Flax classes are deprecated and will be removed in Diffusers v0.40.0. We recommend migrating to PyTorch classes or pinning your version of Diffusers.
Flax classes are deprecated and will be removed in Diffusers v0.40.0. We recommend migrating to PyTorch classes or pinning your version of Diffusers.

Fetching 17 files:   0%|          | 0/17 [00:00<?, ?it/s]
Fetching 17 files: 100%|██████████| 17/17 [00:00<00:00, 8112.77it/s]

Loading pipeline components...:   0%|          | 0/7 [00:00<?, ?it/s]
Loading pipeline components...:  29%|██▊       | 2/7 [00:00<00:00,  5.42it/s]
Loading pipeline components...:  43%|████▎     | 3/7 [00:02<00:03,  1.13it/s]
Loading pipeline components...:  71%|███████▏  | 5/7 [00:02<00:00,  2.22it/s]There are modules in UNet2DConditionModel that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.

Loading pipeline components...:  86%|████████▌ | 6/7 [00:02<00:00,  2.33it/s]There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.

Loading pipeline components...: 100%|██████████| 7/7 [00:04<00:00,  1.45it/s]
Loading pipeline components...: 100%|██████████| 7/7 [00:04<00:00,  1.70it/s]
/usr/local/lib/python3.12/dist-packages/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl.py:748: FutureWarning: `upcast_vae` is deprecated and will be removed in version 1.0.0. `upcast_vae` is deprecated. Please use `pipe.vae.to(torch.float32)`. For more details, please refer to: https://github.com/huggingface/diffusers/pull/12619#issue-3606633695.
  deprecate(
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
  dense case 0: 1.977s
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
  dense case 1: 1.923s
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
  dense case 2: 1.922s
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
  dense case 3: 1.913s

Fetching 17 files:   0%|          | 0/17 [00:00<?, ?it/s]
Fetching 17 files: 100%|██████████| 17/17 [00:00<00:00, 23339.83it/s]

Loading pipeline components...:   0%|          | 0/7 [00:00<?, ?it/s]
Loading pipeline components...:  29%|██▊       | 2/7 [00:00<00:00, 19.92it/s]
Loading pipeline components...:  57%|█████▋    | 4/7 [00:00<00:00,  7.16it/s]There are modules in UNet2DConditionModel that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.

Loading pipeline components...:  86%|████████▌ | 6/7 [00:00<00:00,  5.55it/s]There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.

Loading pipeline components...: 100%|██████████| 7/7 [00:01<00:00,  6.63it/s]

========================================================================================
structured_20 PACKAGE 1/7 targets=36
========================================================================================
  down_blocks.1.attentions.0.transformer_blocks.0.ff: H=2560 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.0.transformer_blocks.0.attn1: H=640 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.0.transformer_blocks.0.attn2: H=640 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.0.transformer_blocks.1.ff: H=2560 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.0.transformer_blocks.1.attn1: H=640 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.0.transformer_blocks.1.attn2: H=640 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.1.transformer_blocks.0.ff: H=2560 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.1.transformer_blocks.0.attn1: H=640 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.1.transformer_blocks.0.attn2: H=640 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.1.transformer_blocks.1.ff: H=2560 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.1.transformer_blocks.1.attn1: H=640 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.1.transformer_blocks.1.attn2: H=640 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.0.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.0.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.0.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.1.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.1.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.1.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.2.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.2.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.2.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.3.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.3.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.3.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.4.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.4.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.4.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.5.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.5.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.5.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.6.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.6.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.6.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.7.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.7.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.7.attn2: H=1280 calls=64 sampled_tokens=8192
  prune down_blocks.1.attentions.0.transformer_blocks.0.ff: kind=ff ratio=20%
  prune down_blocks.1.attentions.0.transformer_blocks.0.attn1: kind=attention ratio=20%
  prune down_blocks.1.attentions.0.transformer_blocks.0.attn2: kind=attention ratio=20%
  prune down_blocks.1.attentions.0.transformer_blocks.1.ff: kind=ff ratio=20%
  prune down_blocks.1.attentions.0.transformer_blocks.1.attn1: kind=attention ratio=20%
  prune down_blocks.1.attentions.0.transformer_blocks.1.attn2: kind=attention ratio=20%
  prune down_blocks.1.attentions.1.transformer_blocks.0.ff: kind=ff ratio=20%
  prune down_blocks.1.attentions.1.transformer_blocks.0.attn1: kind=attention ratio=20%
  prune down_blocks.1.attentions.1.transformer_blocks.0.attn2: kind=attention ratio=20%
  prune down_blocks.1.attentions.1.transformer_blocks.1.ff: kind=ff ratio=20%
  prune down_blocks.1.attentions.1.transformer_blocks.1.attn1: kind=attention ratio=20%
  prune down_blocks.1.attentions.1.transformer_blocks.1.attn2: kind=attention ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.0.ff: kind=ff ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.0.attn1: kind=attention ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.0.attn2: kind=attention ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.1.ff: kind=ff ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.1.attn1: kind=attention ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.1.attn2: kind=attention ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.2.ff: kind=ff ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.2.attn1: kind=attention ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.2.attn2: kind=attention ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.3.ff: kind=ff ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.3.attn1: kind=attention ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.3.attn2: kind=attention ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.4.ff: kind=ff ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.4.attn1: kind=attention ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.4.attn2: kind=attention ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.5.ff: kind=ff ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.5.attn1: kind=attention ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.5.attn2: kind=attention ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.6.ff: kind=ff ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.6.attn1: kind=attention ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.6.attn2: kind=attention ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.7.ff: kind=ff ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.7.attn1: kind=attention ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.7.attn2: kind=attention ratio=20%

========================================================================================
structured_20 PACKAGE 2/7 targets=27
========================================================================================
  down_blocks.2.attentions.0.transformer_blocks.8.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.8.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.8.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.9.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.9.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.9.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.0.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.0.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.0.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.1.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.1.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.1.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.2.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.2.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.2.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.3.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.3.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.3.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.4.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.4.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.4.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.5.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.5.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.5.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.6.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.6.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.6.attn2: H=1280 calls=64 sampled_tokens=8192
  prune down_blocks.2.attentions.0.transformer_blocks.8.ff: kind=ff ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.8.attn1: kind=attention ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.8.attn2: kind=attention ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.9.ff: kind=ff ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.9.attn1: kind=attention ratio=20%
  prune down_blocks.2.attentions.0.transformer_blocks.9.attn2: kind=attention ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.0.ff: kind=ff ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.0.attn1: kind=attention ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.0.attn2: kind=attention ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.1.ff: kind=ff ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.1.attn1: kind=attention ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.1.attn2: kind=attention ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.2.ff: kind=ff ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.2.attn1: kind=attention ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.2.attn2: kind=attention ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.3.ff: kind=ff ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.3.attn1: kind=attention ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.3.attn2: kind=attention ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.4.ff: kind=ff ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.4.attn1: kind=attention ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.4.attn2: kind=attention ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.5.ff: kind=ff ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.5.attn1: kind=attention ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.5.attn2: kind=attention ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.6.ff: kind=ff ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.6.attn1: kind=attention ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.6.attn2: kind=attention ratio=20%

========================================================================================
structured_20 PACKAGE 3/7 targets=27
========================================================================================
  down_blocks.2.attentions.1.transformer_blocks.7.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.7.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.7.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.8.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.8.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.8.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.9.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.9.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.9.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.0.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.0.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.0.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.1.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.1.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.1.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.2.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.2.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.2.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.3.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.3.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.3.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.4.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.4.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.4.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.5.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.5.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.5.attn2: H=1280 calls=64 sampled_tokens=8192
  prune down_blocks.2.attentions.1.transformer_blocks.7.ff: kind=ff ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.7.attn1: kind=attention ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.7.attn2: kind=attention ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.8.ff: kind=ff ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.8.attn1: kind=attention ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.8.attn2: kind=attention ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.9.ff: kind=ff ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.9.attn1: kind=attention ratio=20%
  prune down_blocks.2.attentions.1.transformer_blocks.9.attn2: kind=attention ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.0.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.0.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.0.attn2: kind=attention ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.1.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.1.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.1.attn2: kind=attention ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.2.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.2.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.2.attn2: kind=attention ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.3.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.3.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.3.attn2: kind=attention ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.4.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.4.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.4.attn2: kind=attention ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.5.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.5.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.5.attn2: kind=attention ratio=20%

