Upload folder using huggingface_hub

.gitattributes

@@ -134,3 +134,61 @@ base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b64_btn1024_vector_q
 base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b64_btn1024_vector_quantize_rcs8192__/wandb/run-20250929_000209-p2mcszig/files/media/videos/vis_train_dataset_950_a048891b4bb2e302eb8c.mp4 filter=lfs diff=lfs merge=lfs -text
 base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b64_btn1024_vector_quantize_rcs8192__/wandb/run-20250929_000209-p2mcszig/files/media/videos/vis_train_dataset_975_dad91748c15462b26fb3.mp4 filter=lfs diff=lfs merge=lfs -text
 base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b64_btn1024_vector_quantize_rcs8192__/wandb/run-20250929_000209-p2mcszig/run-p2mcszig.wandb filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_10_0dcd7543861eb951d9eb.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_11_c440b37dfe11e1ecb291.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_12_2c6abd01594c9f04b701.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_13_07ed7abcba2550022a65.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_14_1d19ba838a2636f6bbb9.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_15_e4b12dc26b33b48899cd.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_16_be836e9a23353717c38d.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_17_db636a27733aa82a9288.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_18_8520ee5928f1948584de.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_19_9902d6a8e4fcef1843c7.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_1_4905dd5ddf36de8fa967.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_20_430fdf292106f4cdd169.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_21_b16346a94669cd22074d.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_22_b9cd0a98c6efd38dd326.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_23_16c5160b84b0b93c4d9a.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_24_caa9fc05e947e5c68fe5.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_2_d15bd6c3ab28bcf2d818.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_3_53d7f7f93663c75403c5.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_4_518c61a588fe34e8b97b.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_5_a718cd301e53d86c413d.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_6_75eb65b267e94c10e7e6.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_7_7f140abb6a7cc547d17d.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_8_4b22b9cea46a63503ee2.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_9_09d0cffdb29ebf65107e.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_10_cb55eb08e79a819f9319.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_11_83c1cb0239fed9f79371.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_12_5b3c2426eecabf472d8d.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_13_6876091a79457be67ea7.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_14_e23bb82f59567ab38280.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_15_08fbe0df36c11295fe0e.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_16_6dd373b8d21986a89617.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_17_0b8984f5b47aeb710bec.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_18_37b7c06cbf94b4d1290d.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_19_756caade2d1fd4a30c0a.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_1_cb69ac202a830cd54f3c.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_20_f34ba9f9b4ff9969489b.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_21_91d3de1a852a74091ca3.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_22_b8ec8ae339ed7f41fe8b.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_23_fcb034794f3447789b5c.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_24_d4dac4cd93c9f9849e0c.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_2_a619d3adb2f03b80bc21.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_3_256510a892d28f79e85e.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_4_b55a09705365f1f7bc7c.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_5_e65019ee1ebe9882b0b7.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_6_f4f71170b27199f67c46.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_7_ee4dfc69e0f0ff992917.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_8_3a191d37e945e0e0dc25.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_9_e121e2ba5f686fd4950c.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/run-c3u3silm.wandb filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251027_052202-c3u3silm/files/media/videos/vis_test_dataset_25_f90c3d650c4a707c1ec3.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251027_052202-c3u3silm/files/media/videos/vis_test_dataset_26_fcb0b2b66064e1cc523a.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251027_052202-c3u3silm/files/media/videos/vis_test_dataset_27_fb495aa0725107fd4084.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251027_052202-c3u3silm/files/media/videos/vis_test_dataset_28_f3d85415155c518fccc7.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251027_052202-c3u3silm/files/media/videos/vis_train_dataset_25_9060c16221f1e99f7b6e.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251027_052202-c3u3silm/files/media/videos/vis_train_dataset_26_aed6f3de431c91b8cae8.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251027_052202-c3u3silm/files/media/videos/vis_train_dataset_27_31f4a1a3e263ac0f4120.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251027_052202-c3u3silm/files/media/videos/vis_train_dataset_28_3aa6f2f161e847e89032.mp4 filter=lfs diff=lfs merge=lfs -text
+k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251027_052202-c3u3silm/run-c3u3silm.wandb filter=lfs diff=lfs merge=lfs -text

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/best_fvd_105.24.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:94381c6f59e56b006cfb2704277816901a6222f0470205e5ad06ea787ef13447
+size 8223495896

