blocks.2.1.memory_block.reference_qkv.weight", + "diffusion_model.model.down_blocks.2.1.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.down_blocks.2.1.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.down_blocks.2.1.memory_block.reference_output.weight", + "diffusion_model.model.down_blocks.2.1.memory_block.reference_output.bias", + "diffusion_model.model.down_blocks.2.1.memory_block.reference_ffn.0.weight", + "diffusion_model.model.down_blocks.2.1.memory_block.reference_ffn.0.bias", + "diffusion_model.model.down_blocks.2.1.memory_block.reference_ffn.2.weight", + "diffusion_model.model.down_blocks.2.1.memory_block.reference_ffn.2.bias", + "diffusion_model.model.down_blocks.2.1.memory_block.reference_adaln.1.weight", + "diffusion_model.model.down_blocks.2.1.memory_block.reference_adaln.1.bias", + "diffusion_model.model.down_blocks.2.2.memory_block.reference_qkv.weight", + "diffusion_model.model.down_blocks.2.2.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.down_blocks.2.2.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.down_blocks.2.2.memory_block.reference_output.weight", + "diffusion_model.model.down_blocks.2.2.memory_block.reference_output.bias", + "diffusion_model.model.down_blocks.2.2.memory_block.reference_ffn.0.weight", + "diffusion_model.model.down_blocks.2.2.memory_block.reference_ffn.0.bias", + "diffusion_model.model.down_blocks.2.2.memory_block.reference_ffn.2.weight", + "diffusion_model.model.down_blocks.2.2.memory_block.reference_ffn.2.bias", + "diffusion_model.model.down_blocks.2.2.memory_block.reference_adaln.1.weight", + "diffusion_model.model.down_blocks.2.2.memory_block.reference_adaln.1.bias", + "diffusion_model.model.down_blocks.2.3.memory_block.reference_qkv.weight", + "diffusion_model.model.down_blocks.2.3.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.down_blocks.2.3.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.down_blocks.2.3.memory_block.reference_output.weight", + "diffusion_model.model.down_blocks.2.3.memory_block.reference_output.bias", + "diffusion_model.model.down_blocks.2.3.memory_block.reference_ffn.0.weight", + "diffusion_model.model.down_blocks.2.3.memory_block.reference_ffn.0.bias", + "diffusion_model.model.down_blocks.2.3.memory_block.reference_ffn.2.weight", + "diffusion_model.model.down_blocks.2.3.memory_block.reference_ffn.2.bias", + "diffusion_model.model.down_blocks.2.3.memory_block.reference_adaln.1.weight", + "diffusion_model.model.down_blocks.2.3.memory_block.reference_adaln.1.bias", + "diffusion_model.model.down_blocks.2.4.memory_block.reference_qkv.weight", + "diffusion_model.model.down_blocks.2.4.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.down_blocks.2.4.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.down_blocks.2.4.memory_block.reference_output.weight", + "diffusion_model.model.down_blocks.2.4.memory_block.reference_output.bias", + "diffusion_model.model.down_blocks.2.4.memory_block.reference_ffn.0.weight", + "diffusion_model.model.down_blocks.2.4.memory_block.reference_ffn.0.bias", + "diffusion_model.model.down_blocks.2.4.memory_block.reference_ffn.2.weight", + "diffusion_model.model.down_blocks.2.4.memory_block.reference_ffn.2.bias", + "diffusion_model.model.down_blocks.2.4.memory_block.reference_adaln.1.weight", + "diffusion_model.model.down_blocks.2.4.memory_block.reference_adaln.1.bias", + "diffusion_model.model.down_blocks.2.5.memory_block.reference_qkv.weight", + "diffusion_model.model.down_blocks.2.5.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.down_blocks.2.5.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.down_blocks.2.5.memory_block.reference_output.weight", + "diffusion_model.model.down_blocks.2.5.memory_block.reference_output.bias", + "diffusion_model.model.down_blocks.2.5.memory_block.reference_ffn.0.weight", + "diffusion_model.model.down_blocks.2.5.memory_block.reference_ffn.0.bias", + "diffusion_model.model.down_blocks.2.5.memory_block.reference_ffn.2.weight", + "diffusion_model.model.down_blocks.2.5.memory_block.reference_ffn.2.bias", + "diffusion_model.model.down_blocks.2.5.memory_block.reference_adaln.1.weight", + "diffusion_model.model.down_blocks.2.5.memory_block.reference_adaln.1.bias", + "diffusion_model.model.up_blocks.0.1.memory_block.reference_qkv.weight", + "diffusion_model.model.up_blocks.0.1.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.up_blocks.0.1.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.up_blocks.0.1.memory_block.reference_output.weight", + "diffusion_model.model.up_blocks.0.1.memory_block.reference_output.bias", + "diffusion_model.model.up_blocks.0.1.memory_block.reference_ffn.0.weight", + "diffusion_model.model.up_blocks.0.1.memory_block.reference_ffn.0.bias", + "diffusion_model.model.up_blocks.0.1.memory_block.reference_ffn.2.weight", + "diffusion_model.model.up_blocks.0.1.memory_block.reference_ffn.2.bias", + "diffusion_model.model.up_blocks.0.1.memory_block.reference_adaln.1.weight", + "diffusion_model.model.up_blocks.0.1.memory_block.reference_adaln.1.bias", + "diffusion_model.model.up_blocks.0.2.memory_block.reference_qkv.weight", + "diffusion_model.model.up_blocks.0.2.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.up_blocks.0.2.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.up_blocks.0.2.memory_block.reference_output.weight", + "diffusion_model.model.up_blocks.0.2.memory_block.reference_output.bias", + "diffusion_model.model.up_blocks.0.2.memory_block.reference_ffn.0.weight", + "diffusion_model.model.up_blocks.0.2.memory_block.reference_ffn.0.bias", + "diffusion_model.model.up_blocks.0.2.memory_block.reference_ffn.2.weight", + "diffusion_model.model.up_blocks.0.2.memory_block.reference_ffn.2.bias", + "diffusion_model.model.up_blocks.0.2.memory_block.reference_adaln.1.weight", + "diffusion_model.model.up_blocks.0.2.memory_block.reference_adaln.1.bias", + "diffusion_model.model.up_blocks.0.3.memory_block.reference_qkv.weight", + "diffusion_model.model.up_blocks.0.3.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.up_blocks.0.3.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.up_blocks.0.3.memory_block.reference_output.weight", + "diffusion_model.model.up_blocks.0.3.memory_block.reference_output.bias", + "diffusion_model.model.up_blocks.0.3.memory_block.reference_ffn.0.weight", + "diffusion_model.model.up_blocks.0.3.memory_block.reference_ffn.0.bias", + "diffusion_model.model.up_blocks.0.3.memory_block.reference_ffn.2.weight", + "diffusion_model.model.up_blocks.0.3.memory_block.reference_ffn.2.bias", + "diffusion_model.model.up_blocks.0.3.memory_block.reference_adaln.1.weight", + "diffusion_model.model.up_blocks.0.3.memory_block.reference_adaln.1.bias", + "diffusion_model.model.up_blocks.0.4.memory_block.reference_qkv.weight", + "diffusion_model.model.up_blocks.0.4.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.up_blocks.0.4.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.up_blocks.0.4.memory_block.reference_output.weight", + "diffusion_model.model.up_blocks.0.4.memory_block.reference_output.bias", + "diffusion_model.model.up_blocks.0.4.memory_block.reference_ffn.0.weight", + "diffusion_model.model.up_blocks.0.4.memory_block.reference_ffn.0.bias", + "diffusion_model.model.up_blocks.0.4.memory_block.reference_ffn.2.weight", + "diffusion_model.model.up_blocks.0.4.memory_block.reference_ffn.2.bias", + "diffusion_model.model.up_blocks.0.4.memory_block.reference_adaln.1.weight", + "diffusion_model.model.up_blocks.0.4.memory_block.reference_adaln.1.bias", + "diffusion_model.model.up_blocks.0.5.memory_block.reference_qkv.weight", + "diffusion_model.model.up_blocks.0.5.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.up_blocks.0.5.