========================================================================================
structured_20 PACKAGE 4/7 targets=27
========================================================================================
  up_blocks.0.attentions.0.transformer_blocks.6.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.6.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.6.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.7.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.7.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.7.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.8.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.8.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.8.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.9.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.9.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.9.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.0.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.0.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.0.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.1.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.1.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.1.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.2.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.2.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.2.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.3.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.3.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.3.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.4.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.4.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.4.attn2: H=1280 calls=64 sampled_tokens=8192
  prune up_blocks.0.attentions.0.transformer_blocks.6.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.6.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.6.attn2: kind=attention ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.7.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.7.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.7.attn2: kind=attention ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.8.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.8.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.8.attn2: kind=attention ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.9.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.9.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.0.transformer_blocks.9.attn2: kind=attention ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.0.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.0.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.0.attn2: kind=attention ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.1.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.1.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.1.attn2: kind=attention ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.2.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.2.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.2.attn2: kind=attention ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.3.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.3.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.3.attn2: kind=attention ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.4.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.4.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.4.attn2: kind=attention ratio=20%

========================================================================================
structured_20 PACKAGE 5/7 targets=27
========================================================================================
  up_blocks.0.attentions.1.transformer_blocks.5.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.5.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.5.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.6.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.6.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.6.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.7.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.7.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.7.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.8.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.8.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.8.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.9.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.9.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.9.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.0.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.0.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.0.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.1.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.1.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.1.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.2.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.2.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.2.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.3.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.3.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.3.attn2: H=1280 calls=64 sampled_tokens=8192
  prune up_blocks.0.attentions.1.transformer_blocks.5.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.5.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.5.attn2: kind=attention ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.6.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.6.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.6.attn2: kind=attention ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.7.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.7.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.7.attn2: kind=attention ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.8.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.8.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.8.attn2: kind=attention ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.9.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.9.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.1.transformer_blocks.9.attn2: kind=attention ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.0.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.0.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.0.attn2: kind=attention ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.1.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.1.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.1.attn2: kind=attention ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.2.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.2.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.2.attn2: kind=attention ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.3.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.3.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.3.attn2: kind=attention ratio=20%

========================================================================================
structured_20 PACKAGE 6/7 targets=39
========================================================================================
  up_blocks.0.attentions.2.transformer_blocks.4.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.4.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.4.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.5.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.5.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.5.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.6.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.6.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.6.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.7.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.7.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.7.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.8.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.8.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.8.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.9.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.9.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.9.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.0.transformer_blocks.0.ff: H=2560 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.0.transformer_blocks.0.attn1: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.0.transformer_blocks.0.attn2: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.0.transformer_blocks.1.ff: H=2560 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.0.transformer_blocks.1.attn1: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.0.transformer_blocks.1.attn2: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.1.transformer_blocks.0.ff: H=2560 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.1.transformer_blocks.0.attn1: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.1.transformer_blocks.0.attn2: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.1.transformer_blocks.1.ff: H=2560 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.1.transformer_blocks.1.attn1: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.1.transformer_blocks.1.attn2: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.2.transformer_blocks.0.ff: H=2560 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.2.transformer_blocks.0.attn1: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.2.transformer_blocks.0.attn2: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.2.transformer_blocks.1.ff: H=2560 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.2.transformer_blocks.1.attn1: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.2.transformer_blocks.1.attn2: H=640 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.0.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.0.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.0.attn2: H=1280 calls=64 sampled_tokens=8192
  prune up_blocks.0.attentions.2.transformer_blocks.4.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.4.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.4.attn2: kind=attention ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.5.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.5.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.5.attn2: kind=attention ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.6.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.6.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.6.attn2: kind=attention ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.7.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.7.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.7.attn2: kind=attention ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.8.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.8.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.8.attn2: kind=attention ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.9.ff: kind=ff ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.9.attn1: kind=attention ratio=20%
  prune up_blocks.0.attentions.2.transformer_blocks.9.attn2: kind=attention ratio=20%
  prune up_blocks.1.attentions.0.transformer_blocks.0.ff: kind=ff ratio=20%
  prune up_blocks.1.attentions.0.transformer_blocks.0.attn1: kind=attention ratio=20%
  prune up_blocks.1.attentions.0.transformer_blocks.0.attn2: kind=attention ratio=20%
  prune up_blocks.1.attentions.0.transformer_blocks.1.ff: kind=ff ratio=20%
  prune up_blocks.1.attentions.0.transformer_blocks.1.attn1: kind=attention ratio=20%
  prune up_blocks.1.attentions.0.transformer_blocks.1.attn2: kind=attention ratio=20%
  prune up_blocks.1.attentions.1.transformer_blocks.0.ff: kind=ff ratio=20%
  prune up_blocks.1.attentions.1.transformer_blocks.0.attn1: kind=attention ratio=20%
  prune up_blocks.1.attentions.1.transformer_blocks.0.attn2: kind=attention ratio=20%
  prune up_blocks.1.attentions.1.transformer_blocks.1.ff: kind=ff ratio=20%
  prune up_blocks.1.attentions.1.transformer_blocks.1.attn1: kind=attention ratio=20%
  prune up_blocks.1.attentions.1.transformer_blocks.1.attn2: kind=attention ratio=20%
  prune up_blocks.1.attentions.2.transformer_blocks.0.ff: kind=ff ratio=20%
  prune up_blocks.1.attentions.2.transformer_blocks.0.attn1: kind=attention ratio=20%
  prune up_blocks.1.attentions.2.transformer_blocks.0.attn2: kind=attention ratio=20%
  prune up_blocks.1.attentions.2.transformer_blocks.1.ff: kind=ff ratio=20%
  prune up_blocks.1.attentions.2.transformer_blocks.1.attn1: kind=attention ratio=20%
  prune up_blocks.1.attentions.2.transformer_blocks.1.attn2: kind=attention ratio=20%
  prune mid_block.attentions.0.transformer_blocks.0.ff: kind=ff ratio=20%
  prune mid_block.attentions.0.transformer_blocks.0.attn1: kind=attention ratio=20%
  prune mid_block.attentions.0.transformer_blocks.0.attn2: kind=attention ratio=20%

========================================================================================
structured_20 PACKAGE 7/7 targets=27
========================================================================================
  mid_block.attentions.0.transformer_blocks.1.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.1.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.1.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.2.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.2.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.2.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.3.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.3.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.3.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.4.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.4.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.4.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.5.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.5.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.5.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.6.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.6.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.6.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.7.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.7.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.7.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.8.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.8.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.8.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.9.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.9.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.9.attn2: H=1280 calls=64 sampled_tokens=8192
  prune mid_block.attentions.0.transformer_blocks.1.ff: kind=ff ratio=20%
  prune mid_block.attentions.0.transformer_blocks.1.attn1: kind=attention ratio=20%
  prune mid_block.attentions.0.transformer_blocks.1.attn2: kind=attention ratio=20%
  prune mid_block.attentions.0.transformer_blocks.2.ff: kind=ff ratio=20%
  prune mid_block.attentions.0.transformer_blocks.2.attn1: kind=attention ratio=20%
  prune mid_block.attentions.0.transformer_blocks.2.attn2: kind=attention ratio=20%
  prune mid_block.attentions.0.transformer_blocks.3.ff: kind=ff ratio=20%
  prune mid_block.attentions.0.transformer_blocks.3.attn1: kind=attention ratio=20%
  prune mid_block.attentions.0.transformer_blocks.3.attn2: kind=attention ratio=20%
  prune mid_block.attentions.0.transformer_blocks.4.ff: kind=ff ratio=20%
  prune mid_block.attentions.0.transformer_blocks.4.attn1: kind=attention ratio=20%
  prune mid_block.attentions.0.transformer_blocks.4.attn2: kind=attention ratio=20%
  prune mid_block.attentions.0.transformer_blocks.5.ff: kind=ff ratio=20%
  prune mid_block.attentions.0.transformer_blocks.5.attn1: kind=attention ratio=20%
  prune mid_block.attentions.0.transformer_blocks.5.attn2: kind=attention ratio=20%
  prune mid_block.attentions.0.transformer_blocks.6.ff: kind=ff ratio=20%
  prune mid_block.attentions.0.transformer_blocks.6.attn1: kind=attention ratio=20%
  prune mid_block.attentions.0.transformer_blocks.6.attn2: kind=attention ratio=20%
  prune mid_block.attentions.0.transformer_blocks.7.ff: kind=ff ratio=20%
  prune mid_block.attentions.0.transformer_blocks.7.attn1: kind=attention ratio=20%
  prune mid_block.attentions.0.transformer_blocks.7.attn2: kind=attention ratio=20%
  prune mid_block.attentions.0.transformer_blocks.8.ff: kind=ff ratio=20%
  prune mid_block.attentions.0.transformer_blocks.8.attn1: kind=attention ratio=20%
  prune mid_block.attentions.0.transformer_blocks.8.attn2: kind=attention ratio=20%
  prune mid_block.attentions.0.transformer_blocks.9.ff: kind=ff ratio=20%
  prune mid_block.attentions.0.transformer_blocks.9.attn1: kind=attention ratio=20%
  prune mid_block.attentions.0.transformer_blocks.9.attn2: kind=attention ratio=20%
structured_20: 2,130,533,124 params (2.1305B), physical reduction 17.0180%
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
  structured_20 case 0: 1.902s
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
  structured_20 case 1: 1.872s
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
  structured_20 case 2: 1.871s
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
  structured_20 case 3: 1.856s
There are modules in UNet2DConditionModel that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
  standalone torch.load verification: PASS