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/cfg.yaml

@@ -0,0 +1,477 @@
+!!python/object/new:easydict.EasyDict
+state:
+  trainer: our_tokenizer_trainer
+  train_dataset: &id020 !!python/object/new:easydict.EasyDict
+    state:
+      name: video_dataset
+      args: &id001 !!python/object/new:easydict.EasyDict
+        state:
+          root_path: data/metadata
+          split: train
+          frame_num: 16
+          rand_augment: 'no'
+          csv_file: k600_train.csv+ucf101_train.csv
+          cls_vid_num: -1_-1
+          crop_size: 128
+          scale: 1.0
+          aspect_ratio: 1.0
+          rand_flip: 'yes'
+          use_all_frames: false
+          pre_load: false
+        dictitems:
+          root_path: data/metadata
+          split: train
+          frame_num: 16
+          rand_augment: 'no'
+          csv_file: k600_train.csv+ucf101_train.csv
+          cls_vid_num: -1_-1
+          crop_size: 128
+          scale: 1.0
+          aspect_ratio: 1.0
+          rand_flip: 'yes'
+          use_all_frames: false
+          pre_load: false
+      loader: &id002 !!python/object/new:easydict.EasyDict
+        state:
+          batch_size: 128
+          num_workers: 32
+        dictitems:
+          batch_size: 128
+          num_workers: 32
+    dictitems:
+      name: video_dataset
+      args: *id001
+      loader: *id002
+  test_dataset: &id021 !!python/object/new:easydict.EasyDict
+    state:
+      name: video_dataset
+      args: &id003 !!python/object/new:easydict.EasyDict
+        state:
+          root_path: data/metadata
+          frame_num: 16
+          cls_vid_num: -1_-1
+          crop_size: 128
+          use_all_frames: false
+          pre_load: false
+        dictitems:
+          root_path: data/metadata
+          frame_num: 16
+          cls_vid_num: -1_-1
+          crop_size: 128
+          use_all_frames: false
+          pre_load: false
+      csv_paths: &id004 !!python/object/new:easydict.EasyDict
+        state:
+          ucf101_val: ucf101_val.csv
+        dictitems:
+          ucf101_val: ucf101_val.csv
+      loader: &id005 !!python/object/new:easydict.EasyDict
+        state:
+          batch_size: 128
+          num_workers: 32
+        dictitems:
+          batch_size: 128
+          num_workers: 32
+    dictitems:
+      name: video_dataset
+      args: *id003
+      csv_paths: *id004
+      loader: *id005
+  model: &id022 !!python/object/new:easydict.EasyDict
+    state:
+      name: dyn_tokenizer
+      args: &id016 !!python/object/new:easydict.EasyDict
+        state:
+          noise_schedule: &id012 !!python/object/new:easydict.EasyDict
+            state:
+              name: min_rf_noise_module
+              args: &id006 !!python/object/new:easydict.EasyDict
+                state:
+                  clean_data_read_key: clean_data
+                  noised_data_write_key: noisy_input
+                  noise_write_key: flow_noise
+                  timesteps_write_key: timesteps
+                  sigmas_write_key: sigmas
+                  ln: false
+                  stratisfied: false
+                  mode_scale: 0.25
+                dictitems:
+                  clean_data_read_key: clean_data
+                  noised_data_write_key: noisy_input
+                  noise_write_key: flow_noise
+                  timesteps_write_key: timesteps
+                  sigmas_write_key: sigmas
+                  ln: false
+                  stratisfied: false
+                  mode_scale: 0.25
+            dictitems:
+              name: min_rf_noise_module
+              args: *id006
+          bottleneck: &id013 !!python/object/new:easydict.EasyDict
+            state:
+              name: bottleneck
+              args: &id009 !!python/object/new:easydict.EasyDict
+                state:
+                  regularizer: &id008 !!python/object/new:easydict.EasyDict
+                    state:
+                      name: vector_quantize
+                      args: &id007 !!python/object/new:easydict.EasyDict
+                        state:
+                          codebook_dim: 16
+                          codebook_size: 8192
+                          ema_update: true
+                          decay: 0.99
+                          kmeans_init: true
+                          kmeans_iters: 10
+                          threshold_ema_dead_code: 0.2
+                          use_cosine_sim: true
+                          commitment_weight: 1.0
+                          diversity_weight: 0.0
+                          smart_re_K: 0
+                          continuous: false
+                          reg:
+                          - 0.1
+                          - 0.3
+                          reset_cluster_size: 0.2
+                          ema_entropy_ratio: 0.8
+                          vq_start_step: 0
+                        dictitems:
+                          codebook_dim: 16
+                          codebook_size: 8192
+                          ema_update: true
+                          decay: 0.99
+                          kmeans_init: true
+                          kmeans_iters: 10
+                          threshold_ema_dead_code: 0.2
+                          use_cosine_sim: true
+                          commitment_weight: 1.0
+                          diversity_weight: 0.0
+                          smart_re_K: 0
+                          continuous: false
+                          reg:
+                          - 0.1
+                          - 0.3
+                          reset_cluster_size: 0.2
+                          ema_entropy_ratio: 0.8
+                          vq_start_step: 0
+                    dictitems:
+                      name: vector_quantize
+                      args: *id007
+                dictitems:
+                  regularizer: *id008
+            dictitems:
+              name: bottleneck
+              args: *id009
+          prior_model: &id014 !!python/object/new:easydict.EasyDict
+            state:
+              name: none
+              use_mix_ss: true
+              mix_ss_max_ratio: 0.5
+              mix_ss_peak_steps_ratio: 0.3
+              n_rounds: 2
+              avg_loss_over_rounds: true
+              no_grad_before_last_round: false
+              no_dropout: false
+              latent_ce_temperature: 1.0
+              args: &id010 !!python/object/new:easydict.EasyDict
+                state:
+                  l2_normalized: true
+                dictitems:
+                  l2_normalized: true
+            dictitems:
+              name: none
+              use_mix_ss: true
+              mix_ss_max_ratio: 0.5
+              mix_ss_peak_steps_ratio: 0.3
+              n_rounds: 2
+              avg_loss_over_rounds: true
+              no_grad_before_last_round: false
+              no_dropout: false
+              latent_ce_temperature: 1.0
+              args: *id010
+          dec_time_embedder: &id015 !!python/object/new:easydict.EasyDict
+            state:
+              name: timestep_embedder
+              args: &id011 !!python/object/new:easydict.EasyDict
+                state:
+                  timesteps_read_key: timesteps
+                  time_embedding_write_key: dec_temb
+                  dim: 1152
+                  frequency_embedding_size: 256
+                  max_timestep: 1000.0
+                dictitems:
+                  timesteps_read_key: timesteps
+                  time_embedding_write_key: dec_temb
+                  dim: 1152
+                  frequency_embedding_size: 256
+                  max_timestep: 1000.0
+            dictitems:
+              name: timestep_embedder
+              args: *id011
+          transformer_name: transformer_encoder_parallel
+          encoder_name: none
+          decoder_name: transformer_AdaLN_decoder_parallel
+          bottleneck_token_num: 1024
+          input_size: 128
+          frame_num: 16
+          temporal_patch_size: 4
+          patch_size: 8
+          decoder_temporal_patch_size: 4
+          decoder_patch_size: 8
+          in_channels: 3
+          encoder_hidden_size: 768
+          decoder_hidden_size: 1152
+          encoder_num_heads: 12
+          decoder_num_heads: 18
+          encoder_depth: 12
+          decoder_depth: 18
+          encoder_block_name: block_timm
+          decoder_block_name: adaLN_block_timm
+          encoder_mask_mode: full
+          decoder_mask_mode: full
+          learned_encoder_patch_pe: false
+          learned_encoder_latent_query_embed: true
+          learned_decoder_latent_pe: false
+          learned_decoder_patch_query_embed: false
+          use_encoder_patch_token_type_embed: false
+          use_encoder_latent_query_token_type_embed: false
+          enable_decoder_query: false
+          learned_decoder_pe: false
+          use_decoder_latent_token_type_embed: false
+          use_decoder_patch_query_token_type_embed: true
+          encoder_query_gaussian_init: true
+          latent_pe_scale_factor: 10000
+          query_init_std: 0.02
+          adaLN_expansion: 2
+          final_layer_init: xavier_uniform
+          enable_vq: true
+          qk_norm: true
+          use_rope: true
+          rope_dim:
+          - 16
+          - 24
+          - 24
+          final_layer_type: adanorm
+        dictitems:
+          noise_schedule: *id012
+          bottleneck: *id013
+          prior_model: *id014
+          dec_time_embedder: *id015
+          transformer_name: transformer_encoder_parallel
+          encoder_name: none
+          decoder_name: transformer_AdaLN_decoder_parallel
+          bottleneck_token_num: 1024
+          input_size: 128
+          frame_num: 16
+          temporal_patch_size: 4
+          patch_size: 8
+          decoder_temporal_patch_size: 4
+          decoder_patch_size: 8
+          in_channels: 3
+          encoder_hidden_size: 768
+          decoder_hidden_size: 1152
+          encoder_num_heads: 12
+          decoder_num_heads: 18
+          encoder_depth: 12
+          decoder_depth: 18
+          encoder_block_name: block_timm
+          decoder_block_name: adaLN_block_timm
+          encoder_mask_mode: full
+          decoder_mask_mode: full
+          learned_encoder_patch_pe: false
+          learned_encoder_latent_query_embed: true
+          learned_decoder_latent_pe: false
+          learned_decoder_patch_query_embed: false
+          use_encoder_patch_token_type_embed: false
+          use_encoder_latent_query_token_type_embed: false
+          enable_decoder_query: false
+          learned_decoder_pe: false
+          use_decoder_latent_token_type_embed: false
+          use_decoder_patch_query_token_type_embed: true
+          encoder_query_gaussian_init: true
+          latent_pe_scale_factor: 10000
+          query_init_std: 0.02
+          adaLN_expansion: 2
+          final_layer_init: xavier_uniform
+          enable_vq: true
+          qk_norm: true
+          use_rope: true
+          rope_dim:
+          - 16
+          - 24
+          - 24
+          final_layer_type: adanorm
+    dictitems:
+      name: dyn_tokenizer
+      args: *id016
+  loss: &id023 !!python/object/new:easydict.EasyDict
+    state:
+      name: fm_disc_loss
+      args: &id017 !!python/object/new:easydict.EasyDict
+        state:
+          disc_type: transformer
+          disc_start: 999999
+          disc_self_start: -1
+          perceptual_weight: 0.5
+          perceptual_loss: lpips
+          perceptual_fp16: false
+          lecam_weight: 0.001
+          disc_loss: ns_smooth
+          disc_weight: 0.0
+          r1_gp_weight: 0.0
+          d_update_freq: 5
+          spectral_norm: false
+          disc_tran_hidden_size: 512
+          disc_tran_n_heads: 8
+          disc_tran_n_layers: 12
+          disc_tran_temporal_patch_size: 4
+          disc_tran_patch_size: 8
+          input_spatial_size: 128
+          frame_num: 16
+          fm_loss_weight: 1.0
+        dictitems:
+          disc_type: transformer
+          disc_start: 999999
+          disc_self_start: -1
+          perceptual_weight: 0.5
+          perceptual_loss: lpips
+          perceptual_fp16: false
+          lecam_weight: 0.001
+          disc_loss: ns_smooth
+          disc_weight: 0.0
+          r1_gp_weight: 0.0
+          d_update_freq: 5
+          spectral_norm: false
+          disc_tran_hidden_size: 512
+          disc_tran_n_heads: 8
+          disc_tran_n_layers: 12
+          disc_tran_temporal_patch_size: 4
+          disc_tran_patch_size: 8
+          input_spatial_size: 128
+          frame_num: 16
+          fm_loss_weight: 1.0
+    dictitems:
+      name: fm_disc_loss
+      args: *id017
+  optimizer: &id024 !!python/object/new:easydict.EasyDict
+    state:
+      name: adamw
+      loss_name: adam
+      args: &id018 !!python/object/new:easydict.EasyDict
+        state:
+          lr: 0.0001
+          betas:
+          - 0.9
+          - 0.99
+        dictitems:
+          lr: 0.0001
+          betas:
+          - 0.9
+          - 0.99
+      loss_args: &id019 !!python/object/new:easydict.EasyDict
+        state:
+          lr: 3.0e-05
+          betas:
+          - 0.5
+          - 0.9
+        dictitems:
+          lr: 3.0e-05
+          betas:
+          - 0.5
+          - 0.9
+      lr_type: step
+      lr_step_pcts: 0.9_0.95
+      warmup_epoch: 10
+      min_lr_mult: 0.1
+      prior_lr_mult: 1.0
+      emb_lr_mult: 1.0
+    dictitems:
+      name: adamw
+      loss_name: adam
+      args: *id018
+      loss_args: *id019
+      lr_type: step
+      lr_step_pcts: 0.9_0.95
+      warmup_epoch: 10
+      min_lr_mult: 0.1
+      prior_lr_mult: 1.0
+      emb_lr_mult: 1.0
+  max_epoch: 200
+  eval_epoch: 1
+  vis_epoch: 1
+  latest_interval: 1
+  save_epoch: 100000000
+  save_best: true
+  stepwise_logging: false
+  ema_decay: _
+  use_amp: true
+  amp_dtype: float16
+  compile: true
+  compile_mode: default
+  flash_attn: false
+  loss_q_weight: 0.1
+  loss_q_warmup: '1.0_1'
+  loss_kl_weight: 0.0
+  kl_decay_epoch: -1
+  loss_latent_ce_weight: 0.0
+  sqt_start_end_epoch: 0.0_0.0_0
+  clip_grad_max_norm: 0.0
+  init_checkpoint: ''
+  timesteps: 25
+  verbose: false
+  guidance_scale: 1.0
+  env: &id025 !!python/object/new:easydict.EasyDict
+    state:
+      tot_gpus: 8
+      cudnn: false
+      wandb_upload: true
+      wandb_entity: lingmin
+      wandb_project: dyn_tokenizer
+      exp_name: k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__
+      save_dir: /scratch/Projects/CFP-01/CFP01-CF-033/lingmin/1d_tokenizer/save/tokenizer_ours/tokenizer_e12_d18_k600/k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__
+      port: '12778'
+    dictitems:
+      tot_gpus: 8
+      cudnn: false
+      wandb_upload: true
+      wandb_entity: lingmin
+      wandb_project: dyn_tokenizer
+      exp_name: k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__
+      save_dir: /scratch/Projects/CFP-01/CFP01-CF-033/lingmin/1d_tokenizer/save/tokenizer_ours/tokenizer_e12_d18_k600/k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__
+      port: '12778'
+  comment: ''
+  manualSeed: 66667
+dictitems:
+  trainer: our_tokenizer_trainer
+  train_dataset: *id020
+  test_dataset: *id021
+  model: *id022
+  loss: *id023
+  optimizer: *id024
+  max_epoch: 200
+  eval_epoch: 1
+  vis_epoch: 1
+  latest_interval: 1
+  save_epoch: 100000000
+  save_best: true
+  stepwise_logging: false
+  ema_decay: _
+  use_amp: true
+  amp_dtype: float16
+  compile: true
+  compile_mode: default
+  flash_attn: false
+  loss_q_weight: 0.1
+  loss_q_warmup: '1.0_1'
+  loss_kl_weight: 0.0
+  kl_decay_epoch: -1
+  loss_latent_ce_weight: 0.0
+  sqt_start_end_epoch: 0.0_0.0_0
+  clip_grad_max_norm: 0.0
+  init_checkpoint: ''
+  timesteps: 25
+  verbose: false
+  guidance_scale: 1.0
+  env: *id025
+  comment: ''
+  manualSeed: 66667

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/epoch-last.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:94381c6f59e56b006cfb2704277816901a6222f0470205e5ad06ea787ef13447
+size 8223495896

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/log_2025_10_26_03_58_19.txt