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.up_blocks.0.5.memory_block.reference_output.weight", + "diffusion_model.model.up_blocks.0.5.memory_block.reference_output.bias", + "diffusion_model.model.up_blocks.0.5.memory_block.reference_ffn.0.weight", + "diffusion_model.model.up_blocks.0.5.memory_block.reference_ffn.0.bias", + "diffusion_model.model.up_blocks.0.5.memory_block.reference_ffn.2.weight", + "diffusion_model.model.up_blocks.0.5.memory_block.reference_ffn.2.bias", + "diffusion_model.model.up_blocks.0.5.memory_block.reference_adaln.1.weight", + "diffusion_model.model.up_blocks.0.5.memory_block.reference_adaln.1.bias", + "diffusion_model.model.up_blocks.0.6.memory_block.reference_qkv.weight", + "diffusion_model.model.up_blocks.0.6.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.up_blocks.0.6.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.up_blocks.0.6.memory_block.reference_output.weight", + "diffusion_model.model.up_blocks.0.6.memory_block.reference_output.bias", + "diffusion_model.model.up_blocks.0.6.memory_block.reference_ffn.0.weight", + "diffusion_model.model.up_blocks.0.6.memory_block.reference_ffn.0.bias", + "diffusion_model.model.up_blocks.0.6.memory_block.reference_ffn.2.weight", + "diffusion_model.model.up_blocks.0.6.memory_block.reference_ffn.2.bias", + "diffusion_model.model.up_blocks.0.6.memory_block.reference_adaln.1.weight", + "diffusion_model.model.up_blocks.0.6.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.0.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.0.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.0.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.0.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.0.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.0.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.0.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.0.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.0.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.0.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.0.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.1.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.1.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.1.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.1.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.1.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.1.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.1.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.1.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.1.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.1.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.1.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.2.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.2.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.2.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.2.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.2.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.2.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.2.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.2.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.2.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.2.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.2.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.3.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.3.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.3.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.3.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.3.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.3.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.3.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.3.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.3.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.3.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.3.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.4.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.4.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.4.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.4.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.4.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.4.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.4.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.4.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.4.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.4.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.4.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.5.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.5.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.5.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.5.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.5.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.5.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.5.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.5.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.5.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.5.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.5.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.6.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.6.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.6.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.6.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.6.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.6.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.6.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.6.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.6.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.6.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.6.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.7.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.7.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.7.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.7.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.7.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.7.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.7.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.7.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.7.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.7.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.7.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.8.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.8.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.8.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.8.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.8.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.8.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.8.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.8.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.8.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.8.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.8.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.9.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.9.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.9.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.9.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.9.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.9.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.9.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.9.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.9.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.9.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.9.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.10.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.10.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.10.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.10.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.10.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.10.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.10.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.10.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.10.