Fetching 17 files:   0%|          | 0/17 [00:00<?, ?it/s]
Fetching 17 files: 100%|██████████| 17/17 [00:00<00:00, 22773.29it/s]

Loading pipeline components...:   0%|          | 0/7 [00:00<?, ?it/s]
Loading pipeline components...:  29%|██▊       | 2/7 [00:00<00:00, 19.89it/s]
Loading pipeline components...:  57%|█████▋    | 4/7 [00:00<00:00,  7.75it/s]There are modules in UNet2DConditionModel that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.

Loading pipeline components...:  86%|████████▌ | 6/7 [00:00<00:00,  5.79it/s]There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.

Loading pipeline components...: 100%|██████████| 7/7 [00:01<00:00,  7.00it/s]

========================================================================================
structured_30 PACKAGE 1/7 targets=36
========================================================================================
  down_blocks.1.attentions.0.transformer_blocks.0.ff: H=2560 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.0.transformer_blocks.0.attn1: H=640 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.0.transformer_blocks.0.attn2: H=640 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.0.transformer_blocks.1.ff: H=2560 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.0.transformer_blocks.1.attn1: H=640 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.0.transformer_blocks.1.attn2: H=640 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.1.transformer_blocks.0.ff: H=2560 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.1.transformer_blocks.0.attn1: H=640 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.1.transformer_blocks.0.attn2: H=640 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.1.transformer_blocks.1.ff: H=2560 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.1.transformer_blocks.1.attn1: H=640 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.1.transformer_blocks.1.attn2: H=640 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.0.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.0.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.0.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.1.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.1.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.1.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.2.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.2.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.2.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.3.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.3.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.3.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.4.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.4.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.4.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.5.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.5.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.5.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.6.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.6.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.6.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.7.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.7.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.7.attn2: H=1280 calls=64 sampled_tokens=8192
  prune down_blocks.1.attentions.0.transformer_blocks.0.ff: kind=ff ratio=30%
  prune down_blocks.1.attentions.0.transformer_blocks.0.attn1: kind=attention ratio=30%
  prune down_blocks.1.attentions.0.transformer_blocks.0.attn2: kind=attention ratio=30%
  prune down_blocks.1.attentions.0.transformer_blocks.1.ff: kind=ff ratio=30%
  prune down_blocks.1.attentions.0.transformer_blocks.1.attn1: kind=attention ratio=30%
  prune down_blocks.1.attentions.0.transformer_blocks.1.attn2: kind=attention ratio=30%
  prune down_blocks.1.attentions.1.transformer_blocks.0.ff: kind=ff ratio=30%
  prune down_blocks.1.attentions.1.transformer_blocks.0.attn1: kind=attention ratio=30%
  prune down_blocks.1.attentions.1.transformer_blocks.0.attn2: kind=attention ratio=30%
  prune down_blocks.1.attentions.1.transformer_blocks.1.ff: kind=ff ratio=30%
  prune down_blocks.1.attentions.1.transformer_blocks.1.attn1: kind=attention ratio=30%
  prune down_blocks.1.attentions.1.transformer_blocks.1.attn2: kind=attention ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.0.ff: kind=ff ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.0.attn1: kind=attention ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.0.attn2: kind=attention ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.1.ff: kind=ff ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.1.attn1: kind=attention ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.1.attn2: kind=attention ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.2.ff: kind=ff ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.2.attn1: kind=attention ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.2.attn2: kind=attention ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.3.ff: kind=ff ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.3.attn1: kind=attention ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.3.attn2: kind=attention ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.4.ff: kind=ff ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.4.attn1: kind=attention ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.4.attn2: kind=attention ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.5.ff: kind=ff ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.5.attn1: kind=attention ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.5.attn2: kind=attention ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.6.ff: kind=ff ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.6.attn1: kind=attention ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.6.attn2: kind=attention ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.7.ff: kind=ff ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.7.attn1: kind=attention ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.7.attn2: kind=attention ratio=30%

========================================================================================
structured_30 PACKAGE 2/7 targets=27
========================================================================================
  down_blocks.2.attentions.0.transformer_blocks.8.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.8.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.8.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.9.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.9.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.9.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.0.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.0.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.0.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.1.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.1.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.1.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.2.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.2.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.2.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.3.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.3.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.3.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.4.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.4.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.4.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.5.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.5.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.5.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.6.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.6.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.6.attn2: H=1280 calls=64 sampled_tokens=8192
  prune down_blocks.2.attentions.0.transformer_blocks.8.ff: kind=ff ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.8.attn1: kind=attention ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.8.attn2: kind=attention ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.9.ff: kind=ff ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.9.attn1: kind=attention ratio=30%
  prune down_blocks.2.attentions.0.transformer_blocks.9.attn2: kind=attention ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.0.ff: kind=ff ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.0.attn1: kind=attention ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.0.attn2: kind=attention ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.1.ff: kind=ff ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.1.attn1: kind=attention ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.1.attn2: kind=attention ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.2.ff: kind=ff ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.2.attn1: kind=attention ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.2.attn2: kind=attention ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.3.ff: kind=ff ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.3.attn1: kind=attention ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.3.attn2: kind=attention ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.4.ff: kind=ff ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.4.attn1: kind=attention ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.4.attn2: kind=attention ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.5.ff: kind=ff ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.5.attn1: kind=attention ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.5.attn2: kind=attention ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.6.ff: kind=ff ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.6.attn1: kind=attention ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.6.attn2: kind=attention ratio=30%

========================================================================================
structured_30 PACKAGE 3/7 targets=27
========================================================================================
  down_blocks.2.attentions.1.transformer_blocks.7.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.7.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.7.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.8.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.8.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.8.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.9.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.9.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.9.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.0.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.0.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.0.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.1.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.1.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.1.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.2.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.2.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.2.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.3.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.3.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.3.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.4.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.4.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.4.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.5.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.5.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.5.attn2: H=1280 calls=64 sampled_tokens=8192
  prune down_blocks.2.attentions.1.transformer_blocks.7.ff: kind=ff ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.7.attn1: kind=attention ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.7.attn2: kind=attention ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.8.ff: kind=ff ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.8.attn1: kind=attention ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.8.attn2: kind=attention ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.9.ff: kind=ff ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.9.attn1: kind=attention ratio=30%
  prune down_blocks.2.attentions.1.transformer_blocks.9.attn2: kind=attention ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.0.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.0.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.0.attn2: kind=attention ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.1.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.1.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.1.attn2: kind=attention ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.2.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.2.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.2.attn2: kind=attention ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.3.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.3.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.3.attn2: kind=attention ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.4.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.4.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.4.attn2: kind=attention ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.5.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.5.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.5.attn2: kind=attention ratio=30%

========================================================================================
structured_30 PACKAGE 4/7 targets=27
========================================================================================
  up_blocks.0.attentions.0.transformer_blocks.6.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.6.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.6.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.7.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.7.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.7.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.8.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.8.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.8.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.9.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.9.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.9.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.0.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.0.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.0.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.1.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.1.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.1.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.2.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.2.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.2.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.3.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.3.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.3.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.4.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.4.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.4.attn2: H=1280 calls=64 sampled_tokens=8192
  prune up_blocks.0.attentions.0.transformer_blocks.6.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.6.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.6.attn2: kind=attention ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.7.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.7.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.7.attn2: kind=attention ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.8.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.8.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.8.attn2: kind=attention ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.9.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.9.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.0.transformer_blocks.9.attn2: kind=attention ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.0.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.0.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.0.attn2: kind=attention ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.1.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.1.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.1.attn2: kind=attention ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.2.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.2.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.2.attn2: kind=attention ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.3.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.3.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.3.attn2: kind=attention ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.4.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.4.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.4.attn2: kind=attention ratio=30%