@@ -0,0 +1,393 @@
+[10-26 03:58:19] Distributed training enabled.
+[10-26 03:58:20] Environment setup done.
+[10-26 03:58:21] Train dataset: len=435743
+[10-26 03:58:21] Test dataset: ucf101_val, len=3783
+[10-26 03:58:34] DYNTokenizer(
+  (x_embedder): PatchEmbed3D(
+    (proj): Conv3d(3, 768, kernel_size=(4, 8, 8), stride=(4, 8, 8))
+    (norm): Identity()
+  )
+  (encoder): TransformerEncoderParallel(
+    (blocks): ModuleList(
+      (0-11): 12 x AttentionBlock(
+        (norm1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
+        (attn): Attention(
+          (qkv): Linear(in_features=768, out_features=2304, bias=False)
+          (q_norm): LayerNorm((64,), eps=1e-06, elementwise_affine=True)
+          (k_norm): LayerNorm((64,), eps=1e-06, elementwise_affine=True)
+          (attn_drop): Dropout(p=0.0, inplace=False)
+          (norm): Identity()
+          (proj): Linear(in_features=768, out_features=768, bias=True)
+          (proj_drop): Dropout(p=0.0, inplace=False)
+        )
+        (ls1): Identity()
+        (drop_path1): Identity()
+        (norm2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
+        (mlp): Mlp(
+          (fc1): Linear(in_features=768, out_features=3072, bias=True)
+          (act): GELU(approximate='none')
+          (drop1): Dropout(p=0.0, inplace=False)
+          (norm): Identity()
+          (fc2): Linear(in_features=3072, out_features=768, bias=True)
+          (drop2): Dropout(p=0.0, inplace=False)
+        )
+        (ls2): Identity()
+        (drop_path2): Identity()
+      )
+    )
+  )
+  (decoder): TransformerAdaLNDecoderParallel(
+    (blocks): ModuleList(
+      (0-17): 18 x DiffusionAttentionBlock(
+        (norm1): LayerNorm((1152,), eps=1e-06, elementwise_affine=True)
+        (attn): Attention(
+          (qkv): Linear(in_features=1152, out_features=3456, bias=False)
+          (q_norm): LayerNorm((64,), eps=1e-06, elementwise_affine=True)
+          (k_norm): LayerNorm((64,), eps=1e-06, elementwise_affine=True)
+          (attn_drop): Dropout(p=0.0, inplace=False)
+          (norm): Identity()
+          (proj): Linear(in_features=1152, out_features=1152, bias=True)
+          (proj_drop): Dropout(p=0.0, inplace=False)
+        )
+        (ls1): Identity()
+        (drop_path1): Identity()
+        (norm2): LayerNorm((1152,), eps=1e-06, elementwise_affine=True)
+        (mlp): Mlp(
+          (fc1): Linear(in_features=1152, out_features=4608, bias=True)
+          (act): GELU(approximate='none')
+          (drop1): Dropout(p=0.0, inplace=False)
+          (norm): Identity()
+          (fc2): Linear(in_features=4608, out_features=1152, bias=True)
+          (drop2): Dropout(p=0.0, inplace=False)
+        )
+        (ls2): Identity()
+        (drop_path2): Identity()
+        (adaLN_modulation): Sequential(
+          (0): SiLU()
+          (1): Linear(in_features=1152, out_features=13824, bias=True)
+        )
+      )
+    )
+    (rope): HunyuanVideoRotaryPosEmbed()
+  )
+  (bottleneck): VectorQuantize(
+    (project_in): Linear(in_features=768, out_features=16, bias=True)
+    (project_out): Linear(in_features=16, out_features=1152, bias=True)
+    (_codebook): CosineSimCodebook()
+  )
+  (final_layer): AdaLNOutputLayer(
+    (norm_final): AdaLayerNormContinuous(
+      (silu): SiLU()
+      (linear): Linear(in_features=1152, out_features=2304, bias=True)
+      (norm): LayerNorm((1152,), eps=1e-06, elementwise_affine=False)
+    )
+    (linear): Linear(in_features=1152, out_features=768, bias=True)
+  )
+  (flow_matching_noise_module): MinRFNoiseModule()
+  (dec_time_embedder): TimestepEmbedder(
+    (mlp): Sequential(
+      (0): Linear(in_features=256, out_features=1152, bias=True)
+      (1): SiLU()
+      (2): Linear(in_features=1152, out_features=1152, bias=True)
+    )
+  )
+  (dec_x_embedder): PatchEmbed3D(
+    (proj): Conv3d(3, 1152, kernel_size=(4, 8, 8), stride=(4, 8, 8))
+    (norm): Identity()
+  )
+)
+[10-26 03:58:34] Model: #params=666.3M
+[10-26 03:58:34] SLURM_JOB_ID: None
+[10-26 03:58:34] SLUMR_ARRAY_JOB_ID: None
+[10-26 03:58:34] SLURM_ARRAY_TASK_ID: None
+[10-26 03:58:34] wandb_name: k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192_____
+[10-26 03:58:34] Compiling model with mode: default
+[10-26 03:58:55] Discriminator: #params=38.2M
+[10-26 03:58:55] compiling loss with mode default
+[10-26 03:58:55] Epoch 1 started.
+[10-26 05:34:28] Epoch 1 training done. Time: 5732.70s
+[10-26 06:03:25] Calculating FVD with running real stats
+[10-26 06:04:53] Converting video data to uint8
+[10-26 06:05:59] Converting video data to uint8
+[10-26 06:05:59] Preparing to save rng states...
+[10-26 06:05:59] Saving checkpoint...
+[10-26 06:06:35] New best checkpoint saved: best_fvd_7983.63.pth
+[10-26 06:06:35] Epoch 1, train: fm_loss=0.4965 perceptual_loss=0.7230 rp_loss=0.8580 g_loss=0.0000 g_loss_weight=0.0000 length_loss=0.0000 ref_loss=0.0000 ema_model_loss=0.0000 ce_loss=0.0000 psnr=11.8697 ssim=0.1345 loss_q=0.0000 index_usage_batch=0.0224 perplexity=6358.6162 loss_commit=0.0000 loss=0.8580 fps=24.2793, 
+ eval: ucf101_val_psnr=6.4218 ucf101_val_ssim=0.0048 ucf101_val_fps=3.1429 ucf101_val_fvd=7983.6285, 
+Latest checkpoint saved. Time: 35.78s
+, 2.1h (d 0.47) 2.1h/425.6h
+[10-26 06:06:35] Epoch 2 started.
+[10-26 06:58:07] Epoch 2 training done. Time: 3091.38s
+[10-26 07:00:34] Calculating FVD with running real stats
+[10-26 07:01:54] Converting video data to uint8
+[10-26 07:03:06] Converting video data to uint8
+[10-26 07:03:07] Preparing to save rng states...
+[10-26 07:03:07] Saving checkpoint...
+[10-26 07:03:44] New best checkpoint saved: best_fvd_4078.66.pth
+[10-26 07:03:44] Epoch 2, train: fm_loss=0.2040 perceptual_loss=0.6439 rp_loss=0.5259 g_loss=0.0000 g_loss_weight=0.0000 length_loss=0.0000 ref_loss=0.0000 ema_model_loss=0.0000 ce_loss=0.0000 psnr=15.9498 ssim=0.1968 loss_q=0.0002 index_usage_batch=0.0195 perplexity=7807.2646 loss_commit=0.0002 loss=0.5259 fps=23.9054, 
+ eval: ucf101_val_psnr=8.9663 ucf101_val_ssim=0.0189 ucf101_val_fps=3.1781 ucf101_val_fvd=4078.6604, 
+Latest checkpoint saved. Time: 37.56s
+, 57.2m (d 0.04) 3.1h/308.0h
+[10-26 07:03:44] Epoch 3 started.
+[10-26 07:56:18] Epoch 3 training done. Time: 3154.18s
+[10-26 07:58:45] Calculating FVD with running real stats
+[10-26 08:00:09] Converting video data to uint8
+[10-26 08:01:18] Converting video data to uint8
+[10-26 08:01:18] Preparing to save rng states...
+[10-26 08:01:18] Saving checkpoint...
+[10-26 08:01:58] New best checkpoint saved: best_fvd_3830.81.pth
+[10-26 08:01:58] Epoch 3, train: fm_loss=0.1239 perceptual_loss=0.5882 rp_loss=0.4180 g_loss=0.0000 g_loss_weight=0.0000 length_loss=0.0000 ref_loss=0.0000 ema_model_loss=0.0000 ce_loss=0.0000 psnr=18.5751 ssim=0.2626 loss_q=0.0003 index_usage_batch=0.0115 perplexity=7967.7891 loss_commit=0.0003 loss=0.4181 fps=23.6399, 
+ eval: ucf101_val_psnr=9.5385 ucf101_val_ssim=0.0304 ucf101_val_fps=3.1865 ucf101_val_fvd=3830.8149, 
+Latest checkpoint saved. Time: 39.61s
+, 58.2m (d 0.02) 4.1h/270.1h
+[10-26 08:01:58] Epoch 4 started.
+[10-26 08:54:48] Epoch 4 training done. Time: 3169.99s
+[10-26 08:57:15] Calculating FVD with running real stats
+[10-26 08:58:33] Converting video data to uint8
+[10-26 08:59:41] Converting video data to uint8
+[10-26 08:59:43] Preparing to save rng states...
+[10-26 08:59:49] Saving checkpoint...
+[10-26 09:00:27] New best checkpoint saved: best_fvd_3380.49.pth
+[10-26 09:00:27] Epoch 4, train: fm_loss=0.0941 perceptual_loss=0.5426 rp_loss=0.3654 g_loss=0.0000 g_loss_weight=0.0000 length_loss=0.0000 ref_loss=0.0000 ema_model_loss=0.0000 ce_loss=0.0000 psnr=20.0186 ssim=0.3119 loss_q=0.0008 index_usage_batch=0.0103 perplexity=8027.0205 loss_commit=0.0008 loss=0.3654 fps=23.6290, 
+ eval: ucf101_val_psnr=9.6446 ucf101_val_ssim=0.0359 ucf101_val_fps=3.1718 ucf101_val_fvd=3380.4911, 
+Latest checkpoint saved. Time: 44.50s
+, 58.5m (d 0.03) 5.0h/251.3h
+[10-26 09:00:27] Epoch 5 started.
+[10-26 09:53:14] Epoch 5 training done. Time: 3167.01s
+[10-26 09:55:41] Calculating FVD with running real stats
+[10-26 09:57:00] Converting video data to uint8
+[10-26 09:58:08] Converting video data to uint8
+[10-26 09:58:09] Preparing to save rng states...
+[10-26 09:58:10] Saving checkpoint...
+[10-26 09:58:47] New best checkpoint saved: best_fvd_2447.01.pth
+[10-26 09:58:47] Epoch 5, train: fm_loss=0.0751 perceptual_loss=0.5013 rp_loss=0.3258 g_loss=0.0000 g_loss_weight=0.0000 length_loss=0.0000 ref_loss=0.0000 ema_model_loss=0.0000 ce_loss=0.0000 psnr=21.1530 ssim=0.3623 loss_q=0.0020 index_usage_batch=0.0285 perplexity=8064.0542 loss_commit=0.0020 loss=0.3260 fps=23.6636, 