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.10.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.10.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.11.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.11.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.11.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.11.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.11.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.11.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.11.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.11.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.11.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.11.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.11.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.12.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.12.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.12.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.12.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.12.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.12.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.12.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.12.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.12.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.12.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.12.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.13.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.13.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.13.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.13.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.13.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.13.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.13.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.13.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.13.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.13.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.13.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.14.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.14.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.14.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.14.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.14.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.14.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.14.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.14.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.14.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.14.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.14.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.15.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.15.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.15.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.15.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.15.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.15.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.15.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.15.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.15.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.15.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.15.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.16.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.16.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.16.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.16.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.16.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.16.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.16.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.16.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.16.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.16.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.16.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.17.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.17.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.17.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.17.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.17.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.17.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.17.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.17.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.17.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.17.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.17.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.18.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.18.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.18.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.18.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.18.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.18.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.18.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.18.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.18.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.18.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.18.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.19.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.19.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.19.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.19.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.19.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.19.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.19.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.19.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.19.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.19.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.19.memory_block.reference_adaln.1.bias", + "diffusion_model.model.memory_conditioner.pose_encoder.576.0.weight", + "diffusion_model.model.memory_conditioner.pose_encoder.576.0.bias", + "diffusion_model.model.memory_conditioner.pose_encoder.576.2.weight", + "diffusion_model.model.memory_conditioner.pose_encoder.576.2.bias", + "diffusion_model.model.memory_conditioner.pose_encoder.1152.0.weight", + "diffusion_model.model.memory_conditioner.pose_encoder.1152.0.bias", + "diffusion_model.model.memory_conditioner.pose_encoder.1152.2.weight", + "diffusion_model.model.memory_conditioner.pose_encoder.1152.2.bias", + "diffusion_model.model.memory_conditioner.timestamp_encoder.576.0.weight", + "diffusion_model.model.memory_conditioner.timestamp_encoder.576.0.bias", + "diffusion_model.model.memory_conditioner.timestamp_encoder.576.2.weight", + "diffusion_model.model.memory_conditioner.timestamp_encoder.576.2.bias", + "diffusion_model.model.memory_conditioner.timestamp_encoder.1152.0.weight", + "diffusion_model.model.memory_conditioner.timestamp_encoder.1152.0.bias", + "diffusion_model.model.memory_conditioner.timestamp_encoder.1152.2.weight", + "diffusion_model.model.memory_conditioner.timestamp_encoder.1152.2.bias", + "diffusion_model.model.memory_conditioner.timestep_encoder.0.weight", + "diffusion_model.model.memory_conditioner.timestep_encoder.0.bias" + ], + "reference_numel": 300503744, + "reference_projection_initialization": "independent_xavier_bias_free_qkv_v1", + "reference_tensor_count": 370, + "reference_timestamp_conditioning": true, + "schema_version": 1, + "selector_version": "re10k_faithful_trajectory_h100_f399_v1", + "trainable_names": [ + "diffusion_model.model.down_blocks.2.0.memory_block.reference_qkv.weight", + "diffusion_model.model.down_blocks.2.0.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.down_blocks.2.0.