outputs/nvfp4/deploy/save_r32.log

W0613 17:22:07.102000 24497 torch/utils/_pytree.py:630] <enum 'KernelPreference'> is an Enum subclass and is now natively supported by torch.compile as an opaque value type. Calling register_constant() on Enum subclasses is deprecated and will be an error in a future release.
W0613 17:22:07.133000 24497 torch/utils/_pytree.py:630] <enum 'ScaleCalculationMode'> is an Enum subclass and is now natively supported by torch.compile as an opaque value type. Calling register_constant() on Enum subclasses is deprecated and will be an error in a future release.
Unable to import `torchao` Tensor objects. This may affect loading checkpoints serialized with `torchao`
=== save real NVFP4 W4A4 r32 -> outputs/nvfp4/deploy/klein4b_nvfp4_w4a4_r32.safetensors ===

Loading pipeline components...:   0%|          | 0/5 [00:00<?, ?it/s]

Loading weights:   0%|          | 0/398 [00:00<?, ?it/s]

Loading weights:  17%|█▋        | 66/398 [00:00<00:00, 635.65it/s]

Loading weights:  34%|███▍      | 137/398 [00:00<00:00, 630.36it/s]

Loading weights:  54%|█████▎    | 213/398 [00:00<00:00, 682.73it/s]

Loading weights:  71%|███████   | 282/398 [00:00<00:00, 554.71it/s]

Loading weights:  92%|█████████▏| 367/398 [00:00<00:00, 644.44it/s]
Loading weights: 100%|██████████| 398/398 [00:00<00:00, 636.19it/s]

Loading pipeline components...:  20%|██        | 1/5 [00:01<00:07,  1.96s/it]
Loading pipeline components...:  40%|████      | 2/5 [00:02<00:03,  1.23s/it]
Loading pipeline components...:  80%|████████  | 4/5 [00:04<00:00,  1.12it/s]
Loading pipeline components...: 100%|██████████| 5/5 [00:04<00:00,  1.47it/s]
Loading pipeline components...: 100%|██████████| 5/5 [00:04<00:00,  1.17it/s]
teacher gen (reference)...

  0%|          | 0/4 [00:00<?, ?it/s]
 25%|██▌       | 1/4 [00:00<00:00,  6.97it/s]
100%|██████████| 4/4 [00:00<00:00, 12.84it/s]
100%|██████████| 4/4 [00:00<00:00, 12.07it/s]

  0%|          | 0/4 [00:00<?, ?it/s]
 75%|███████▌  | 3/4 [00:00<00:00, 16.26it/s]
100%|██████████| 4/4 [00:00<00:00, 14.48it/s]
/usr/local/lib/python3.11/dist-packages/nunchaku/lora/flux/utils.py:93: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:347.)
  result[[slice(0, extent) for extent in tensor.shape]] = tensor
converted 120 Linears in 462s
SAVED -> outputs/nvfp4/deploy/klein4b_nvfp4_w4a4_r32.safetensors (2.56GB)
quant gen (correctness check)...

  0%|          | 0/4 [00:00<?, ?it/s]
 25%|██▌       | 1/4 [00:00<00:00,  6.53it/s]
100%|██████████| 4/4 [00:00<00:00, 17.11it/s]

  0%|          | 0/4 [00:00<?, ?it/s]
100%|██████████| 4/4 [00:00<00:00, 39.46it/s]
100%|██████████| 4/4 [00:00<00:00, 39.42it/s]
DONE r32 -> outputs/nvfp4/deploy/klein4b_nvfp4_w4a4_r32.safetensors  (montages: outputs/nvfp4/deploy/cmp_r32_*.png)