+ eval: ucf101_val_psnr=11.4611 ucf101_val_ssim=0.0623 ucf101_val_fps=3.1759 ucf101_val_fvd=2447.0120, 
+Latest checkpoint saved. Time: 37.86s
+, 58.3m (d 0.01) 6.0h/239.9h
+[10-26 09:58:47] Epoch 6 started.
+[10-26 10:50:47] Epoch 6 training done. Time: 3120.14s
+[10-26 10:53:16] Calculating FVD with running real stats
+[10-26 10:54:37] Converting video data to uint8
+[10-26 10:55:44] Converting video data to uint8
+[10-26 10:55:46] Preparing to save rng states...
+[10-26 10:55:49] Saving checkpoint...
+[10-26 10:56:25] New best checkpoint saved: best_fvd_2160.28.pth
+[10-26 10:56:25] Epoch 6, train: fm_loss=0.0625 perceptual_loss=0.4607 rp_loss=0.2928 g_loss=0.0000 g_loss_weight=0.0000 length_loss=0.0000 ref_loss=0.0000 ema_model_loss=0.0000 ce_loss=0.0000 psnr=22.0782 ssim=0.4143 loss_q=0.0022 index_usage_batch=0.1266 perplexity=8087.0312 loss_commit=0.0022 loss=0.2930 fps=23.6909, 
+ eval: ucf101_val_psnr=13.6264 ucf101_val_ssim=0.0992 ucf101_val_fps=3.1386 ucf101_val_fvd=2160.2844, 
+Latest checkpoint saved. Time: 39.57s
+, 57.6m (d 0.00) 7.0h/231.9h
+[10-26 10:56:25] Epoch 7 started.
+[10-26 11:45:07] Epoch 7 training done. Time: 2921.50s
+[10-26 11:47:34] Calculating FVD with running real stats
+[10-26 11:48:53] Converting video data to uint8
+[10-26 11:50:02] Converting video data to uint8
+[10-26 11:50:03] Preparing to save rng states...
+[10-26 11:50:06] Saving checkpoint...
+[10-26 11:50:42] New best checkpoint saved: best_fvd_1674.94.pth
+[10-26 11:50:42] Epoch 7, train: fm_loss=0.0541 perceptual_loss=0.4081 rp_loss=0.2581 g_loss=0.0000 g_loss_weight=0.0000 length_loss=0.0000 ref_loss=0.0000 ema_model_loss=0.0000 ce_loss=0.0000 psnr=22.7474 ssim=0.4579 loss_q=0.0018 index_usage_batch=0.2620 perplexity=8086.5469 loss_commit=0.0018 loss=0.2583 fps=24.0509, 
+ eval: ucf101_val_psnr=15.8265 ucf101_val_ssim=0.1793 ucf101_val_fps=3.1762 ucf101_val_fvd=1674.9365, 
+Latest checkpoint saved. Time: 38.41s
+, 54.3m (d 0.00) 7.9h/224.7h
+[10-26 11:50:42] Epoch 8 started.
+[10-26 12:38:25] Epoch 8 training done. Time: 2862.67s
+[10-26 12:40:51] Calculating FVD with running real stats
+[10-26 12:42:13] Converting video data to uint8
+[10-26 12:43:24] Converting video data to uint8
+[10-26 12:43:24] Preparing to save rng states...
+[10-26 12:43:24] Saving checkpoint...
+[10-26 12:43:58] New best checkpoint saved: best_fvd_1113.78.pth
+[10-26 12:43:58] Epoch 8, train: fm_loss=0.0474 perceptual_loss=0.3506 rp_loss=0.2227 g_loss=0.0000 g_loss_weight=0.0000 length_loss=0.0000 ref_loss=0.0000 ema_model_loss=0.0000 ce_loss=0.0000 psnr=23.4041 ssim=0.5031 loss_q=0.0020 index_usage_batch=0.4202 perplexity=8058.2388 loss_commit=0.0020 loss=0.2229 fps=24.2708, 
+ eval: ucf101_val_psnr=16.7288 ucf101_val_ssim=0.2578 ucf101_val_fps=3.1883 ucf101_val_fvd=1113.7755, 
+Latest checkpoint saved. Time: 33.97s
+, 53.3m (d 0.02) 8.8h/218.8h
+[10-26 12:43:58] Epoch 9 started.
+[10-26 13:31:57] Epoch 9 training done. Time: 2878.38s
+[10-26 13:34:23] Calculating FVD with running real stats
+[10-26 13:35:46] Converting video data to uint8
+[10-26 13:36:54] Converting video data to uint8
+[10-26 13:36:54] Preparing to save rng states...
+[10-26 13:36:54] Saving checkpoint...
+[10-26 13:37:29] New best checkpoint saved: best_fvd_827.66.pth
+[10-26 13:37:29] Epoch 9, train: fm_loss=0.0423 perceptual_loss=0.3054 rp_loss=0.1950 g_loss=0.0000 g_loss_weight=0.0000 length_loss=0.0000 ref_loss=0.0000 ema_model_loss=0.0000 ce_loss=0.0000 psnr=23.9759 ssim=0.5432 loss_q=0.0027 index_usage_batch=0.5526 perplexity=8024.5972 loss_commit=0.0027 loss=0.1952 fps=24.1793, 
+ eval: ucf101_val_psnr=18.4337 ucf101_val_ssim=0.3501 ucf101_val_fps=3.1849 ucf101_val_fvd=827.6603, 
+Latest checkpoint saved. Time: 34.83s
+, 53.5m (d 0.04) 9.6h/214.3h
+[10-26 13:37:29] Epoch 10 started.
+[10-26 14:25:32] Epoch 10 training done. Time: 2882.85s
+[10-26 14:27:58] Calculating FVD with running real stats
+[10-26 14:29:22] Converting video data to uint8
+[10-26 14:30:28] Converting video data to uint8
+[10-26 14:30:29] Preparing to save rng states...
+[10-26 14:30:29] Saving checkpoint...
+[10-26 14:31:05] New best checkpoint saved: best_fvd_610.79.pth
+[10-26 14:31:05] Epoch 10, train: fm_loss=0.0383 perceptual_loss=0.2705 rp_loss=0.1736 g_loss=0.0000 g_loss_weight=0.0000 length_loss=0.0000 ref_loss=0.0000 ema_model_loss=0.0000 ce_loss=0.0000 psnr=24.5157 ssim=0.5798 loss_q=0.0037 index_usage_batch=0.6788 perplexity=8023.9727 loss_commit=0.0037 loss=0.1740 fps=24.1696, 
+ eval: ucf101_val_psnr=19.2263 ucf101_val_ssim=0.4042 ucf101_val_fps=3.1910 ucf101_val_fvd=610.7945, 
+Latest checkpoint saved. Time: 36.02s
+, 53.6m (d 0.02) 10.5h/210.7h
+[10-26 14:31:05] Epoch 11 started.
+[10-26 15:19:04] Epoch 11 training done. Time: 2879.19s
+[10-26 15:21:30] Calculating FVD with running real stats
+[10-26 15:22:50] Converting video data to uint8
+[10-26 15:24:00] Converting video data to uint8
+[10-26 15:24:01] Preparing to save rng states...
+[10-26 15:24:01] Saving checkpoint...
+[10-26 15:24:37] New best checkpoint saved: best_fvd_381.65.pth
+[10-26 15:24:37] Epoch 11, train: fm_loss=0.0351 perceptual_loss=0.2409 rp_loss=0.1555 g_loss=0.0000 g_loss_weight=0.0000 length_loss=0.0000 ref_loss=0.0000 ema_model_loss=0.0000 ce_loss=0.0000 psnr=25.0424 ssim=0.6155 loss_q=0.0043 index_usage_batch=0.7751 perplexity=8051.9331 loss_commit=0.0043 loss=0.1559 fps=24.1750, 
+ eval: ucf101_val_psnr=20.3549 ucf101_val_ssim=0.4929 ucf101_val_fps=3.1925 ucf101_val_fvd=381.6456, 
+Latest checkpoint saved. Time: 36.06s
+, 53.5m (d 0.07) 11.4h/207.8h
+[10-26 15:24:37] Epoch 12 started.
+[10-26 16:12:57] Epoch 12 training done. Time: 2900.10s
+[10-26 16:15:24] Calculating FVD with running real stats
+[10-26 16:16:51] Converting video data to uint8
+[10-26 16:18:01] Converting video data to uint8
+[10-26 16:18:02] Preparing to save rng states...
+[10-26 16:18:02] Saving checkpoint...
+[10-26 16:18:37] New best checkpoint saved: best_fvd_323.49.pth
+[10-26 16:18:37] Epoch 12, train: fm_loss=0.0328 perceptual_loss=0.2185 rp_loss=0.1420 g_loss=0.0000 g_loss_weight=0.0000 length_loss=0.0000 ref_loss=0.0000 ema_model_loss=0.0000 ce_loss=0.0000 psnr=25.5186 ssim=0.6458 loss_q=0.0044 index_usage_batch=0.8184 perplexity=8070.7065 loss_commit=0.0044 loss=0.1425 fps=24.0804, 
+ eval: ucf101_val_psnr=21.4852 ucf101_val_ssim=0.5541 ucf101_val_fps=3.1760 ucf101_val_fvd=323.4891, 
+Latest checkpoint saved. Time: 35.00s
+, 54.0m (d 0.03) 12.3h/205.5h
+[10-26 16:18:37] Epoch 13 started.
+[10-26 17:07:51] Epoch 13 training done. Time: 2954.12s
+[10-26 17:10:17] Calculating FVD with running real stats
+[10-26 17:11:39] Converting video data to uint8
+[10-26 17:12:46] Converting video data to uint8
+[10-26 17:12:47] Preparing to save rng states...
+[10-26 17:12:47] Saving checkpoint...
+[10-26 17:13:23] New best checkpoint saved: best_fvd_270.49.pth
+[10-26 17:13:23] Epoch 13, train: fm_loss=0.0308 perceptual_loss=0.2025 rp_loss=0.1320 g_loss=0.0000 g_loss_weight=0.0000 length_loss=0.0000 ref_loss=0.0000 ema_model_loss=0.0000 ce_loss=0.0000 psnr=25.8723 ssim=0.6692 loss_q=0.0044 index_usage_batch=0.8237 perplexity=8076.1089 loss_commit=0.0044 loss=0.1325 fps=24.0107, 
+ eval: ucf101_val_psnr=21.3858 ucf101_val_ssim=0.5712 ucf101_val_fps=3.2044 ucf101_val_fvd=270.4910, 
+Latest checkpoint saved. Time: 35.80s
+, 54.8m (d 0.06) 13.2h/203.7h
+[10-26 17:13:23] Epoch 14 started.
+[10-26 18:01:49] Epoch 14 training done. Time: 2905.86s
+[10-26 18:04:15] Calculating FVD with running real stats
+[10-26 18:05:37] Converting video data to uint8
+[10-26 18:06:44] Converting video data to uint8
+[10-26 18:06:45] Preparing to save rng states...
+[10-26 18:06:45] Saving checkpoint...
+[10-26 18:07:23] New best checkpoint saved: best_fvd_266.42.pth
+[10-26 18:07:23] Epoch 14, train: fm_loss=0.0294 perceptual_loss=0.1902 rp_loss=0.1245 g_loss=0.0000 g_loss_weight=0.0000 length_loss=0.0000 ref_loss=0.0000 ema_model_loss=0.0000 ce_loss=0.0000 psnr=26.1650 ssim=0.6876 loss_q=0.0045 index_usage_batch=0.8265 perplexity=8077.9287 loss_commit=0.0045 loss=0.1249 fps=24.1597, 
+ eval: ucf101_val_psnr=22.1036 ucf101_val_ssim=0.6195 ucf101_val_fps=3.1930 ucf101_val_fvd=266.4189, 
+Latest checkpoint saved. Time: 37.61s
+, 54.0m (d 0.06) 14.1h/202.0h
+[10-26 18:07:23] Epoch 15 started.
+[10-26 18:55:32] Epoch 15 training done. Time: 2889.90s
+[10-26 18:57:59] Calculating FVD with running real stats
+[10-26 18:59:21] Converting video data to uint8
+[10-26 19:00:29] Converting video data to uint8
+[10-26 19:00:29] Preparing to save rng states...