========================================================================================
structured_30 PACKAGE 5/7 targets=27
========================================================================================
  up_blocks.0.attentions.1.transformer_blocks.5.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.5.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.5.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.6.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.6.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.6.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.7.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.7.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.7.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.8.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.8.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.8.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.9.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.9.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.9.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.0.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.0.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.0.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.1.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.1.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.1.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.2.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.2.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.2.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.3.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.3.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.3.attn2: H=1280 calls=64 sampled_tokens=8192
  prune up_blocks.0.attentions.1.transformer_blocks.5.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.5.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.5.attn2: kind=attention ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.6.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.6.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.6.attn2: kind=attention ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.7.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.7.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.7.attn2: kind=attention ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.8.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.8.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.8.attn2: kind=attention ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.9.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.9.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.1.transformer_blocks.9.attn2: kind=attention ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.0.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.0.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.0.attn2: kind=attention ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.1.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.1.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.1.attn2: kind=attention ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.2.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.2.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.2.attn2: kind=attention ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.3.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.3.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.3.attn2: kind=attention ratio=30%

========================================================================================
structured_30 PACKAGE 6/7 targets=39
========================================================================================
  up_blocks.0.attentions.2.transformer_blocks.4.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.4.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.4.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.5.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.5.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.5.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.6.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.6.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.6.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.7.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.7.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.7.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.8.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.8.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.8.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.9.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.9.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.9.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.0.transformer_blocks.0.ff: H=2560 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.0.transformer_blocks.0.attn1: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.0.transformer_blocks.0.attn2: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.0.transformer_blocks.1.ff: H=2560 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.0.transformer_blocks.1.attn1: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.0.transformer_blocks.1.attn2: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.1.transformer_blocks.0.ff: H=2560 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.1.transformer_blocks.0.attn1: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.1.transformer_blocks.0.attn2: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.1.transformer_blocks.1.ff: H=2560 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.1.transformer_blocks.1.attn1: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.1.transformer_blocks.1.attn2: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.2.transformer_blocks.0.ff: H=2560 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.2.transformer_blocks.0.attn1: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.2.transformer_blocks.0.attn2: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.2.transformer_blocks.1.ff: H=2560 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.2.transformer_blocks.1.attn1: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.2.transformer_blocks.1.attn2: H=640 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.0.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.0.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.0.attn2: H=1280 calls=64 sampled_tokens=8192
  prune up_blocks.0.attentions.2.transformer_blocks.4.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.4.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.4.attn2: kind=attention ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.5.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.5.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.5.attn2: kind=attention ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.6.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.6.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.6.attn2: kind=attention ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.7.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.7.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.7.attn2: kind=attention ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.8.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.8.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.8.attn2: kind=attention ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.9.ff: kind=ff ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.9.attn1: kind=attention ratio=30%
  prune up_blocks.0.attentions.2.transformer_blocks.9.attn2: kind=attention ratio=30%
  prune up_blocks.1.attentions.0.transformer_blocks.0.ff: kind=ff ratio=30%
  prune up_blocks.1.attentions.0.transformer_blocks.0.attn1: kind=attention ratio=30%
  prune up_blocks.1.attentions.0.transformer_blocks.0.attn2: kind=attention ratio=30%
  prune up_blocks.1.attentions.0.transformer_blocks.1.ff: kind=ff ratio=30%
  prune up_blocks.1.attentions.0.transformer_blocks.1.attn1: kind=attention ratio=30%
  prune up_blocks.1.attentions.0.transformer_blocks.1.attn2: kind=attention ratio=30%
  prune up_blocks.1.attentions.1.transformer_blocks.0.ff: kind=ff ratio=30%
  prune up_blocks.1.attentions.1.transformer_blocks.0.attn1: kind=attention ratio=30%
  prune up_blocks.1.attentions.1.transformer_blocks.0.attn2: kind=attention ratio=30%
  prune up_blocks.1.attentions.1.transformer_blocks.1.ff: kind=ff ratio=30%
  prune up_blocks.1.attentions.1.transformer_blocks.1.attn1: kind=attention ratio=30%
  prune up_blocks.1.attentions.1.transformer_blocks.1.attn2: kind=attention ratio=30%
  prune up_blocks.1.attentions.2.transformer_blocks.0.ff: kind=ff ratio=30%
  prune up_blocks.1.attentions.2.transformer_blocks.0.attn1: kind=attention ratio=30%
  prune up_blocks.1.attentions.2.transformer_blocks.0.attn2: kind=attention ratio=30%
  prune up_blocks.1.attentions.2.transformer_blocks.1.ff: kind=ff ratio=30%
  prune up_blocks.1.attentions.2.transformer_blocks.1.attn1: kind=attention ratio=30%
  prune up_blocks.1.attentions.2.transformer_blocks.1.attn2: kind=attention ratio=30%
  prune mid_block.attentions.0.transformer_blocks.0.ff: kind=ff ratio=30%
  prune mid_block.attentions.0.transformer_blocks.0.attn1: kind=attention ratio=30%
  prune mid_block.attentions.0.transformer_blocks.0.attn2: kind=attention ratio=30%

========================================================================================
structured_30 PACKAGE 7/7 targets=27
========================================================================================
  mid_block.attentions.0.transformer_blocks.1.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.1.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.1.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.2.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.2.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.2.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.3.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.3.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.3.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.4.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.4.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.4.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.5.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.5.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.5.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.6.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.6.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.6.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.7.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.7.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.7.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.8.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.8.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.8.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.9.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.9.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.9.attn2: H=1280 calls=64 sampled_tokens=8192
  prune mid_block.attentions.0.transformer_blocks.1.ff: kind=ff ratio=30%
  prune mid_block.attentions.0.transformer_blocks.1.attn1: kind=attention ratio=30%
  prune mid_block.attentions.0.transformer_blocks.1.attn2: kind=attention ratio=30%
  prune mid_block.attentions.0.transformer_blocks.2.ff: kind=ff ratio=30%
  prune mid_block.attentions.0.transformer_blocks.2.attn1: kind=attention ratio=30%
  prune mid_block.attentions.0.transformer_blocks.2.attn2: kind=attention ratio=30%
  prune mid_block.attentions.0.transformer_blocks.3.ff: kind=ff ratio=30%
  prune mid_block.attentions.0.transformer_blocks.3.attn1: kind=attention ratio=30%
  prune mid_block.attentions.0.transformer_blocks.3.attn2: kind=attention ratio=30%
  prune mid_block.attentions.0.transformer_blocks.4.ff: kind=ff ratio=30%
  prune mid_block.attentions.0.transformer_blocks.4.attn1: kind=attention ratio=30%
  prune mid_block.attentions.0.transformer_blocks.4.attn2: kind=attention ratio=30%
  prune mid_block.attentions.0.transformer_blocks.5.ff: kind=ff ratio=30%
  prune mid_block.attentions.0.transformer_blocks.5.attn1: kind=attention ratio=30%
  prune mid_block.attentions.0.transformer_blocks.5.attn2: kind=attention ratio=30%
  prune mid_block.attentions.0.transformer_blocks.6.ff: kind=ff ratio=30%
  prune mid_block.attentions.0.transformer_blocks.6.attn1: kind=attention ratio=30%
  prune mid_block.attentions.0.transformer_blocks.6.attn2: kind=attention ratio=30%
  prune mid_block.attentions.0.transformer_blocks.7.ff: kind=ff ratio=30%
  prune mid_block.attentions.0.transformer_blocks.7.attn1: kind=attention ratio=30%
  prune mid_block.attentions.0.transformer_blocks.7.attn2: kind=attention ratio=30%
  prune mid_block.attentions.0.transformer_blocks.8.ff: kind=ff ratio=30%
  prune mid_block.attentions.0.transformer_blocks.8.attn1: kind=attention ratio=30%
  prune mid_block.attentions.0.transformer_blocks.8.attn2: kind=attention ratio=30%
  prune mid_block.attentions.0.transformer_blocks.9.ff: kind=ff ratio=30%
  prune mid_block.attentions.0.transformer_blocks.9.attn1: kind=attention ratio=30%
  prune mid_block.attentions.0.transformer_blocks.9.attn2: kind=attention ratio=30%
structured_30: 1,912,067,844 params (1.9121B), physical reduction 25.5270%
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
  structured_30 case 0: 1.808s
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
  structured_30 case 1: 1.773s
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
  structured_30 case 2: 1.779s
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
  structured_30 case 3: 1.778s
There are modules in UNet2DConditionModel that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
  standalone torch.load verification: PASS

Fetching 17 files:   0%|          | 0/17 [00:00<?, ?it/s]
Fetching 17 files: 100%|██████████| 17/17 [00:00<00:00, 19911.52it/s]

Loading pipeline components...:   0%|          | 0/7 [00:00<?, ?it/s]
Loading pipeline components...:  29%|██▊       | 2/7 [00:00<00:00, 19.92it/s]
Loading pipeline components...:  57%|█████▋    | 4/7 [00:00<00:00,  7.30it/s]There are modules in UNet2DConditionModel that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.

