Assertion Error when starting decomposition using Pipeline.py

[Sagnik2810Sarkar]

I got the following error when trying the mansion.zip pipeline example provided in the readme.

Traceback (most recent call last): File "G:\SAGNIK_Material_Palette\Material_Palette\MaterialPalette\pipeline.py", line 25, in module = capture.get_inference_module(pt='model.ckpt') File "G:\SAGNIK_Material_Palette\Material_Palette\MaterialPalette\capture\utils\model.py", line 41, in get_inference_module assert Path(pt).exists() AssertionError

I cant figure out what the model.ckpt is supposed to be referencing because there is no ckpt file in the capture directory itself.

[wonjunior]

Hi @Sagnik2810Sarkar, did you download the model checkpoint from here?

[Sagnik2810Sarkar]

Sorry for the late reply, but yeah the model file actually downloaded in the wrong folder. Due to some hardware constraints further attempts at decomposition keep getting interrupted so I'll post an update as soon as it works or I get stuck somewhere else.

[Sagnik2810Sarkar]

After putting in the model file the whole process finishes with no errors but there are no outputs in the mansion directory. The logs do say predict returned no results. Here are the logs:

Lightning automatically upgraded your loaded checkpoint from v1.5.10 to v2.2.1. To apply the upgrade to your files permanently, run python -m pytorch_lightning.utilities.upgrade_checkpoint G:\SAGNIK_Material_Palette\Material_Palette\MaterialPalette\model.ckpt C:\Users\cheta\anaconda3\envs\matpal\lib\site-packages\lightning_fabric\connector.py:563: precision=16 is supported for historical reasons but its usage is discouraged. Please set your precision to 16-mixed instead! Using 16bit Automatic Mixed Precision (AMP) GPU available: True (cuda), used: True TPU available: False, using: 0 TPU cores IPU available: False, using: 0 IPUs HPU available: False, using: 0 HPUs You are using a CUDA device ('NVIDIA GeForce RTX 3060') that has Tensor Cores. To properly utilize them, you should set torch.set_float32_matmul_precision('medium' | 'high') which will trade-off precision for performance. For more details, read https://pytorch.org/docs/stable/generated/torch.set_float32_matmul_precision.html#torch.set_float32_matmul_precision Missing logger folder: ..\mansion\lightning_logs total=6 after=6 StableDiffusion list=None:all=[6/6] LOCAL_RANK: 0 - CUDA_VISIBLE_DEVICES: [0] C:\Users\cheta\anaconda3\envs\matpal\lib\site-packages\pytorch_lightning\trainer\connectors\data_connector.py:436: Consider setting persistent_workers=True in 'predict_dataloader' to speed up the dataloader worker initialization. Predicting DataLoader 0: 0%| | 0/6 [00:00<?, ?it/s]C:\Users\cheta\anaconda3\envs\matpal\lib\site-packages\pytorch_lightning\loops\prediction_loop.py:255: predict returned None if it was on purpose, ignore this warning... Predicting DataLoader 0: 100%|███████████████████████████████████████████████████████████| 6/6 [00:02<00:00, 2.81it/s]

[wonjunior]

There are no logs from Lightning, the outputs should be located in the same directory as the LoRA checkpoint. For example, the outputs of the grass region in the mansion example will be in

mansion\weights\grass\an_object_with_azertyuiop_texture\checkpoint-800

There, you will find a outputs directory - and if renorm is enabled - a out_renorm directory.

[Sagnik2810Sarkar]

Got it, I thought the outputs for decomposition would be in a separate folder so I didn't check these.

[wonjunior]

I will close this issue for now but don't hesitate to come back to me if you need more assistance.