+[10-26 19:00:29] Saving checkpoint...
+[10-26 19:01:04] New best checkpoint saved: best_fvd_218.01.pth
+[10-26 19:01:04] Epoch 15, train: fm_loss=0.0283 perceptual_loss=0.1776 rp_loss=0.1171 g_loss=0.0000 g_loss_weight=0.0000 length_loss=0.0000 ref_loss=0.0000 ema_model_loss=0.0000 ce_loss=0.0000 psnr=26.4998 ssim=0.7085 loss_q=0.0048 index_usage_batch=0.8284 perplexity=8078.5635 loss_commit=0.0048 loss=0.1176 fps=24.3035, 
+ eval: ucf101_val_psnr=22.0786 ucf101_val_ssim=0.6269 ucf101_val_fps=3.1910 ucf101_val_fvd=218.0063, 
+Latest checkpoint saved. Time: 34.62s
+, 53.7m (d 0.04) 15.0h/200.5h
+[10-26 19:01:04] Epoch 16 started.
+[10-26 19:49:32] Epoch 16 training done. Time: 2907.65s
+[10-26 19:51:57] Calculating FVD with running real stats
+[10-26 19:53:20] Converting video data to uint8
+[10-26 19:54:27] Converting video data to uint8
+[10-26 19:54:28] Preparing to save rng states...
+[10-26 19:54:28] Saving checkpoint...
+[10-26 19:54:39] Epoch 16, train: fm_loss=0.0273 perceptual_loss=0.1666 rp_loss=0.1106 g_loss=0.0000 g_loss_weight=0.0000 length_loss=0.0000 ref_loss=0.0000 ema_model_loss=0.0000 ce_loss=0.0000 psnr=26.7899 ssim=0.7275 loss_q=0.0060 index_usage_batch=0.8263 perplexity=8079.1172 loss_commit=0.0060 loss=0.1112 fps=24.1462, 
+ eval: ucf101_val_psnr=22.4575 ucf101_val_ssim=0.6505 ucf101_val_fps=3.2147 ucf101_val_fvd=221.7525, 
+Latest checkpoint saved. Time: 10.53s
+, 53.6m (d 0.02) 15.9h/199.1h
+[10-26 19:54:39] Epoch 17 started.
+[10-26 20:43:02] Epoch 17 training done. Time: 2903.26s
+[10-26 20:45:28] Calculating FVD with running real stats
+[10-26 20:46:53] Converting video data to uint8
+[10-26 20:47:59] Converting video data to uint8
+[10-26 20:48:00] Preparing to save rng states...
+[10-26 20:48:00] Saving checkpoint...
+[10-26 20:48:35] New best checkpoint saved: best_fvd_200.12.pth
+[10-26 20:48:35] Epoch 17, train: fm_loss=0.0262 perceptual_loss=0.1568 rp_loss=0.1046 g_loss=0.0000 g_loss_weight=0.0000 length_loss=0.0000 ref_loss=0.0000 ema_model_loss=0.0000 ce_loss=0.0000 psnr=27.0340 ssim=0.7445 loss_q=0.0061 index_usage_batch=0.8285 perplexity=8079.9019 loss_commit=0.0061 loss=0.1052 fps=24.0968, 
+ eval: ucf101_val_psnr=22.6389 ucf101_val_ssim=0.6659 ucf101_val_fps=3.1874 ucf101_val_fvd=200.1184, 
+Latest checkpoint saved. Time: 35.77s
+, 53.9m (d 0.07) 16.8h/198.0h
+[10-26 20:48:35] Epoch 18 started.
+[10-26 21:36:21] Epoch 18 training done. Time: 2865.34s
+[10-26 21:38:47] Calculating FVD with running real stats
+[10-26 21:40:11] Converting video data to uint8
+[10-26 21:41:17] Converting video data to uint8
+[10-26 21:41:18] Preparing to save rng states...
+[10-26 21:41:18] Saving checkpoint...
+[10-26 21:41:52] New best checkpoint saved: best_fvd_187.29.pth
+[10-26 21:41:52] Epoch 18, train: fm_loss=0.0255 perceptual_loss=0.1484 rp_loss=0.0997 g_loss=0.0000 g_loss_weight=0.0000 length_loss=0.0000 ref_loss=0.0000 ema_model_loss=0.0000 ce_loss=0.0000 psnr=27.2793 ssim=0.7591 loss_q=0.0062 index_usage_batch=0.8297 perplexity=8080.2930 loss_commit=0.0062 loss=0.1003 fps=24.1556, 
+ eval: ucf101_val_psnr=22.6294 ucf101_val_ssim=0.6638 ucf101_val_fps=3.2034 ucf101_val_fvd=187.2859, 
+Latest checkpoint saved. Time: 34.07s
+, 53.3m (d 0.07) 17.7h/196.8h
+[10-26 21:41:52] Epoch 19 started.
+[10-26 22:29:32] Epoch 19 training done. Time: 2860.19s
+[10-26 22:31:59] Calculating FVD with running real stats
+[10-26 22:33:21] Converting video data to uint8
+[10-26 22:34:32] Converting video data to uint8
+[10-26 22:34:33] Preparing to save rng states...
+[10-26 22:34:33] Saving checkpoint...
+[10-26 22:35:10] New best checkpoint saved: best_fvd_177.69.pth
+[10-26 22:35:10] Epoch 19, train: fm_loss=0.0250 perceptual_loss=0.1422 rp_loss=0.0961 g_loss=0.0000 g_loss_weight=0.0000 length_loss=0.0000 ref_loss=0.0000 ema_model_loss=0.0000 ce_loss=0.0000 psnr=27.4619 ssim=0.7701 loss_q=0.0062 index_usage_batch=0.8307 perplexity=8080.1904 loss_commit=0.0062 loss=0.0967 fps=24.0560, 
+ eval: ucf101_val_psnr=23.0714 ucf101_val_ssim=0.6980 ucf101_val_fps=3.1797 ucf101_val_fvd=177.6935, 
+Latest checkpoint saved. Time: 37.20s
+, 53.3m (d 0.05) 18.6h/195.8h
+[10-26 22:35:10] Epoch 20 started.
+[10-26 23:23:17] Epoch 20 training done. Time: 2886.99s
+[10-26 23:25:44] Calculating FVD with running real stats
+[10-26 23:27:10] Converting video data to uint8
+[10-26 23:28:17] Converting video data to uint8
+[10-26 23:28:18] Preparing to save rng states...
+[10-26 23:28:18] Saving checkpoint...
+[10-26 23:28:52] New best checkpoint saved: best_fvd_164.04.pth
+[10-26 23:28:52] Epoch 20, train: fm_loss=0.0244 perceptual_loss=0.1366 rp_loss=0.0927 g_loss=0.0000 g_loss_weight=0.0000 length_loss=0.0000 ref_loss=0.0000 ema_model_loss=0.0000 ce_loss=0.0000 psnr=27.6406 ssim=0.7801 loss_q=0.0062 index_usage_batch=0.8317 perplexity=8079.9258 loss_commit=0.0062 loss=0.0933 fps=24.3060, 
+ eval: ucf101_val_psnr=23.0348 ucf101_val_ssim=0.7059 ucf101_val_fps=3.1822 ucf101_val_fvd=164.0369, 
+Latest checkpoint saved. Time: 34.57s
+, 53.7m (d 0.03) 19.5h/195.0h
+[10-26 23:28:52] Epoch 21 started.
+[10-27 00:16:29] Epoch 21 training done. Time: 2856.47s
+[10-27 00:18:57] Calculating FVD with running real stats
+[10-27 00:20:23] Converting video data to uint8
+[10-27 00:21:30] Converting video data to uint8
+[10-27 00:21:30] Preparing to save rng states...
+[10-27 00:21:30] Saving checkpoint...
+[10-27 00:21:41] Epoch 21, train: fm_loss=0.0239 perceptual_loss=0.1321 rp_loss=0.0900 g_loss=0.0000 g_loss_weight=0.0000 length_loss=0.0000 ref_loss=0.0000 ema_model_loss=0.0000 ce_loss=0.0000 psnr=27.7904 ssim=0.7884 loss_q=0.0062 index_usage_batch=0.8325 perplexity=8079.4365 loss_commit=0.0062 loss=0.0906 fps=24.3157, 
+ eval: ucf101_val_psnr=22.6612 ucf101_val_ssim=0.6816 ucf101_val_fps=3.1624 ucf101_val_fvd=164.3932, 
+Latest checkpoint saved. Time: 10.43s
+, 52.8m (d 0.06) 20.4h/194.1h
+[10-27 00:21:41] Epoch 22 started.
+[10-27 01:09:35] Epoch 22 training done. Time: 2874.68s
+[10-27 01:12:03] Calculating FVD with running real stats
+[10-27 01:13:29] Converting video data to uint8
+[10-27 01:14:36] Converting video data to uint8
+[10-27 01:14:36] Preparing to save rng states...
+[10-27 01:14:36] Saving checkpoint...
+[10-27 01:15:11] New best checkpoint saved: best_fvd_151.94.pth
+[10-27 01:15:11] Epoch 22, train: fm_loss=0.0234 perceptual_loss=0.1282 rp_loss=0.0875 g_loss=0.0000 g_loss_weight=0.0000 length_loss=0.0000 ref_loss=0.0000 ema_model_loss=0.0000 ce_loss=0.0000 psnr=27.9169 ssim=0.7950 loss_q=0.0063 index_usage_batch=0.8332 perplexity=8078.8394 loss_commit=0.0063 loss=0.0881 fps=24.1343, 
+ eval: ucf101_val_psnr=23.1404 ucf101_val_ssim=0.6852 ucf101_val_fps=3.1745 ucf101_val_fvd=151.9350, 
+Latest checkpoint saved. Time: 34.47s
+, 53.5m (d 0.05) 21.3h/193.4h
+[10-27 01:15:11] Epoch 23 started.
+[10-27 02:03:16] Epoch 23 training done. Time: 2885.42s
+[10-27 02:05:44] Calculating FVD with running real stats
+[10-27 02:07:08] Converting video data to uint8
+[10-27 02:08:16] Converting video data to uint8
+[10-27 02:08:17] Preparing to save rng states...
+[10-27 02:08:17] Saving checkpoint...
+[10-27 02:08:53] New best checkpoint saved: best_fvd_138.58.pth
+[10-27 02:08:53] Epoch 23, train: fm_loss=0.0231 perceptual_loss=0.1252 rp_loss=0.0857 g_loss=0.0000 g_loss_weight=0.0000 length_loss=0.0000 ref_loss=0.0000 ema_model_loss=0.0000 ce_loss=0.0000 psnr=28.0422 ssim=0.8006 loss_q=0.0063 index_usage_batch=0.8338 perplexity=8077.9961 loss_commit=0.0063 loss=0.0864 fps=24.0963, 
+ eval: ucf101_val_psnr=23.3009 ucf101_val_ssim=0.7058 ucf101_val_fps=3.1746 ucf101_val_fvd=138.5815, 
+Latest checkpoint saved. Time: 36.40s
+, 53.7m (d 0.05) 22.2h/192.8h
+[10-27 02:08:53] Epoch 24 started.
+[10-27 02:57:06] Epoch 24 training done. Time: 2892.43s
+[10-27 02:59:32] Calculating FVD with running real stats
+[10-27 03:00:55] Converting video data to uint8
+[10-27 03:02:04] Converting video data to uint8
+[10-27 03:02:05] Preparing to save rng states...
+[10-27 03:02:05] Saving checkpoint...
+[10-27 03:02:40] New best checkpoint saved: best_fvd_130.39.pth
+[10-27 03:02:40] Epoch 24, train: fm_loss=0.0227 perceptual_loss=0.1221 rp_loss=0.0837 g_loss=0.0000 g_loss_weight=0.0000 length_loss=0.0000 ref_loss=0.0000 ema_model_loss=0.0000 ce_loss=0.0000 psnr=28.1448 ssim=0.8057 loss_q=0.0063 index_usage_batch=0.8342 perplexity=8077.2231 loss_commit=0.0063 loss=0.0843 fps=24.2673, 
+ eval: ucf101_val_psnr=23.4273 ucf101_val_ssim=0.7316 ucf101_val_fps=3.1914 ucf101_val_fvd=130.3938, 
+Latest checkpoint saved. Time: 35.15s
+, 53.8m (d 0.04) 23.1h/192.2h
+[10-27 03:02:40] Epoch 25 started.