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.down_blocks.2.0.memory_block.reference_output.weight", + "diffusion_model.model.down_blocks.2.0.memory_block.reference_output.bias", + "diffusion_model.model.down_blocks.2.0.memory_block.reference_ffn.0.weight", + "diffusion_model.model.down_blocks.2.0.memory_block.reference_ffn.0.bias", + "diffusion_model.model.down_blocks.2.0.memory_block.reference_ffn.2.weight", + "diffusion_model.model.down_blocks.2.0.memory_block.reference_ffn.2.bias", + "diffusion_model.model.down_blocks.2.0.memory_block.reference_adaln.1.weight", + "diffusion_model.model.down_blocks.2.0.memory_block.reference_adaln.1.bias", + "diffusion_model.model.down_blocks.2.0.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.down_blocks.2.0.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.down_blocks.2.0.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.down_blocks.2.0.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.down_blocks.2.0.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.down_blocks.2.0.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.down_blocks.2.1.memory_block.reference_qkv.weight", + "diffusion_model.model.down_blocks.2.1.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.down_blocks.2.1.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.down_blocks.2.1.memory_block.reference_output.weight", + "diffusion_model.model.down_blocks.2.1.memory_block.reference_output.bias", + "diffusion_model.model.down_blocks.2.1.memory_block.reference_ffn.0.weight", + "diffusion_model.model.down_blocks.2.1.memory_block.reference_ffn.0.bias", + "diffusion_model.model.down_blocks.2.1.memory_block.reference_ffn.2.weight", + "diffusion_model.model.down_blocks.2.1.memory_block.reference_ffn.2.bias", + "diffusion_model.model.down_blocks.2.1.memory_block.reference_adaln.1.weight", + "diffusion_model.model.down_blocks.2.1.memory_block.reference_adaln.1.bias", + "diffusion_model.model.down_blocks.2.1.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.down_blocks.2.1.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.down_blocks.2.1.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.down_blocks.2.1.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.down_blocks.2.1.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.down_blocks.2.1.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.down_blocks.2.2.memory_block.reference_qkv.weight", + "diffusion_model.model.down_blocks.2.2.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.down_blocks.2.2.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.down_blocks.2.2.memory_block.reference_output.weight", + "diffusion_model.model.down_blocks.2.2.memory_block.reference_output.bias", + "diffusion_model.model.down_blocks.2.2.memory_block.reference_ffn.0.weight", + "diffusion_model.model.down_blocks.2.2.memory_block.reference_ffn.0.bias", + "diffusion_model.model.down_blocks.2.2.memory_block.reference_ffn.2.weight", + "diffusion_model.model.down_blocks.2.2.memory_block.reference_ffn.2.bias", + "diffusion_model.model.down_blocks.2.2.memory_block.reference_adaln.1.weight", + "diffusion_model.model.down_blocks.2.2.memory_block.reference_adaln.1.bias", + "diffusion_model.model.down_blocks.2.2.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.down_blocks.2.2.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.down_blocks.2.2.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.down_blocks.2.2.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.down_blocks.2.2.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.down_blocks.2.2.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.down_blocks.2.3.memory_block.reference_qkv.weight", + "diffusion_model.model.down_blocks.2.3.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.down_blocks.2.3.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.down_blocks.2.3.memory_block.reference_output.weight", + "diffusion_model.model.down_blocks.2.3.memory_block.reference_output.bias", + "diffusion_model.model.down_blocks.2.3.memory_block.reference_ffn.0.weight", + "diffusion_model.model.down_blocks.2.3.memory_block.reference_ffn.0.bias", + "diffusion_model.model.down_blocks.2.3.memory_block.reference_ffn.2.weight", + "diffusion_model.model.down_blocks.2.3.memory_block.reference_ffn.2.bias", + "diffusion_model.model.down_blocks.2.3.memory_block.reference_adaln.1.weight", + "diffusion_model.model.down_blocks.2.3.memory_block.reference_adaln.1.bias", + "diffusion_model.model.down_blocks.2.3.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.down_blocks.2.3.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.down_blocks.2.3.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.down_blocks.2.3.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.down_blocks.2.3.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.down_blocks.2.3.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.down_blocks.2.4.memory_block.reference_qkv.weight", + "diffusion_model.model.down_blocks.2.4.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.down_blocks.2.4.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.down_blocks.2.4.memory_block.reference_output.weight", + "diffusion_model.model.down_blocks.2.4.memory_block.reference_output.bias", + "diffusion_model.model.down_blocks.2.4.memory_block.reference_ffn.0.weight", + "diffusion_model.model.down_blocks.2.4.memory_block.reference_ffn.0.bias", + "diffusion_model.model.down_blocks.2.4.memory_block.reference_ffn.2.weight", + "diffusion_model.model.down_blocks.2.4.memory_block.reference_ffn.2.bias", + "diffusion_model.model.down_blocks.2.4.memory_block.reference_adaln.1.weight", + "diffusion_model.model.down_blocks.2.4.memory_block.reference_adaln.1.bias", + "diffusion_model.model.down_blocks.2.4.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.down_blocks.2.4.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.down_blocks.2.4.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.down_blocks.2.4.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.down_blocks.2.4.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.down_blocks.2.4.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.down_blocks.2.5.memory_block.reference_qkv.weight", + "diffusion_model.model.down_blocks.2.5.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.down_blocks.2.5.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.down_blocks.2.5.memory_block.reference_output.weight", + "diffusion_model.model.down_blocks.2.5.memory_block.reference_output.bias", + "diffusion_model.model.down_blocks.2.5.memory_block.reference_ffn.0.weight", + "diffusion_model.model.down_blocks.2.5.memory_block.reference_ffn.0.bias", + "diffusion_model.model.down_blocks.2.5.memory_block.reference_ffn.2.weight", + "diffusion_model.model.down_blocks.2.5.memory_block.reference_ffn.2.bias", + "diffusion_model.model.down_blocks.2.5.memory_block.reference_adaln.1.weight", + "diffusion_model.model.down_blocks.2.5.memory_block.reference_adaln.1.bias", + "diffusion_model.model.down_blocks.2.5.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.down_blocks.2.5.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.down_blocks.2.5.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.down_blocks.2.5.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.down_blocks.2.5.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.down_blocks.2.5.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.up_blocks.0.1.memory_block.reference_qkv.weight", + "diffusion_model.model.up_blocks.0.1.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.up_blocks.0.1.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.up_blocks.0.1.memory_block.reference_output.weight", + "diffusion_model.model.up_blocks.0.1.memory_block.reference_output.bias", + "diffusion_model.model.up_blocks.0.1.memory_block.reference_ffn.0.weight", + "diffusion_model.model.up_blocks.0.1.memory_block.reference_ffn.0.bias", + "diffusion_model.model.up_blocks.0.1.memory_block.reference_ffn.2.weight", + "diffusion_model.model.up_blocks.0.1.memory_block.reference_ffn.2.bias", + "diffusion_model.model.up_blocks.0.1.memory_block.reference_adaln.1.weight", + "diffusion_model.model.up_blocks.0.1.memory_block.reference_adaln.1.bias", + "diffusion_model.model.up_blocks.0.1.