Loading pipeline components...:  86%|████████▌ | 6/7 [00:00<00:00,  5.54it/s]There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.

Loading pipeline components...: 100%|██████████| 7/7 [00:01<00:00,  6.75it/s]

========================================================================================
structured_40 PACKAGE 1/7 targets=36
========================================================================================
  down_blocks.1.attentions.0.transformer_blocks.0.ff: H=2560 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.0.transformer_blocks.0.attn1: H=640 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.0.transformer_blocks.0.attn2: H=640 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.0.transformer_blocks.1.ff: H=2560 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.0.transformer_blocks.1.attn1: H=640 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.0.transformer_blocks.1.attn2: H=640 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.1.transformer_blocks.0.ff: H=2560 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.1.transformer_blocks.0.attn1: H=640 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.1.transformer_blocks.0.attn2: H=640 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.1.transformer_blocks.1.ff: H=2560 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.1.transformer_blocks.1.attn1: H=640 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.1.transformer_blocks.1.attn2: H=640 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.0.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.0.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.0.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.1.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.1.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.1.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.2.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.2.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.2.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.3.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.3.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.3.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.4.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.4.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.4.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.5.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.5.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.5.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.6.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.6.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.6.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.7.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.7.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.7.attn2: H=1280 calls=64 sampled_tokens=8192
  prune down_blocks.1.attentions.0.transformer_blocks.0.ff: kind=ff ratio=40%
  prune down_blocks.1.attentions.0.transformer_blocks.0.attn1: kind=attention ratio=40%
  prune down_blocks.1.attentions.0.transformer_blocks.0.attn2: kind=attention ratio=40%
  prune down_blocks.1.attentions.0.transformer_blocks.1.ff: kind=ff ratio=40%
  prune down_blocks.1.attentions.0.transformer_blocks.1.attn1: kind=attention ratio=40%
  prune down_blocks.1.attentions.0.transformer_blocks.1.attn2: kind=attention ratio=40%
  prune down_blocks.1.attentions.1.transformer_blocks.0.ff: kind=ff ratio=40%
  prune down_blocks.1.attentions.1.transformer_blocks.0.attn1: kind=attention ratio=40%
  prune down_blocks.1.attentions.1.transformer_blocks.0.attn2: kind=attention ratio=40%
  prune down_blocks.1.attentions.1.transformer_blocks.1.ff: kind=ff ratio=40%
  prune down_blocks.1.attentions.1.transformer_blocks.1.attn1: kind=attention ratio=40%
  prune down_blocks.1.attentions.1.transformer_blocks.1.attn2: kind=attention ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.0.ff: kind=ff ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.0.attn1: kind=attention ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.0.attn2: kind=attention ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.1.ff: kind=ff ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.1.attn1: kind=attention ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.1.attn2: kind=attention ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.2.ff: kind=ff ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.2.attn1: kind=attention ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.2.attn2: kind=attention ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.3.ff: kind=ff ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.3.attn1: kind=attention ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.3.attn2: kind=attention ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.4.ff: kind=ff ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.4.attn1: kind=attention ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.4.attn2: kind=attention ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.5.ff: kind=ff ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.5.attn1: kind=attention ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.5.attn2: kind=attention ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.6.ff: kind=ff ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.6.attn1: kind=attention ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.6.attn2: kind=attention ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.7.ff: kind=ff ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.7.attn1: kind=attention ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.7.attn2: kind=attention ratio=40%

========================================================================================
structured_40 PACKAGE 2/7 targets=27
========================================================================================
  down_blocks.2.attentions.0.transformer_blocks.8.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.8.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.8.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.9.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.9.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.9.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.0.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.0.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.0.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.1.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.1.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.1.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.2.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.2.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.2.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.3.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.3.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.3.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.4.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.4.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.4.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.5.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.5.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.5.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.6.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.6.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.6.attn2: H=1280 calls=64 sampled_tokens=8192
  prune down_blocks.2.attentions.0.transformer_blocks.8.ff: kind=ff ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.8.attn1: kind=attention ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.8.attn2: kind=attention ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.9.ff: kind=ff ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.9.attn1: kind=attention ratio=40%
  prune down_blocks.2.attentions.0.transformer_blocks.9.attn2: kind=attention ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.0.ff: kind=ff ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.0.attn1: kind=attention ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.0.attn2: kind=attention ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.1.ff: kind=ff ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.1.attn1: kind=attention ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.1.attn2: kind=attention ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.2.ff: kind=ff ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.2.attn1: kind=attention ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.2.attn2: kind=attention ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.3.ff: kind=ff ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.3.attn1: kind=attention ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.3.attn2: kind=attention ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.4.ff: kind=ff ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.4.attn1: kind=attention ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.4.attn2: kind=attention ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.5.ff: kind=ff ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.5.attn1: kind=attention ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.5.attn2: kind=attention ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.6.ff: kind=ff ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.6.attn1: kind=attention ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.6.attn2: kind=attention ratio=40%

========================================================================================
structured_40 PACKAGE 3/7 targets=27
========================================================================================
  down_blocks.2.attentions.1.transformer_blocks.7.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.7.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.7.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.8.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.8.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.8.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.9.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.9.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.9.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.0.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.0.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.0.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.1.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.1.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.1.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.2.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.2.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.2.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.3.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.3.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.3.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.4.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.4.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.4.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.5.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.5.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.5.attn2: H=1280 calls=64 sampled_tokens=8192
  prune down_blocks.2.attentions.1.transformer_blocks.7.ff: kind=ff ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.7.attn1: kind=attention ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.7.attn2: kind=attention ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.8.ff: kind=ff ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.8.attn1: kind=attention ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.8.attn2: kind=attention ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.9.ff: kind=ff ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.9.attn1: kind=attention ratio=40%
  prune down_blocks.2.attentions.1.transformer_blocks.9.attn2: kind=attention ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.0.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.0.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.0.attn2: kind=attention ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.1.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.1.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.1.attn2: kind=attention ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.2.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.2.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.2.attn2: kind=attention ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.3.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.3.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.3.attn2: kind=attention ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.4.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.4.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.4.attn2: kind=attention ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.5.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.5.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.5.attn2: kind=attention ratio=40%

========================================================================================
structured_40 PACKAGE 4/7 targets=27
========================================================================================
  up_blocks.0.attentions.0.transformer_blocks.6.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.6.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.6.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.7.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.7.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.7.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.8.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.8.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.8.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.9.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.9.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.9.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.0.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.0.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.0.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.1.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.1.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.1.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.2.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.2.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.2.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.3.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.3.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.3.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.4.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.4.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.4.attn2: H=1280 calls=64 sampled_tokens=8192
  prune up_blocks.0.attentions.0.transformer_blocks.6.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.6.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.6.attn2: kind=attention ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.7.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.7.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.7.attn2: kind=attention ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.8.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.8.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.8.attn2: kind=attention ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.9.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.9.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.0.transformer_blocks.9.attn2: kind=attention ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.0.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.0.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.0.attn2: kind=attention ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.1.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.1.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.1.attn2: kind=attention ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.2.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.2.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.2.attn2: kind=attention ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.3.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.3.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.3.attn2: kind=attention ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.4.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.4.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.4.attn2: kind=attention ratio=40%

========================================================================================
structured_40 PACKAGE 5/7 targets=27
========================================================================================
  up_blocks.0.attentions.1.transformer_blocks.5.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.5.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.5.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.6.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.6.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.6.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.7.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.7.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.7.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.8.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.8.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.8.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.9.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.9.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.9.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.0.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.0.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.0.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.1.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.1.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.1.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.2.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.2.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.2.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.3.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.3.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.3.attn2: H=1280 calls=64 sampled_tokens=8192
  prune up_blocks.0.attentions.1.transformer_blocks.5.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.5.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.5.attn2: kind=attention ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.6.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.6.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.6.attn2: kind=attention ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.7.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.7.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.7.attn2: kind=attention ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.8.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.8.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.8.attn2: kind=attention ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.9.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.9.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.1.transformer_blocks.9.attn2: kind=attention ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.0.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.0.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.0.attn2: kind=attention ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.1.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.1.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.1.attn2: kind=attention ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.2.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.2.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.2.attn2: kind=attention ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.3.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.3.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.3.attn2: kind=attention ratio=40%