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/log_2025_10_27_05_21_46.txt

@@ -0,0 +1,156 @@
+[10-27 05:21:48] Distributed training enabled.
+[10-27 05:21:48] Environment setup done.
+[10-27 05:21:50] Train dataset: len=435743
+[10-27 05:21:50] Test dataset: ucf101_val, len=3783
+[10-27 05:21:50] Resuming training from /scratch/Projects/CFP-01/CFP01-CF-033/lingmin/1d_tokenizer/save/tokenizer_ours/tokenizer_e12_d18_k600/k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/epoch-last.pth
+[10-27 05:22:20] DYNTokenizer(
+  (x_embedder): PatchEmbed3D(
+    (proj): Conv3d(3, 768, kernel_size=(4, 8, 8), stride=(4, 8, 8))
+    (norm): Identity()
+  )
+  (encoder): TransformerEncoderParallel(
+    (blocks): ModuleList(
+      (0-11): 12 x AttentionBlock(
+        (norm1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
+        (attn): Attention(
+          (qkv): Linear(in_features=768, out_features=2304, bias=False)
+          (q_norm): LayerNorm((64,), eps=1e-06, elementwise_affine=True)
+          (k_norm): LayerNorm((64,), eps=1e-06, elementwise_affine=True)
+          (attn_drop): Dropout(p=0.0, inplace=False)
+          (norm): Identity()
+          (proj): Linear(in_features=768, out_features=768, bias=True)
+          (proj_drop): Dropout(p=0.0, inplace=False)
+        )
+        (ls1): Identity()
+        (drop_path1): Identity()
+        (norm2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
+        (mlp): Mlp(
+          (fc1): Linear(in_features=768, out_features=3072, bias=True)
+          (act): GELU(approximate='none')
+          (drop1): Dropout(p=0.0, inplace=False)
+          (norm): Identity()
+          (fc2): Linear(in_features=3072, out_features=768, bias=True)
+          (drop2): Dropout(p=0.0, inplace=False)
+        )
+        (ls2): Identity()
+        (drop_path2): Identity()
+      )
+    )
+  )
+  (decoder): TransformerAdaLNDecoderParallel(
+    (blocks): ModuleList(
+      (0-17): 18 x DiffusionAttentionBlock(
+        (norm1): LayerNorm((1152,), eps=1e-06, elementwise_affine=True)
+        (attn): Attention(
+          (qkv): Linear(in_features=1152, out_features=3456, bias=False)
+          (q_norm): LayerNorm((64,), eps=1e-06, elementwise_affine=True)
+          (k_norm): LayerNorm((64,), eps=1e-06, elementwise_affine=True)
+          (attn_drop): Dropout(p=0.0, inplace=False)
+          (norm): Identity()
+          (proj): Linear(in_features=1152, out_features=1152, bias=True)
+          (proj_drop): Dropout(p=0.0, inplace=False)
+        )
+        (ls1): Identity()
+        (drop_path1): Identity()
+        (norm2): LayerNorm((1152,), eps=1e-06, elementwise_affine=True)
+        (mlp): Mlp(
+          (fc1): Linear(in_features=1152, out_features=4608, bias=True)
+          (act): GELU(approximate='none')
+          (drop1): Dropout(p=0.0, inplace=False)
+          (norm): Identity()
+          (fc2): Linear(in_features=4608, out_features=1152, bias=True)
+          (drop2): Dropout(p=0.0, inplace=False)
+        )
+        (ls2): Identity()
+        (drop_path2): Identity()
+        (adaLN_modulation): Sequential(
+          (0): SiLU()
+          (1): Linear(in_features=1152, out_features=13824, bias=True)
+        )
+      )
+    )
+    (rope): HunyuanVideoRotaryPosEmbed()
+  )
+  (bottleneck): VectorQuantize(
+    (project_in): Linear(in_features=768, out_features=16, bias=True)
+    (project_out): Linear(in_features=16, out_features=1152, bias=True)
+    (_codebook): CosineSimCodebook()
+  )
+  (final_layer): AdaLNOutputLayer(
+    (norm_final): AdaLayerNormContinuous(
+      (silu): SiLU()
+      (linear): Linear(in_features=1152, out_features=2304, bias=True)
+      (norm): LayerNorm((1152,), eps=1e-06, elementwise_affine=False)
+    )
+    (linear): Linear(in_features=1152, out_features=768, bias=True)
+  )
+  (flow_matching_noise_module): MinRFNoiseModule()
+  (dec_time_embedder): TimestepEmbedder(
+    (mlp): Sequential(
+      (0): Linear(in_features=256, out_features=1152, bias=True)
+      (1): SiLU()
+      (2): Linear(in_features=1152, out_features=1152, bias=True)
+    )
+  )
+  (dec_x_embedder): PatchEmbed3D(
+    (proj): Conv3d(3, 1152, kernel_size=(4, 8, 8), stride=(4, 8, 8))
+    (norm): Identity()
+  )
+)
+[10-27 05:22:20] Model: #params=666.3M
+[10-27 05:22:20] SLURM_JOB_ID: None
+[10-27 05:22:20] SLUMR_ARRAY_JOB_ID: None
+[10-27 05:22:20] SLURM_ARRAY_TASK_ID: None
+[10-27 05:22:20] wandb_name: k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192_____
+[10-27 05:22:20] Compiling model with mode: default
+[10-27 05:22:37] Discriminator: #params=38.2M
+[10-27 05:22:37] compiling loss with mode default
+[10-27 05:22:37] Epoch 25 started.
+[10-27 06:55:17] Epoch 25 training done. Time: 5560.63s
+[10-27 07:24:16] Calculating FVD with running real stats
+[10-27 07:25:45] Converting video data to uint8
+[10-27 07:26:48] Converting video data to uint8
+[10-27 07:26:48] Preparing to save rng states...
+[10-27 07:26:48] Saving checkpoint...
+[10-27 07:27:24] New best checkpoint saved: best_fvd_121.09.pth
+[10-27 07:27:24] Epoch 25, train: fm_loss=0.0224 perceptual_loss=0.1194 rp_loss=0.0821 g_loss=0.0000 g_loss_weight=0.0000 length_loss=0.0000 ref_loss=0.0000 ema_model_loss=0.0000 ce_loss=0.0000 psnr=28.2606 ssim=0.8104 loss_q=0.0063 index_usage_batch=0.8348 perplexity=8076.3428 loss_commit=0.0063 loss=0.0827 fps=24.2888, 
+ eval: ucf101_val_psnr=23.3652 ucf101_val_ssim=0.7205 ucf101_val_fps=3.1465 ucf101_val_fvd=121.0894, 
+Latest checkpoint saved. Time: 35.08s
+, 2.1h (d 0.47) 2.1h/415.9h
+[10-27 07:27:24] Epoch 26 started.
+[10-27 08:15:04] Epoch 26 training done. Time: 2860.67s
+[10-27 08:17:31] Calculating FVD with running real stats
+[10-27 08:18:53] Converting video data to uint8
+[10-27 08:20:01] Converting video data to uint8
+[10-27 08:20:02] Preparing to save rng states...
+[10-27 08:20:02] Saving checkpoint...
+[10-27 08:20:37] New best checkpoint saved: best_fvd_107.29.pth
+[10-27 08:20:37] Epoch 26, train: fm_loss=0.0219 perceptual_loss=0.1175 rp_loss=0.0807 g_loss=0.0000 g_loss_weight=0.0000 length_loss=0.0000 ref_loss=0.0000 ema_model_loss=0.0000 ce_loss=0.0000 psnr=28.3334 ssim=0.8138 loss_q=0.0063 index_usage_batch=0.8352 perplexity=8075.3994 loss_commit=0.0063 loss=0.0813 fps=24.1446, 
+ eval: ucf101_val_psnr=23.3940 ucf101_val_ssim=0.7037 ucf101_val_fps=3.1935 ucf101_val_fvd=107.2903, 
+Latest checkpoint saved. Time: 35.02s
+, 53.2m (d 0.06) 3.0h/296.7h
+[10-27 08:20:37] Epoch 27 started.
+[10-27 09:08:13] Epoch 27 training done. Time: 2856.72s
+[10-27 09:10:40] Calculating FVD with running real stats
+[10-27 09:12:02] Converting video data to uint8
+[10-27 09:13:10] Converting video data to uint8
+[10-27 09:13:11] Preparing to save rng states...
+[10-27 09:13:11] Saving checkpoint...
+[10-27 09:13:47] New best checkpoint saved: best_fvd_106.67.pth
+[10-27 09:13:47] Epoch 27, train: fm_loss=0.0217 perceptual_loss=0.1151 rp_loss=0.0792 g_loss=0.0000 g_loss_weight=0.0000 length_loss=0.0000 ref_loss=0.0000 ema_model_loss=0.0000 ce_loss=0.0000 psnr=28.4303 ssim=0.8178 loss_q=0.0063 index_usage_batch=0.8351 perplexity=8074.3740 loss_commit=0.0063 loss=0.0799 fps=24.0925, 
+ eval: ucf101_val_psnr=23.7231 ucf101_val_ssim=0.7382 ucf101_val_fps=3.1962 ucf101_val_fvd=106.6714, 
+Latest checkpoint saved. Time: 35.97s
+, 53.2m (d 0.02) 3.9h/256.9h
+[10-27 09:13:47] Epoch 28 started.
+[10-27 10:01:01] Epoch 28 training done. Time: 2834.39s
+[10-27 10:03:27] Calculating FVD with running real stats
+[10-27 10:04:50] Converting video data to uint8
+[10-27 10:05:58] Converting video data to uint8
+[10-27 10:05:59] Preparing to save rng states...
+[10-27 10:05:59] Saving checkpoint...
+[10-27 10:06:35] New best checkpoint saved: best_fvd_105.24.pth
+[10-27 10:06:35] Epoch 28, train: fm_loss=0.0215 perceptual_loss=0.1129 rp_loss=0.0780 g_loss=0.0000 g_loss_weight=0.0000 length_loss=0.0000 ref_loss=0.0000 ema_model_loss=0.0000 ce_loss=0.0000 psnr=28.5272 ssim=0.8215 loss_q=0.0063 index_usage_batch=0.8354 perplexity=8073.2065 loss_commit=0.0063 loss=0.0786 fps=24.1403, 
+ eval: ucf101_val_psnr=23.6395 ucf101_val_ssim=0.7394 ucf101_val_fps=3.2026 ucf101_val_fvd=105.2365, 
+Latest checkpoint saved. Time: 35.69s
+, 52.8m (d 0.03) 4.7h/236.6h
+[10-27 10:06:35] Epoch 29 started.