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.up_blocks.0.1.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.up_blocks.0.1.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.up_blocks.0.1.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.up_blocks.0.1.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.up_blocks.0.1.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.up_blocks.0.2.memory_block.reference_qkv.weight", + "diffusion_model.model.up_blocks.0.2.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.up_blocks.0.2.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.up_blocks.0.2.memory_block.reference_output.weight", + "diffusion_model.model.up_blocks.0.2.memory_block.reference_output.bias", + "diffusion_model.model.up_blocks.0.2.memory_block.reference_ffn.0.weight", + "diffusion_model.model.up_blocks.0.2.memory_block.reference_ffn.0.bias", + "diffusion_model.model.up_blocks.0.2.memory_block.reference_ffn.2.weight", + "diffusion_model.model.up_blocks.0.2.memory_block.reference_ffn.2.bias", + "diffusion_model.model.up_blocks.0.2.memory_block.reference_adaln.1.weight", + "diffusion_model.model.up_blocks.0.2.memory_block.reference_adaln.1.bias", + "diffusion_model.model.up_blocks.0.2.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.up_blocks.0.2.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.up_blocks.0.2.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.up_blocks.0.2.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.up_blocks.0.2.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.up_blocks.0.2.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.up_blocks.0.3.memory_block.reference_qkv.weight", + "diffusion_model.model.up_blocks.0.3.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.up_blocks.0.3.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.up_blocks.0.3.memory_block.reference_output.weight", + "diffusion_model.model.up_blocks.0.3.memory_block.reference_output.bias", + "diffusion_model.model.up_blocks.0.3.memory_block.reference_ffn.0.weight", + "diffusion_model.model.up_blocks.0.3.memory_block.reference_ffn.0.bias", + "diffusion_model.model.up_blocks.0.3.memory_block.reference_ffn.2.weight", + "diffusion_model.model.up_blocks.0.3.memory_block.reference_ffn.2.bias", + "diffusion_model.model.up_blocks.0.3.memory_block.reference_adaln.1.weight", + "diffusion_model.model.up_blocks.0.3.memory_block.reference_adaln.1.bias", + "diffusion_model.model.up_blocks.0.3.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.up_blocks.0.3.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.up_blocks.0.3.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.up_blocks.0.3.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.up_blocks.0.3.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.up_blocks.0.3.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.up_blocks.0.4.memory_block.reference_qkv.weight", + "diffusion_model.model.up_blocks.0.4.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.up_blocks.0.4.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.up_blocks.0.4.memory_block.reference_output.weight", + "diffusion_model.model.up_blocks.0.4.memory_block.reference_output.bias", + "diffusion_model.model.up_blocks.0.4.memory_block.reference_ffn.0.weight", + "diffusion_model.model.up_blocks.0.4.memory_block.reference_ffn.0.bias", + "diffusion_model.model.up_blocks.0.4.memory_block.reference_ffn.2.weight", + "diffusion_model.model.up_blocks.0.4.memory_block.reference_ffn.2.bias", + "diffusion_model.model.up_blocks.0.4.memory_block.reference_adaln.1.weight", + "diffusion_model.model.up_blocks.0.4.memory_block.reference_adaln.1.bias", + "diffusion_model.model.up_blocks.0.4.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.up_blocks.0.4.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.up_blocks.0.4.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.up_blocks.0.4.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.up_blocks.0.4.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.up_blocks.0.4.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.up_blocks.0.5.memory_block.reference_qkv.weight", + "diffusion_model.model.up_blocks.0.5.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.up_blocks.0.5.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.up_blocks.0.5.memory_block.reference_output.weight", + "diffusion_model.model.up_blocks.0.5.memory_block.reference_output.bias", + "diffusion_model.model.up_blocks.0.5.memory_block.reference_ffn.0.weight", + "diffusion_model.model.up_blocks.0.5.memory_block.reference_ffn.0.bias", + "diffusion_model.model.up_blocks.0.5.memory_block.reference_ffn.2.weight", + "diffusion_model.model.up_blocks.0.5.memory_block.reference_ffn.2.bias", + "diffusion_model.model.up_blocks.0.5.memory_block.reference_adaln.1.weight", + "diffusion_model.model.up_blocks.0.5.memory_block.reference_adaln.1.bias", + "diffusion_model.model.up_blocks.0.5.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.up_blocks.0.5.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.up_blocks.0.5.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.up_blocks.0.5.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.up_blocks.0.5.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.up_blocks.0.5.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.up_blocks.0.6.memory_block.reference_qkv.weight", + "diffusion_model.model.up_blocks.0.6.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.up_blocks.0.6.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.up_blocks.0.6.memory_block.reference_output.weight", + "diffusion_model.model.up_blocks.0.6.memory_block.reference_output.bias", + "diffusion_model.model.up_blocks.0.6.memory_block.reference_ffn.0.weight", + "diffusion_model.model.up_blocks.0.6.memory_block.reference_ffn.0.bias", + "diffusion_model.model.up_blocks.0.6.memory_block.reference_ffn.2.weight", + "diffusion_model.model.up_blocks.0.6.memory_block.reference_ffn.2.bias", + "diffusion_model.model.up_blocks.0.6.memory_block.reference_adaln.1.weight", + "diffusion_model.model.up_blocks.0.6.memory_block.reference_adaln.1.bias", + "diffusion_model.model.up_blocks.0.6.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.up_blocks.0.6.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.up_blocks.0.6.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.up_blocks.0.6.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.up_blocks.0.6.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.up_blocks.0.6.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.0.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.0.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.0.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.0.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.0.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.0.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.0.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.0.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.0.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.0.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.0.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.0.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.0.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.0.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.0.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.0.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.0.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.1.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.1.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.1.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.1.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.1.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.1.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.1.