========================================================================================
structured_40 PACKAGE 6/7 targets=39
========================================================================================
  up_blocks.0.attentions.2.transformer_blocks.4.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.4.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.4.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.5.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.5.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.5.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.6.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.6.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.6.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.7.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.7.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.7.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.8.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.8.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.8.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.9.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.9.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.9.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.0.transformer_blocks.0.ff: H=2560 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.0.transformer_blocks.0.attn1: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.0.transformer_blocks.0.attn2: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.0.transformer_blocks.1.ff: H=2560 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.0.transformer_blocks.1.attn1: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.0.transformer_blocks.1.attn2: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.1.transformer_blocks.0.ff: H=2560 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.1.transformer_blocks.0.attn1: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.1.transformer_blocks.0.attn2: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.1.transformer_blocks.1.ff: H=2560 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.1.transformer_blocks.1.attn1: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.1.transformer_blocks.1.attn2: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.2.transformer_blocks.0.ff: H=2560 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.2.transformer_blocks.0.attn1: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.2.transformer_blocks.0.attn2: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.2.transformer_blocks.1.ff: H=2560 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.2.transformer_blocks.1.attn1: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.2.transformer_blocks.1.attn2: H=640 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.0.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.0.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.0.attn2: H=1280 calls=64 sampled_tokens=8192
  prune up_blocks.0.attentions.2.transformer_blocks.4.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.4.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.4.attn2: kind=attention ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.5.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.5.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.5.attn2: kind=attention ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.6.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.6.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.6.attn2: kind=attention ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.7.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.7.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.7.attn2: kind=attention ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.8.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.8.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.8.attn2: kind=attention ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.9.ff: kind=ff ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.9.attn1: kind=attention ratio=40%
  prune up_blocks.0.attentions.2.transformer_blocks.9.attn2: kind=attention ratio=40%
  prune up_blocks.1.attentions.0.transformer_blocks.0.ff: kind=ff ratio=40%
  prune up_blocks.1.attentions.0.transformer_blocks.0.attn1: kind=attention ratio=40%
  prune up_blocks.1.attentions.0.transformer_blocks.0.attn2: kind=attention ratio=40%
  prune up_blocks.1.attentions.0.transformer_blocks.1.ff: kind=ff ratio=40%
  prune up_blocks.1.attentions.0.transformer_blocks.1.attn1: kind=attention ratio=40%
  prune up_blocks.1.attentions.0.transformer_blocks.1.attn2: kind=attention ratio=40%
  prune up_blocks.1.attentions.1.transformer_blocks.0.ff: kind=ff ratio=40%
  prune up_blocks.1.attentions.1.transformer_blocks.0.attn1: kind=attention ratio=40%
  prune up_blocks.1.attentions.1.transformer_blocks.0.attn2: kind=attention ratio=40%
  prune up_blocks.1.attentions.1.transformer_blocks.1.ff: kind=ff ratio=40%
  prune up_blocks.1.attentions.1.transformer_blocks.1.attn1: kind=attention ratio=40%
  prune up_blocks.1.attentions.1.transformer_blocks.1.attn2: kind=attention ratio=40%
  prune up_blocks.1.attentions.2.transformer_blocks.0.ff: kind=ff ratio=40%
  prune up_blocks.1.attentions.2.transformer_blocks.0.attn1: kind=attention ratio=40%
  prune up_blocks.1.attentions.2.transformer_blocks.0.attn2: kind=attention ratio=40%
  prune up_blocks.1.attentions.2.transformer_blocks.1.ff: kind=ff ratio=40%
  prune up_blocks.1.attentions.2.transformer_blocks.1.attn1: kind=attention ratio=40%
  prune up_blocks.1.attentions.2.transformer_blocks.1.attn2: kind=attention ratio=40%
  prune mid_block.attentions.0.transformer_blocks.0.ff: kind=ff ratio=40%
  prune mid_block.attentions.0.transformer_blocks.0.attn1: kind=attention ratio=40%
  prune mid_block.attentions.0.transformer_blocks.0.attn2: kind=attention ratio=40%

========================================================================================
structured_40 PACKAGE 7/7 targets=27
========================================================================================
  mid_block.attentions.0.transformer_blocks.1.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.1.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.1.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.2.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.2.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.2.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.3.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.3.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.3.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.4.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.4.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.4.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.5.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.5.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.5.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.6.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.6.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.6.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.7.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.7.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.7.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.8.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.8.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.8.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.9.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.9.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.9.attn2: H=1280 calls=64 sampled_tokens=8192
  prune mid_block.attentions.0.transformer_blocks.1.ff: kind=ff ratio=40%
  prune mid_block.attentions.0.transformer_blocks.1.attn1: kind=attention ratio=40%
  prune mid_block.attentions.0.transformer_blocks.1.attn2: kind=attention ratio=40%
  prune mid_block.attentions.0.transformer_blocks.2.ff: kind=ff ratio=40%
  prune mid_block.attentions.0.transformer_blocks.2.attn1: kind=attention ratio=40%
  prune mid_block.attentions.0.transformer_blocks.2.attn2: kind=attention ratio=40%
  prune mid_block.attentions.0.transformer_blocks.3.ff: kind=ff ratio=40%
  prune mid_block.attentions.0.transformer_blocks.3.attn1: kind=attention ratio=40%
  prune mid_block.attentions.0.transformer_blocks.3.attn2: kind=attention ratio=40%
  prune mid_block.attentions.0.transformer_blocks.4.ff: kind=ff ratio=40%
  prune mid_block.attentions.0.transformer_blocks.4.attn1: kind=attention ratio=40%
  prune mid_block.attentions.0.transformer_blocks.4.attn2: kind=attention ratio=40%
  prune mid_block.attentions.0.transformer_blocks.5.ff: kind=ff ratio=40%
  prune mid_block.attentions.0.transformer_blocks.5.attn1: kind=attention ratio=40%
  prune mid_block.attentions.0.transformer_blocks.5.attn2: kind=attention ratio=40%
  prune mid_block.attentions.0.transformer_blocks.6.ff: kind=ff ratio=40%
  prune mid_block.attentions.0.transformer_blocks.6.attn1: kind=attention ratio=40%
  prune mid_block.attentions.0.transformer_blocks.6.attn2: kind=attention ratio=40%
  prune mid_block.attentions.0.transformer_blocks.7.ff: kind=ff ratio=40%
  prune mid_block.attentions.0.transformer_blocks.7.attn1: kind=attention ratio=40%
  prune mid_block.attentions.0.transformer_blocks.7.attn2: kind=attention ratio=40%
  prune mid_block.attentions.0.transformer_blocks.8.ff: kind=ff ratio=40%
  prune mid_block.attentions.0.transformer_blocks.8.attn1: kind=attention ratio=40%
  prune mid_block.attentions.0.transformer_blocks.8.attn2: kind=attention ratio=40%
  prune mid_block.attentions.0.transformer_blocks.9.ff: kind=ff ratio=40%
  prune mid_block.attentions.0.transformer_blocks.9.attn1: kind=attention ratio=40%
  prune mid_block.attentions.0.transformer_blocks.9.attn2: kind=attention ratio=40%
structured_40: 1,693,602,564 params (1.6936B), physical reduction 34.0360%
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
  structured_40 case 0: 1.742s
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
  structured_40 case 1: 1.731s
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
  structured_40 case 2: 1.732s
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
  structured_40 case 3: 1.731s
There are modules in UNet2DConditionModel that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
  standalone torch.load verification: PASS

Fetching 17 files:   0%|          | 0/17 [00:00<?, ?it/s]
Fetching 17 files: 100%|██████████| 17/17 [00:00<00:00, 23736.07it/s]

Loading pipeline components...:   0%|          | 0/7 [00:00<?, ?it/s]
Loading pipeline components...:  43%|████▎     | 3/7 [00:00<00:00,  7.69it/s]
Loading pipeline components...:  71%|███████▏  | 5/7 [00:00<00:00,  8.89it/s]There are modules in UNet2DConditionModel that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.

Loading pipeline components...:  86%|████████▌ | 6/7 [00:00<00:00,  5.56it/s]There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.