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/tensorboard/events.out.tfevents.1761422299.hopper-26.3880041.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0490cee65ce3c964710407235680e3fdf961cf02a6449fc6c9f8687a7a38833e
+size 29464

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/tensorboard/events.out.tfevents.1761513706.hopper-10.716544.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8d35c4937d378b988c0ab9a41b6d878d2282760c87e250ddb62efe3f706a945e
+size 4984

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/debug-internal.log

@@ -0,0 +1,7 @@
+{"time":"2025-10-27T05:22:02.60922833+08:00","level":"INFO","msg":"stream: starting","core version":"0.19.9","symlink path":"/scratch/Projects/CFP-01/CFP01-CF-033/lingmin/1d_tokenizer/save/tokenizer_ours/tokenizer_e12_d18_k600/k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251027_052202-c3u3silm/logs/debug-core.log"}
+{"time":"2025-10-27T05:22:03.20175398+08:00","level":"INFO","msg":"created new stream","id":"c3u3silm"}
+{"time":"2025-10-27T05:22:03.202333506+08:00","level":"INFO","msg":"stream: started","id":"c3u3silm"}
+{"time":"2025-10-27T05:22:03.202357497+08:00","level":"INFO","msg":"writer: Do: started","stream_id":"c3u3silm"}
+{"time":"2025-10-27T05:22:03.203578697+08:00","level":"INFO","msg":"sender: started","stream_id":"c3u3silm"}
+{"time":"2025-10-27T05:22:03.20239453+08:00","level":"INFO","msg":"handler: started","stream_id":"c3u3silm"}
+{"time":"2025-10-27T05:22:03.818476109+08:00","level":"INFO","msg":"Starting system monitor"}

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/debug.log

@@ -0,0 +1,23 @@
+2025-10-27 05:22:02,538 INFO    MainThread:716544 [wandb_setup.py:_flush():67] Current SDK version is 0.19.9
+2025-10-27 05:22:02,539 INFO    MainThread:716544 [wandb_setup.py:_flush():67] Configure stats pid to 716544
+2025-10-27 05:22:02,539 INFO    MainThread:716544 [wandb_setup.py:_flush():67] Loading settings from /home/svu/e0724392/.config/wandb/settings
+2025-10-27 05:22:02,539 INFO    MainThread:716544 [wandb_setup.py:_flush():67] Loading settings from /scratch/e0724392/work4/LARP/wandb/settings
+2025-10-27 05:22:02,540 INFO    MainThread:716544 [wandb_setup.py:_flush():67] Loading settings from environment variables
+2025-10-27 05:22:02,541 INFO    MainThread:716544 [wandb_init.py:setup_run_log_directory():662] Logging user logs to /scratch/Projects/CFP-01/CFP01-CF-033/lingmin/1d_tokenizer/save/tokenizer_ours/tokenizer_e12_d18_k600/k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251027_052202-c3u3silm/logs/debug.log
+2025-10-27 05:22:02,542 INFO    MainThread:716544 [wandb_init.py:setup_run_log_directory():663] Logging internal logs to /scratch/Projects/CFP-01/CFP01-CF-033/lingmin/1d_tokenizer/save/tokenizer_ours/tokenizer_e12_d18_k600/k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251027_052202-c3u3silm/logs/debug-internal.log
+2025-10-27 05:22:02,543 INFO    MainThread:716544 [wandb_init.py:init():781] calling init triggers
+2025-10-27 05:22:02,543 INFO    MainThread:716544 [wandb_init.py:init():786] wandb.init called with sweep_config: {}
+config: {'trainer': 'our_tokenizer_trainer', 'train_dataset': {'name': 'video_dataset', 'args': {'root_path': 'data/metadata', 'split': 'train', 'frame_num': 16, 'rand_augment': 'no', 'csv_file': 'k600_train.csv+ucf101_train.csv', 'cls_vid_num': '-1_-1', 'crop_size': 128, 'scale': 1.0, 'aspect_ratio': 1.0, 'rand_flip': 'yes', 'use_all_frames': False, 'pre_load': False}, 'loader': {'batch_size': 128, 'num_workers': 32}}, 'test_dataset': {'name': 'video_dataset', 'args': {'root_path': 'data/metadata', 'frame_num': 16, 'cls_vid_num': '-1_-1', 'crop_size': 128, 'use_all_frames': False, 'pre_load': False}, 'csv_paths': {'ucf101_val': 'ucf101_val.csv'}, 'loader': {'batch_size': 128, 'num_workers': 32}}, 'model': {'name': 'dyn_tokenizer', 'args': {'noise_schedule': {'name': 'min_rf_noise_module', 'args': {'clean_data_read_key': 'clean_data', 'noised_data_write_key': 'noisy_input', 'noise_write_key': 'flow_noise', 'timesteps_write_key': 'timesteps', 'sigmas_write_key': 'sigmas', 'ln': False, 'stratisfied': False, 'mode_scale': 0.25}}, 'bottleneck': {'name': 'bottleneck', 'args': {'regularizer': {'name': 'vector_quantize', 'args': {'codebook_dim': 16, 'codebook_size': 8192, 'ema_update': True, 'decay': 0.99, 'kmeans_init': True, 'kmeans_iters': 10, 'threshold_ema_dead_code': 0.2, 'use_cosine_sim': True, 'commitment_weight': 1.0, 'diversity_weight': 0.0, 'smart_re_K': 0, 'continuous': False, 'reg': [0.1, 0.3], 'reset_cluster_size': 0.2, 'ema_entropy_ratio': 0.8, 'vq_start_step': 0}}}}, 'prior_model': {'name': 'none', 'use_mix_ss': True, 'mix_ss_max_ratio': 0.5, 'mix_ss_peak_steps_ratio': 0.3, 'n_rounds': 2, 'avg_loss_over_rounds': True, 'no_grad_before_last_round': False, 'no_dropout': False, 'latent_ce_temperature': 1.0, 'args': {'l2_normalized': True}}, 'dec_time_embedder': {'name': 'timestep_embedder', 'args': {'timesteps_read_key': 'timesteps', 'time_embedding_write_key': 'dec_temb', 'dim': 1152, 'frequency_embedding_size': 256, 'max_timestep': 1000.0}}, 'transformer_name': 'transformer_encoder_parallel', 'encoder_name': 'none', 'decoder_name': 'transformer_AdaLN_decoder_parallel', 'bottleneck_token_num': 1024, 'input_size': 128, 'frame_num': 16, 'temporal_patch_size': 4, 'patch_size': 8, 'decoder_temporal_patch_size': 4, 'decoder_patch_size': 8, 'in_channels': 3, 'encoder_hidden_size': 768, 'decoder_hidden_size': 1152, 'encoder_num_heads': 12, 'decoder_num_heads': 18, 'encoder_depth': 12, 'decoder_depth': 18, 'encoder_block_name': 'block_timm', 'decoder_block_name': 'adaLN_block_timm', 'encoder_mask_mode': 'full', 'decoder_mask_mode': 'full', 'learned_encoder_patch_pe': False, 'learned_encoder_latent_query_embed': True, 'learned_decoder_latent_pe': False, 'learned_decoder_patch_query_embed': False, 'use_encoder_patch_token_type_embed': False, 'use_encoder_latent_query_token_type_embed': False, 'enable_decoder_query': False, 'learned_decoder_pe': False, 'use_decoder_latent_token_type_embed': False, 'use_decoder_patch_query_token_type_embed': True, 'encoder_query_gaussian_init': True, 'latent_pe_scale_factor': 10000, 'query_init_std': 0.02, 'adaLN_expansion': 2, 'final_layer_init': 'xavier_uniform', 'enable_vq': True, 'qk_norm': True, 'use_rope': True, 'rope_dim': [16, 24, 24], 'final_layer_type': 'adanorm'}}, 'loss': {'name': 'fm_disc_loss', 'args': {'disc_type': 'transformer', 'disc_start': 999999, 'disc_self_start': -1, 'perceptual_weight': 0.5, 'perceptual_loss': 'lpips', 'perceptual_fp16': False, 'lecam_weight': 0.001, 'disc_loss': 'ns_smooth', 'disc_weight': 0.0, 'r1_gp_weight': 0.0, 'd_update_freq': 5, 'spectral_norm': False, 'disc_tran_hidden_size': 512, 'disc_tran_n_heads': 8, 'disc_tran_n_layers': 12, 'disc_tran_temporal_patch_size': 4, 'disc_tran_patch_size': 8, 'input_spatial_size': 128, 'frame_num': 16, 'fm_loss_weight': 1.0}}, 'optimizer': {'name': 'adamw', 'loss_name': 'adam', 'args': {'lr': 0.0001, 'betas': [0.9, 0.99]}, 'loss_args': {'lr': 3e-05, 'betas': [0.5, 0.9]}, 'lr_type': 'step', 'lr_step_pcts': '0.9_0.95', 'warmup_epoch': 10, 'min_lr_mult': 0.1, 'prior_lr_mult': 1.0, 'emb_lr_mult': 1.0}, 'max_epoch': 200, 'eval_epoch': 1, 'vis_epoch': 1, 'latest_interval': 1, 'save_epoch': 100000000, 'save_best': True, 'stepwise_logging': False, 'ema_decay': '_', 'use_amp': True, 'amp_dtype': 'float16', 'compile': True, 'compile_mode': 'default', 'flash_attn': False, 'loss_q_weight': 0.1, 'loss_q_warmup': '1.0_1', 'loss_kl_weight': 0.0, 'kl_decay_epoch': -1, 'loss_latent_ce_weight': 0.0, 'sqt_start_end_epoch': '0.0_0.0_0', 'clip_grad_max_norm': 0.0, 'init_checkpoint': '', 'timesteps': 25, 'verbose': False, 'guidance_scale': 1.0, 'env': {'tot_gpus': 8, 'cudnn': False, 'wandb_upload': True, 'wandb_entity': 'lingmin', 'wandb_project': 'dyn_tokenizer', 'exp_name': 'k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__', 'save_dir': '/scratch/Projects/CFP-01/CFP01-CF-033/lingmin/1d_tokenizer/save/tokenizer_ours/tokenizer_e12_d18_k600/k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__', 'port': '12778'}, 'comment': '', 'manualSeed': 66667, 'TrainSize': 435743, 'TestSize_ucf101_val': 3783, '_wandb': {}}
+2025-10-27 05:22:02,544 INFO    MainThread:716544 [wandb_init.py:init():809] starting backend
+2025-10-27 05:22:02,544 INFO    MainThread:716544 [wandb_init.py:init():813] sending inform_init request
+2025-10-27 05:22:02,596 INFO    MainThread:716544 [backend.py:_multiprocessing_setup():101] multiprocessing start_methods=fork,spawn,forkserver, using: spawn
+2025-10-27 05:22:02,597 INFO    MainThread:716544 [wandb_init.py:init():823] backend started and connected
+2025-10-27 05:22:02,602 INFO    MainThread:716544 [wandb_init.py:init():915] updated telemetry
+2025-10-27 05:22:02,731 INFO    MainThread:716544 [wandb_init.py:init():939] communicating run to backend with 90.0 second timeout
+2025-10-27 05:22:03,767 INFO    MainThread:716544 [wandb_init.py:init():1009] run resumed
+2025-10-27 05:22:03,771 INFO    MainThread:716544 [wandb_init.py:init():1014] starting run threads in backend
+2025-10-27 05:22:10,370 INFO    MainThread:716544 [wandb_run.py:_console_start():2454] atexit reg
+2025-10-27 05:22:10,371 INFO    MainThread:716544 [wandb_run.py:_redirect():2306] redirect: wrap_raw
+2025-10-27 05:22:10,372 INFO    MainThread:716544 [wandb_run.py:_redirect():2371] Wrapping output streams.
+2025-10-27 05:22:10,373 INFO    MainThread:716544 [wandb_run.py:_redirect():2394] Redirects installed.
+2025-10-27 05:22:10,663 INFO    MainThread:716544 [wandb_init.py:init():1056] run started, returning control to user process