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.1.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.1.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.1.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.1.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.1.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.1.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.1.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.1.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.1.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.1.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.2.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.2.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.2.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.2.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.2.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.2.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.2.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.2.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.2.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.2.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.2.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.2.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.2.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.2.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.2.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.2.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.2.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.3.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.3.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.3.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.3.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.3.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.3.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.3.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.3.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.3.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.3.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.3.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.3.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.3.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.3.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.3.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.3.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.3.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.4.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.4.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.4.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.4.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.4.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.4.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.4.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.4.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.4.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.4.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.4.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.4.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.4.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.4.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.4.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.4.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.4.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.5.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.5.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.5.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.5.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.5.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.5.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.5.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.5.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.5.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.5.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.5.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.5.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.5.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.5.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.5.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.5.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.5.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.6.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.6.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.6.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.6.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.6.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.6.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.6.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.6.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.6.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.6.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.6.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.6.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.6.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.6.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.6.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.6.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.6.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.7.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.7.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.7.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.7.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.7.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.7.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.7.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.7.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.7.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.7.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.7.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.7.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.7.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.7.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.7.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.7.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.7.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.8.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.8.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.8.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.8.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.8.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.8.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.8.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.8.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.8.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.8.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.8.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.8.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.8.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.8.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.8.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.8.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.8.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.9.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.9.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.9.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.9.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.9.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.9.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.9.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.9.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.9.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.9.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.9.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.9.