Upload folder using huggingface_hub

decoder.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:33bc72089283767d00ffc85eac8beca3f5c4158c1049ba44f3e93a188c5de6bb
+size 278084696

decoder_conf.yml

@@ -0,0 +1,82 @@
+model:
+  model_id: dcae
+  sample_size:
+  - 360
+  - 640
+  channels: 3
+  latent_size: 32
+  latent_channels: 16
+  ch_0: 128
+  ch_max: 1024
+  encoder_blocks_per_stage:
+  - 1
+  - 1
+  - 1
+  - 1
+  decoder_blocks_per_stage:
+  - 1
+  - 1
+  - 1
+  - 1
+  use_middle_block: false
+  do_channel_mask: false
+  skip_logvar: false
+  discriminator:
+    model_id: seraena
+    channels: 3
+    latent_channels: 16
+    ch_0: 64
+    sample_size:
+    - 360
+    - 640
+train:
+  trainer_id: distill_dec_seraena
+  data_id: video_dir_loader
+  data_kwargs:
+    source:
+    - /mnt/data/datasets/extracted_tars/kbm/fps/*/*.mp4
+    - /mnt/data/datasets/extracted_tars/kbm/3ps/*/*.mp4
+    - /mnt/data/datasets/extracted_tars/kbm/other/*/*.mp4
+    target_size:
+    - 360
+    - 640
+  target_batch_size: 256
+  batch_size: 32
+  epochs: 200
+  opt: AdamW
+  opt_kwargs:
+    lr: 0.0001
+    weight_decay: 0.0001
+    betas:
+    - 0.9
+    - 0.95
+    eps: 1.0e-15
+  lpips_type: convnext
+  loss_weights:
+    lpips: 12.0
+    l2: 1.0
+    gan: 0.5
+    dwt: 0.0
+  scheduler: LinearWarmupWithCosineDecay
+  scheduler_kwargs:
+    warmup_steps: 1000
+    decay_after: 120000
+    decay_steps: 220000
+    start_mult: 0.1
+    decay_to_mult: 0.1
+  checkpoint_dir: checkpoints/waypoint_1_vae_owlc_f16_c16_dec_distill_adv_fix
+  resume_ckpt: checkpoints/waypoint_1_vae_owlc_f16_c16_dec_distill/step_5000.pt
+  encoder_cfg: configs/waypoint_1/owl_vae_f16_c16_enc_distill.yml
+  encoder_ckpt: checkpoints/waypoint_1_vae_owlc_f16_c16_enc_distill/step_110000.pt
+  teacher_ckpt: checkpoints/waypoint_1_vae_owlc_f16_c16/step_300000.pt
+  teacher_cfg: configs/waypoint_1/owl_vae_f16_c16.yml
+  sample_interval: 1000
+  save_interval: 5000
+  latent_scale: 1.0
+  latent_shift: 0.0
+  delay_adv: 5000
+  warmup_adv: 2000
+wandb:
+  name: shahbuland
+  project: new_vaes_v2
+  run_name: waypoint_1_vae_owlc_f16_c16_dec_distill

encoder.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:364680dc0902b61ff6b836a7f88cccdbfc522f13e3cbf2286b342a4aff048d8d
+size 5674296

encoder_conf.yml

@@ -0,0 +1,73 @@
+model:
+  model_id: dcae
+  sample_size:
+  - 360
+  - 640
+  channels: 3
+  latent_size: 32
+  latent_channels: 16
+  ch_0: 64
+  ch_max: 256
+  encoder_blocks_per_stage:
+  - 1
+  - 1
+  - 1
+  - 1
+  decoder_blocks_per_stage:
+  - 1
+  - 1
+  - 1
+  - 1
+  use_middle_block: false
+  do_channel_mask: false
+  skip_logvar: false
+train:
+  trainer_id: distill_enc
+  data_id: video_dir_loader
+  data_kwargs:
+    source:
+    - /mnt/data/datasets/extracted_tars/kbm/fps/*/*.mp4
+    - /mnt/data/datasets/extracted_tars/kbm/3ps/*/*.mp4
+    - /mnt/data/datasets/extracted_tars/kbm/other/*/*.mp4
+    target_size:
+    - 360
+    - 640
+  target_batch_size: 256
+  batch_size: 32
+  epochs: 200
+  opt: Muon
+  opt_kwargs:
+    lr: 0.0005
+    momentum: 0.95
+    adamw_lr: 0.0001
+    adamw_wd: 0.01
+    adamw_eps: 1.0e-06
+    adamw_betas:
+    - 0.9
+    - 0.95
+    adamw_keys:
+    - conv_in
+    - conv_out
+    - conv_out_logvar
+    - .down.
+    - .residuals.
+  lpips_type: convnext
+  loss_weights:
+    l2: 1.0
+    logvar: 1.0
+  scheduler: LinearWarmup
+  scheduler_kwargs:
+    warmup_steps: 1500
+    min_lr: 1.0e-05
+  checkpoint_dir: checkpoints/waypoint_1_vae_owlc_f16_c16_enc_distill
+  resume_ckpt: null
+  teacher_ckpt: checkpoints/waypoint_1_vae_owlc_f16_c16/step_300000.pt
+  teacher_cfg: configs/waypoint_1/owl_vae_f16_c16.yml
+  sample_interval: 1000
+  save_interval: 5000
+  latent_scale: 1.0
+  latent_shift: 0.0
+wandb:
+  name: shahbuland
+  project: new_vaes_v2
+  run_name: waypoint_1_vae_owlc_f16_c16_enc_distill_highlr