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_10_0dcd7543861eb951d9eb.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0dcd7543861eb951d9eb9128002ad8c13ed359ea83bdae4393d4ce3a895546af
+size 1447271

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_11_c440b37dfe11e1ecb291.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c440b37dfe11e1ecb291dc667a185db861185b5e562915887c95bb58e07c1ad2
+size 1361429

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_12_2c6abd01594c9f04b701.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2c6abd01594c9f04b701e5263b60f0493aa59c673e9fdf04efceac6c15ddebb5
+size 1104671

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_13_07ed7abcba2550022a65.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:07ed7abcba2550022a651a87086ca880859a600b134152317b5406cb35652c37
+size 1061247

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_14_1d19ba838a2636f6bbb9.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1d19ba838a2636f6bbb97e8cdab2a1a1ad413ba47d36269393ae97552278e130
+size 971039

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_15_e4b12dc26b33b48899cd.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e4b12dc26b33b48899cd40408810e3fabd1ad83a452c333eb2304576fb6e7b67
+size 898506

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_16_be836e9a23353717c38d.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:be836e9a23353717c38d1bc7a52e75ac416615e04c76ace755ce177335410c9f
+size 978642

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_17_db636a27733aa82a9288.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:db636a27733aa82a9288e3b388c4ceb5089ba746d4909ed6adbe141d96ac2603
+size 1003627

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_18_8520ee5928f1948584de.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8520ee5928f1948584de506a2dee7748d47912c71657efecd4cc3d6f7e21d6ab
+size 975965

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_19_9902d6a8e4fcef1843c7.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9902d6a8e4fcef1843c7e85204d9a67bb8aea067506b585f8b84b84a1f094404
+size 882054

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_1_4905dd5ddf36de8fa967.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4905dd5ddf36de8fa9671e752a11f477c07e6e64107d7df50f65d37fe9686f15
+size 6065915

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_20_430fdf292106f4cdd169.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:430fdf292106f4cdd169295b60d4b53c1bded466a659a13b42e361724689ed80
+size 937420

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_21_b16346a94669cd22074d.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b16346a94669cd22074d836082da0f4dc64523c512248d294be70e676755758d
+size 899995

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_22_b9cd0a98c6efd38dd326.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b9cd0a98c6efd38dd32640094344de5687e1d0ee16bc09acc0831ccd0a478ccd
+size 952132

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_23_16c5160b84b0b93c4d9a.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:16c5160b84b0b93c4d9a35723ac23dc74080d6816c9c386900a0d6aae7e7cbf3
+size 911203

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_24_caa9fc05e947e5c68fe5.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:caa9fc05e947e5c68fe571c83d411c21619387b375f3c3962e4255d625363608
+size 895463

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_2_d15bd6c3ab28bcf2d818.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d15bd6c3ab28bcf2d818887cc710cd9e8c41e7f000f59b09cb1efcad20a3b4cb
+size 4538287

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_3_53d7f7f93663c75403c5.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:53d7f7f93663c75403c55ecf763a0a4cb179d042312633f62bdd97228db1c9e9
+size 3564898

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_4_518c61a588fe34e8b97b.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:518c61a588fe34e8b97b4d5a847ca4503fc147203f2ba655c6c51036ab9f11e0
+size 3098325

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_5_a718cd301e53d86c413d.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a718cd301e53d86c413d314d81c01ae2be78483c38b325fb0f74a8f015dc934f
+size 2490105

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_6_75eb65b267e94c10e7e6.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:75eb65b267e94c10e7e6736fbc4f28aa33c443da8280a22360cdb26deb1b4fb1
+size 2170170

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_7_7f140abb6a7cc547d17d.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7f140abb6a7cc547d17d4e3b1407f3c67121025bb26aa8918cdaa742227f1963
+size 2086638

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_8_4b22b9cea46a63503ee2.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4b22b9cea46a63503ee26cb4c3c76f9aa4879d911efe88e213f43d5151b4eaba
+size 1653335

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_test_dataset_9_09d0cffdb29ebf65107e.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:09d0cffdb29ebf65107ec0f593ad28c54ff88651007c137a2788cdc878771660
+size 1545553

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_10_cb55eb08e79a819f9319.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cb55eb08e79a819f931958929f93fad04d815ad67d0d36647eda2840616fd94a
+size 1618902

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_11_83c1cb0239fed9f79371.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:83c1cb0239fed9f79371044130e055bdabf559600bf3b6773897ae9a4d3ed8f7
+size 1436148

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_12_5b3c2426eecabf472d8d.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5b3c2426eecabf472d8df6e0d2d5bc162cd8c92f4b00df04ac20fc09b45800cd
+size 1227217

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_13_6876091a79457be67ea7.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6876091a79457be67ea70b38e2ae0ebc86379eb82c564e2265c219f6c894ef26
+size 1215670

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_14_e23bb82f59567ab38280.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e23bb82f59567ab382809cc37267b9248d3731e77f2f071bf4ac329640fb21b1
+size 1120628

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_15_08fbe0df36c11295fe0e.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:08fbe0df36c11295fe0e35f37c7eae508473abaaa9de839230eadf0da9b19364
+size 1071212

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_16_6dd373b8d21986a89617.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6dd373b8d21986a89617b4ec4b380228dd5936a3620fee9f99f8c169a4a1a5c2
+size 1144524

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_17_0b8984f5b47aeb710bec.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0b8984f5b47aeb710bec1f02c175050ac65cf24c1149dc6cce038585da3a856a
+size 1065785

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_18_37b7c06cbf94b4d1290d.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:37b7c06cbf94b4d1290d5c8333878b6a24a44f7cb5faafbc9a294d6b66a30374
+size 1107669

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_19_756caade2d1fd4a30c0a.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:756caade2d1fd4a30c0a7aeabc087b33ef89a52cba4fcb46ced71a39cc4a9984
+size 1067021

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_1_cb69ac202a830cd54f3c.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cb69ac202a830cd54f3cb6cd11867857813b4deee93e0f022d5006948a3380ee
+size 6059426

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_20_f34ba9f9b4ff9969489b.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f34ba9f9b4ff9969489b89ad1eab8ab255a4baa624c880e751379613e52d1089
+size 1040426

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_21_91d3de1a852a74091ca3.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91d3de1a852a74091ca359278c8f7140102923f88f6a772743ee7e9a250d4d08
+size 1045543

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_22_b8ec8ae339ed7f41fe8b.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b8ec8ae339ed7f41fe8b500a26f4e7e2a76ad15a79fb4f97b027da4cebf54dcb
+size 1013515

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_23_fcb034794f3447789b5c.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fcb034794f3447789b5c4bd4647e3c9c83ac6fe73c24f75e45c09ebefb3fc3a7
+size 1055829

k600_base_vq_dim_16_decay_0_99_lr_step_p_loss_1_0_dec_rope_fixed/b128_btn1024_vector_quantize_rcs8192__/wandb/run-20251026_035824-c3u3silm/files/media/videos/vis_train_dataset_24_d4dac4cd93c9f9849e0c.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d4dac4cd93c9f9849e0cb3bc3410531219f1fd8242e050b92935b380c187e07b
+size 1116529