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.9.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.9.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.9.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.9.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.9.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.10.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.10.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.10.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.10.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.10.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.10.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.10.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.10.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.10.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.10.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.10.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.10.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.10.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.10.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.10.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.10.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.10.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.11.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.11.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.11.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.11.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.11.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.11.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.11.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.11.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.11.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.11.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.11.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.11.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.11.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.11.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.11.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.11.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.11.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.12.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.12.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.12.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.12.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.12.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.12.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.12.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.12.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.12.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.12.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.12.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.12.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.12.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.12.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.12.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.12.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.12.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.13.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.13.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.13.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.13.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.13.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.13.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.13.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.13.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.13.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.13.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.13.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.13.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.13.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.13.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.13.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.13.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.13.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.14.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.14.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.14.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.14.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.14.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.14.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.14.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.14.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.14.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.14.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.14.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.14.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.14.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.14.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.14.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.14.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.14.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.15.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.15.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.15.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.15.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.15.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.15.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.15.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.15.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.15.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.15.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.15.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.15.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.15.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.15.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.15.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.15.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.15.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.16.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.16.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.16.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.16.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.16.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.16.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.16.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.16.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.16.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.16.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.16.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.16.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.16.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.16.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.16.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.16.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.16.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.17.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.17.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.17.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.17.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.17.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.17.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.17.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.17.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.17.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.17.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.17.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.17.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.17.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.17.