Loading pipeline components...: 100%|██████████| 7/7 [00:01<00:00,  6.83it/s]

========================================================================================
structured_50 PACKAGE 1/7 targets=36
========================================================================================
  down_blocks.1.attentions.0.transformer_blocks.0.ff: H=2560 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.0.transformer_blocks.0.attn1: H=640 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.0.transformer_blocks.0.attn2: H=640 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.0.transformer_blocks.1.ff: H=2560 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.0.transformer_blocks.1.attn1: H=640 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.0.transformer_blocks.1.attn2: H=640 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.1.transformer_blocks.0.ff: H=2560 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.1.transformer_blocks.0.attn1: H=640 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.1.transformer_blocks.0.attn2: H=640 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.1.transformer_blocks.1.ff: H=2560 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.1.transformer_blocks.1.attn1: H=640 calls=64 sampled_tokens=8192
  down_blocks.1.attentions.1.transformer_blocks.1.attn2: H=640 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.0.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.0.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.0.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.1.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.1.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.1.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.2.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.2.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.2.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.3.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.3.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.3.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.4.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.4.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.4.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.5.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.5.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.5.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.6.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.6.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.6.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.7.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.7.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.7.attn2: H=1280 calls=64 sampled_tokens=8192
  prune down_blocks.1.attentions.0.transformer_blocks.0.ff: kind=ff ratio=50%
  prune down_blocks.1.attentions.0.transformer_blocks.0.attn1: kind=attention ratio=50%
  prune down_blocks.1.attentions.0.transformer_blocks.0.attn2: kind=attention ratio=50%
  prune down_blocks.1.attentions.0.transformer_blocks.1.ff: kind=ff ratio=50%
  prune down_blocks.1.attentions.0.transformer_blocks.1.attn1: kind=attention ratio=50%
  prune down_blocks.1.attentions.0.transformer_blocks.1.attn2: kind=attention ratio=50%
  prune down_blocks.1.attentions.1.transformer_blocks.0.ff: kind=ff ratio=50%
  prune down_blocks.1.attentions.1.transformer_blocks.0.attn1: kind=attention ratio=50%
  prune down_blocks.1.attentions.1.transformer_blocks.0.attn2: kind=attention ratio=50%
  prune down_blocks.1.attentions.1.transformer_blocks.1.ff: kind=ff ratio=50%
  prune down_blocks.1.attentions.1.transformer_blocks.1.attn1: kind=attention ratio=50%
  prune down_blocks.1.attentions.1.transformer_blocks.1.attn2: kind=attention ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.0.ff: kind=ff ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.0.attn1: kind=attention ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.0.attn2: kind=attention ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.1.ff: kind=ff ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.1.attn1: kind=attention ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.1.attn2: kind=attention ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.2.ff: kind=ff ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.2.attn1: kind=attention ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.2.attn2: kind=attention ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.3.ff: kind=ff ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.3.attn1: kind=attention ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.3.attn2: kind=attention ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.4.ff: kind=ff ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.4.attn1: kind=attention ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.4.attn2: kind=attention ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.5.ff: kind=ff ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.5.attn1: kind=attention ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.5.attn2: kind=attention ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.6.ff: kind=ff ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.6.attn1: kind=attention ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.6.attn2: kind=attention ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.7.ff: kind=ff ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.7.attn1: kind=attention ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.7.attn2: kind=attention ratio=50%

========================================================================================
structured_50 PACKAGE 2/7 targets=27
========================================================================================
  down_blocks.2.attentions.0.transformer_blocks.8.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.8.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.8.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.9.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.9.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.0.transformer_blocks.9.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.0.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.0.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.0.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.1.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.1.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.1.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.2.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.2.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.2.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.3.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.3.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.3.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.4.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.4.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.4.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.5.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.5.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.5.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.6.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.6.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.6.attn2: H=1280 calls=64 sampled_tokens=8192
  prune down_blocks.2.attentions.0.transformer_blocks.8.ff: kind=ff ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.8.attn1: kind=attention ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.8.attn2: kind=attention ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.9.ff: kind=ff ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.9.attn1: kind=attention ratio=50%
  prune down_blocks.2.attentions.0.transformer_blocks.9.attn2: kind=attention ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.0.ff: kind=ff ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.0.attn1: kind=attention ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.0.attn2: kind=attention ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.1.ff: kind=ff ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.1.attn1: kind=attention ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.1.attn2: kind=attention ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.2.ff: kind=ff ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.2.attn1: kind=attention ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.2.attn2: kind=attention ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.3.ff: kind=ff ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.3.attn1: kind=attention ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.3.attn2: kind=attention ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.4.ff: kind=ff ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.4.attn1: kind=attention ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.4.attn2: kind=attention ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.5.ff: kind=ff ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.5.attn1: kind=attention ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.5.attn2: kind=attention ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.6.ff: kind=ff ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.6.attn1: kind=attention ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.6.attn2: kind=attention ratio=50%

========================================================================================
structured_50 PACKAGE 3/7 targets=27
========================================================================================
  down_blocks.2.attentions.1.transformer_blocks.7.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.7.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.7.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.8.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.8.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.8.attn2: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.9.ff: H=5120 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.9.attn1: H=1280 calls=64 sampled_tokens=8192
  down_blocks.2.attentions.1.transformer_blocks.9.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.0.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.0.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.0.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.1.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.1.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.1.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.2.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.2.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.2.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.3.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.3.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.3.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.4.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.4.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.4.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.5.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.5.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.5.attn2: H=1280 calls=64 sampled_tokens=8192
  prune down_blocks.2.attentions.1.transformer_blocks.7.ff: kind=ff ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.7.attn1: kind=attention ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.7.attn2: kind=attention ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.8.ff: kind=ff ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.8.attn1: kind=attention ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.8.attn2: kind=attention ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.9.ff: kind=ff ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.9.attn1: kind=attention ratio=50%
  prune down_blocks.2.attentions.1.transformer_blocks.9.attn2: kind=attention ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.0.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.0.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.0.attn2: kind=attention ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.1.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.1.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.1.attn2: kind=attention ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.2.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.2.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.2.attn2: kind=attention ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.3.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.3.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.3.attn2: kind=attention ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.4.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.4.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.4.attn2: kind=attention ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.5.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.5.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.5.attn2: kind=attention ratio=50%

========================================================================================
structured_50 PACKAGE 4/7 targets=27
========================================================================================
  up_blocks.0.attentions.0.transformer_blocks.6.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.6.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.6.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.7.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.7.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.7.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.8.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.8.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.8.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.9.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.9.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.0.transformer_blocks.9.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.0.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.0.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.0.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.1.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.1.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.1.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.2.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.2.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.2.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.3.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.3.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.3.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.4.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.4.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.4.attn2: H=1280 calls=64 sampled_tokens=8192
  prune up_blocks.0.attentions.0.transformer_blocks.6.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.6.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.6.attn2: kind=attention ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.7.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.7.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.7.attn2: kind=attention ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.8.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.8.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.8.attn2: kind=attention ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.9.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.9.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.0.transformer_blocks.9.attn2: kind=attention ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.0.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.0.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.0.attn2: kind=attention ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.1.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.1.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.1.attn2: kind=attention ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.2.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.2.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.2.attn2: kind=attention ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.3.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.3.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.3.attn2: kind=attention ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.4.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.4.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.4.attn2: kind=attention ratio=50%

========================================================================================
structured_50 PACKAGE 5/7 targets=27
========================================================================================
  up_blocks.0.attentions.1.transformer_blocks.5.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.5.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.5.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.6.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.6.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.6.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.7.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.7.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.7.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.8.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.8.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.8.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.9.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.9.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.1.transformer_blocks.9.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.0.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.0.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.0.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.1.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.1.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.1.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.2.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.2.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.2.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.3.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.3.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.3.attn2: H=1280 calls=64 sampled_tokens=8192
  prune up_blocks.0.attentions.1.transformer_blocks.5.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.5.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.5.attn2: kind=attention ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.6.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.6.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.6.attn2: kind=attention ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.7.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.7.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.7.attn2: kind=attention ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.8.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.8.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.8.attn2: kind=attention ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.9.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.9.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.1.transformer_blocks.9.attn2: kind=attention ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.0.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.0.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.0.attn2: kind=attention ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.1.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.1.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.1.attn2: kind=attention ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.2.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.2.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.2.attn2: kind=attention ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.3.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.3.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.3.attn2: kind=attention ratio=50%

