Model failing to train with 'onnx' error

[pbl007]

Hi,

I am running the demo notebook demo_train_pipeline.ipynb and got the following error; code was crashing and I only managed to get the exception by modifying to:
try:
tc.run()
except Exception as e:
print(e)

Image list for training ----->
Total stack number -----> 1
Noise image name -----> fish_localbrain.tif
Noise image shape -----> (6955, 492, 492)
Using 1 GPU(s) for training ----->
[Epoch 1/5] [Batch 3056/3056] [Total loss: 33473.38, L1 Loss: 173.18, L2 Loss: 66773.58] [ETA: 0:33:51] [Time cost: 471 s]
module 'onnx' has no attribute 'load_model_from_string'

[cabooster]

Hi! Make sure you download the latest version of DeepCAD-RT files and install conda environment according to readme. If it still doesn't works, check train_collection.py and comment out the lines contains torch.onnx.export funtion manually.

[LelouchLamperougeVI]

pip install onnx fixes it