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.17.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.17.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.17.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.18.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.18.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.18.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.18.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.18.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.18.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.18.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.18.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.18.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.18.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.18.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.18.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.18.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.18.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.18.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.18.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.18.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.mid_blocks.19.memory_block.reference_qkv.weight", + "diffusion_model.model.mid_blocks.19.memory_block.pose_cond_mlp.weight", + "diffusion_model.model.mid_blocks.19.memory_block.pose_cond_mlp.bias", + "diffusion_model.model.mid_blocks.19.memory_block.reference_output.weight", + "diffusion_model.model.mid_blocks.19.memory_block.reference_output.bias", + "diffusion_model.model.mid_blocks.19.memory_block.reference_ffn.0.weight", + "diffusion_model.model.mid_blocks.19.memory_block.reference_ffn.0.bias", + "diffusion_model.model.mid_blocks.19.memory_block.reference_ffn.2.weight", + "diffusion_model.model.mid_blocks.19.memory_block.reference_ffn.2.bias", + "diffusion_model.model.mid_blocks.19.memory_block.reference_adaln.1.weight", + "diffusion_model.model.mid_blocks.19.memory_block.reference_adaln.1.bias", + "diffusion_model.model.mid_blocks.19.memory_block.base_fused_lora.down.weight", + "diffusion_model.model.mid_blocks.19.memory_block.base_fused_lora.up.weight", + "diffusion_model.model.mid_blocks.19.memory_block.base_attention_output_lora.down.weight", + "diffusion_model.model.mid_blocks.19.memory_block.base_attention_output_lora.up.weight", + "diffusion_model.model.mid_blocks.19.memory_block.base_mlp_output_lora.down.weight", + "diffusion_model.model.mid_blocks.19.memory_block.base_mlp_output_lora.up.weight", + "diffusion_model.model.memory_conditioner.pose_encoder.576.0.weight", + "diffusion_model.model.memory_conditioner.pose_encoder.576.0.bias", + "diffusion_model.model.memory_conditioner.pose_encoder.576.2.weight", + "diffusion_model.model.memory_conditioner.pose_encoder.576.2.bias", + "diffusion_model.model.memory_conditioner.pose_encoder.1152.0.weight", + "diffusion_model.model.memory_conditioner.pose_encoder.1152.0.bias", + "diffusion_model.model.memory_conditioner.pose_encoder.1152.2.weight", + "diffusion_model.model.memory_conditioner.pose_encoder.1152.2.bias", + "diffusion_model.model.memory_conditioner.timestamp_encoder.576.0.weight", + "diffusion_model.model.memory_conditioner.timestamp_encoder.576.0.bias", + "diffusion_model.model.memory_conditioner.timestamp_encoder.576.2.weight", + "diffusion_model.model.memory_conditioner.timestamp_encoder.576.2.bias", + "diffusion_model.model.memory_conditioner.timestamp_encoder.1152.0.weight", + "diffusion_model.model.memory_conditioner.timestamp_encoder.1152.0.bias", + "diffusion_model.model.memory_conditioner.timestamp_encoder.1152.2.weight", + "diffusion_model.model.memory_conditioner.timestamp_encoder.1152.2.bias", + "diffusion_model.model.memory_conditioner.timestep_encoder.0.weight", + "diffusion_model.model.memory_conditioner.timestep_encoder.0.bias" + ], + "trainable_numel": 304097984, + "trainable_tensor_count": 562, + "training_ema_decay": 0.9999 +} diff --git a/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/0ece034988793847.mp4 b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/0ece034988793847.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..284151e0f9d7ccf6ccaf22ed89166baa93e13436 --- /dev/null +++ b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/0ece034988793847.mp4 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8c30cc49297bb760504ed4df2f6d3e9fa66358444d262e00c5318c4cbb0fdbe8 +size 1242094 diff --git a/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/1735d8d1b4015669.mp4 b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/1735d8d1b4015669.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..15dbf8a30502931668ddca51d85a0a16508734fb --- /dev/null +++ b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/1735d8d1b4015669.mp4 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bb9f0b0b55b4520adbdf265f2ee8082654cd9f4254df185c9ef028ba2731181d +size 1492655 diff --git a/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/1cdfd3abfcc3a64e.mp4 b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/1cdfd3abfcc3a64e.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..fa5060c170899a24d34627b9c88bf36e52c5aece --- /dev/null +++ b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/1cdfd3abfcc3a64e.mp4 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:676b339a74b6655b970d43ac64a46e2f843ed3be2f8990803299f3a566c9c9d5 +size 2546246 diff --git a/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/6afff9609b65a604.mp4 b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/6afff9609b65a604.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..270aafa1634d27ee0dca9490b10cd3162c477093 --- /dev/null +++ b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/6afff9609b65a604.mp4 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:116944e329855fa9964d50515303eee3336da054413845063021a83d74b1798b +size 1569548 diff --git a/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/b009fbe3950ebeb2.mp4 b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/b009fbe3950ebeb2.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..e9c3af5a936caaf476012b6416232d0808b23514 --- /dev/null +++ b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/b009fbe3950ebeb2.mp4 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:297d0ef380a093091e820caab4ac38b8bce23f6f9c2aec87be28b5c29c0be31a +size 1759351 diff --git a/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/bcc76b58e2585320.mp4 b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/bcc76b58e2585320.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..6d02e7e3815fac25fcf0537da3a87e6b1f778e81 --- /dev/null +++ b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/bcc76b58e2585320.mp4 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f155fa6d1e071beac028fe98bd0aea4852d60c4fce5cbd4d09e177b4c0c7dc67 +size 1709270 diff --git a/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/c49a14de01d7bd35.mp4 b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/c49a14de01d7bd35.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..45db252fb95cc621d044239b5868be17d0605a25 --- /dev/null +++ b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/c49a14de01d7bd35.mp4 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d505890e93585e21a80f0c158cd24446fa71b448d9123f26d52bd0dd43943b4b +size 1588155 diff --git a/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/f287c679c250a841.mp4 b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/f287c679c250a841.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..111afe3f8f8f1414048f2fce2e66c4170b3152c7 --- /dev/null +++ b/results/re10k_short_reverse_worldmem_clipped_monitor8_train731edd42758e_eval940b6fb72bb1_step30000_job81245/videos/test_loop_true_reverse_h100_f199_v1_step00000000/f287c679c250a841.mp4 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:daf6e6199974b620994829985628ad3192f253b4aa777c90e6af24e307f2b6ef +size 1473196