========================================================================================
structured_50 PACKAGE 6/7 targets=39
========================================================================================
  up_blocks.0.attentions.2.transformer_blocks.4.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.4.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.4.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.5.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.5.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.5.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.6.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.6.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.6.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.7.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.7.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.7.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.8.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.8.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.8.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.9.ff: H=5120 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.9.attn1: H=1280 calls=64 sampled_tokens=8192
  up_blocks.0.attentions.2.transformer_blocks.9.attn2: H=1280 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.0.transformer_blocks.0.ff: H=2560 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.0.transformer_blocks.0.attn1: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.0.transformer_blocks.0.attn2: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.0.transformer_blocks.1.ff: H=2560 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.0.transformer_blocks.1.attn1: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.0.transformer_blocks.1.attn2: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.1.transformer_blocks.0.ff: H=2560 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.1.transformer_blocks.0.attn1: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.1.transformer_blocks.0.attn2: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.1.transformer_blocks.1.ff: H=2560 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.1.transformer_blocks.1.attn1: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.1.transformer_blocks.1.attn2: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.2.transformer_blocks.0.ff: H=2560 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.2.transformer_blocks.0.attn1: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.2.transformer_blocks.0.attn2: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.2.transformer_blocks.1.ff: H=2560 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.2.transformer_blocks.1.attn1: H=640 calls=64 sampled_tokens=8192
  up_blocks.1.attentions.2.transformer_blocks.1.attn2: H=640 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.0.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.0.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.0.attn2: H=1280 calls=64 sampled_tokens=8192
  prune up_blocks.0.attentions.2.transformer_blocks.4.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.4.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.4.attn2: kind=attention ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.5.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.5.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.5.attn2: kind=attention ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.6.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.6.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.6.attn2: kind=attention ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.7.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.7.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.7.attn2: kind=attention ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.8.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.8.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.8.attn2: kind=attention ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.9.ff: kind=ff ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.9.attn1: kind=attention ratio=50%
  prune up_blocks.0.attentions.2.transformer_blocks.9.attn2: kind=attention ratio=50%
  prune up_blocks.1.attentions.0.transformer_blocks.0.ff: kind=ff ratio=50%
  prune up_blocks.1.attentions.0.transformer_blocks.0.attn1: kind=attention ratio=50%
  prune up_blocks.1.attentions.0.transformer_blocks.0.attn2: kind=attention ratio=50%
  prune up_blocks.1.attentions.0.transformer_blocks.1.ff: kind=ff ratio=50%
  prune up_blocks.1.attentions.0.transformer_blocks.1.attn1: kind=attention ratio=50%
  prune up_blocks.1.attentions.0.transformer_blocks.1.attn2: kind=attention ratio=50%
  prune up_blocks.1.attentions.1.transformer_blocks.0.ff: kind=ff ratio=50%
  prune up_blocks.1.attentions.1.transformer_blocks.0.attn1: kind=attention ratio=50%
  prune up_blocks.1.attentions.1.transformer_blocks.0.attn2: kind=attention ratio=50%
  prune up_blocks.1.attentions.1.transformer_blocks.1.ff: kind=ff ratio=50%
  prune up_blocks.1.attentions.1.transformer_blocks.1.attn1: kind=attention ratio=50%
  prune up_blocks.1.attentions.1.transformer_blocks.1.attn2: kind=attention ratio=50%
  prune up_blocks.1.attentions.2.transformer_blocks.0.ff: kind=ff ratio=50%
  prune up_blocks.1.attentions.2.transformer_blocks.0.attn1: kind=attention ratio=50%
  prune up_blocks.1.attentions.2.transformer_blocks.0.attn2: kind=attention ratio=50%
  prune up_blocks.1.attentions.2.transformer_blocks.1.ff: kind=ff ratio=50%
  prune up_blocks.1.attentions.2.transformer_blocks.1.attn1: kind=attention ratio=50%
  prune up_blocks.1.attentions.2.transformer_blocks.1.attn2: kind=attention ratio=50%
  prune mid_block.attentions.0.transformer_blocks.0.ff: kind=ff ratio=50%
  prune mid_block.attentions.0.transformer_blocks.0.attn1: kind=attention ratio=50%
  prune mid_block.attentions.0.transformer_blocks.0.attn2: kind=attention ratio=50%

========================================================================================
structured_50 PACKAGE 7/7 targets=27
========================================================================================
  mid_block.attentions.0.transformer_blocks.1.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.1.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.1.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.2.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.2.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.2.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.3.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.3.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.3.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.4.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.4.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.4.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.5.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.5.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.5.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.6.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.6.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.6.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.7.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.7.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.7.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.8.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.8.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.8.attn2: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.9.ff: H=5120 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.9.attn1: H=1280 calls=64 sampled_tokens=8192
  mid_block.attentions.0.transformer_blocks.9.attn2: H=1280 calls=64 sampled_tokens=8192
  prune mid_block.attentions.0.transformer_blocks.1.ff: kind=ff ratio=50%
  prune mid_block.attentions.0.transformer_blocks.1.attn1: kind=attention ratio=50%
  prune mid_block.attentions.0.transformer_blocks.1.attn2: kind=attention ratio=50%
  prune mid_block.attentions.0.transformer_blocks.2.ff: kind=ff ratio=50%
  prune mid_block.attentions.0.transformer_blocks.2.attn1: kind=attention ratio=50%
  prune mid_block.attentions.0.transformer_blocks.2.attn2: kind=attention ratio=50%
  prune mid_block.attentions.0.transformer_blocks.3.ff: kind=ff ratio=50%
  prune mid_block.attentions.0.transformer_blocks.3.attn1: kind=attention ratio=50%
  prune mid_block.attentions.0.transformer_blocks.3.attn2: kind=attention ratio=50%
  prune mid_block.attentions.0.transformer_blocks.4.ff: kind=ff ratio=50%
  prune mid_block.attentions.0.transformer_blocks.4.attn1: kind=attention ratio=50%
  prune mid_block.attentions.0.transformer_blocks.4.attn2: kind=attention ratio=50%
  prune mid_block.attentions.0.transformer_blocks.5.ff: kind=ff ratio=50%
  prune mid_block.attentions.0.transformer_blocks.5.attn1: kind=attention ratio=50%
  prune mid_block.attentions.0.transformer_blocks.5.attn2: kind=attention ratio=50%
  prune mid_block.attentions.0.transformer_blocks.6.ff: kind=ff ratio=50%
  prune mid_block.attentions.0.transformer_blocks.6.attn1: kind=attention ratio=50%
  prune mid_block.attentions.0.transformer_blocks.6.attn2: kind=attention ratio=50%
  prune mid_block.attentions.0.transformer_blocks.7.ff: kind=ff ratio=50%
  prune mid_block.attentions.0.transformer_blocks.7.attn1: kind=attention ratio=50%
  prune mid_block.attentions.0.transformer_blocks.7.attn2: kind=attention ratio=50%
  prune mid_block.attentions.0.transformer_blocks.8.ff: kind=ff ratio=50%
  prune mid_block.attentions.0.transformer_blocks.8.attn1: kind=attention ratio=50%
  prune mid_block.attentions.0.transformer_blocks.8.attn2: kind=attention ratio=50%
  prune mid_block.attentions.0.transformer_blocks.9.ff: kind=ff ratio=50%
  prune mid_block.attentions.0.transformer_blocks.9.attn1: kind=attention ratio=50%
  prune mid_block.attentions.0.transformer_blocks.9.attn2: kind=attention ratio=50%
structured_50: 1,475,137,284 params (1.4751B), physical reduction 42.5450%
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
  structured_50 case 0: 1.696s
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
  structured_50 case 1: 1.688s
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
  structured_50 case 2: 1.697s
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
There are modules in AutoencoderKL that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
  structured_50 case 3: 1.691s
There are modules in UNet2DConditionModel that should be kept in float32: []. Casting directly with `to()` can lead to inconsistent results; set `torch_dtype` in `from_pretrained()` instead to keep these modules in float32.
  standalone torch.load verification: PASS

========================================================================================
STRUCTURED OBS-DIFF SDXL COMPLETE
========================================================================================
dense           params=2,567,463,684 reduction=0.000% pth=0.000 GiB mean=1.9338s
structured_20   params=2,130,533,124 reduction=17.018% pth=3.969 GiB mean=1.8754s
structured_30   params=1,912,067,844 reduction=25.527% pth=3.562 GiB mean=1.7842s
structured_40   params=1,693,602,564 reduction=34.036% pth=3.155 GiB mean=1.7340s
structured_50   params=1,475,137,284 reduction=42.545% pth=2.749 GiB mean=1.6932s
JSON: /content/obs_diff_sdxl_structured_results/obs_sdxl_structured_compare.json
HTML: /content/obs_diff_sdxl_structured_results/obs_sdxl_structured_compare.html
UNets: /content/obs_diff_sdxl_structured_results/unets

RETURN CODE: 0