Verbose transform logs

[DanBlanaru]

I'm using code slightly modified from https://github.com/Project-MONAI/tutorials/blob/main/3d_segmentation/spleen_segmentation_3d_lightning.ipynb.

However, there is this super verbose print on every step

Epoch 0:  10%|█         | 1/10 [00:00<00:06,  1.30it/s, loss=0.684, v_num=0]train:  tensor(0.6815, device='cuda:0', grad_fn=<AliasBackward0>)

Metadata

	sizeof_hdr: tensor([348], dtype=torch.int32)

	extents: tensor([0], dtype=torch.int32)

	session_error: tensor([0], dtype=torch.int16)

	dim_info: tensor([0], dtype=torch.uint8)

	dim: tensor([[  3, 218, 218,  89,   1,   1,   1,   1]], dtype=torch.int16)

	intent_p1: tensor([0.])

	intent_p2: tensor([0.])

	intent_p3: tensor([0.])

	intent_code: tensor([0], dtype=torch.int16)

	datatype: tensor([64], dtype=torch.int16)

	bitpix: tensor([64], dtype=torch.int16)

	slice_start: tensor([0], dtype=torch.int16)

	pixdim: tensor([[-1.,  2.,  2.,  5.,  0.,  0.,  0.,  0.]])

	vox_offset: tensor([0.])

	scl_slope: tensor([nan])

	scl_inter: tensor([nan])

	slice_end: tensor([0], dtype=torch.int16)

	slice_code: tensor([0], dtype=torch.uint8)

	xyzt_units: tensor([2], dtype=torch.uint8)

	cal_max: tensor([0.])

	cal_min: tensor([0.])

	slice_duration: tensor([0.])

	toffset: tensor([0.])

	glmax: tensor([0], dtype=torch.int32)

	glmin: tensor([0], dtype=torch.int32)

	qform_code: tensor([1], dtype=torch.int16)

	sform_code: tensor([1], dtype=torch.int16)

	quatern_b: tensor([-0.])

	quatern_c: tensor([1.])

	quatern_d: tensor([0.])

	qoffset_x: tensor([233.])

	qoffset_y: tensor([-373.4297])

	qoffset_z: tensor([26.5000])

	srow_x: tensor([[ -2.,  -0.,   0., 233.]])

	srow_y: tensor([[  -0.0000,    2.0000,   -0.0000, -373.4297]])

	srow_z: tensor([[ 0.0000,  0.0000,  5.0000, 26.5000]])

	affine: tensor([[[   2.0000,    0.0000,    0.0000, -207.0000],

         [   0.0000,    2.0000,    0.0000, -379.4297],

         [   0.0000,    0.0000,    5.0000,   11.5000],

         [   0.0000,    0.0000,    0.0000,    1.0000]]])

	original_affine: tensor([[[  -2.0000,   -0.0000,    0.0000,  233.0000],

         [  -0.0000,    2.0000,   -0.0000, -373.4297],

         [   0.0000,    0.0000,    5.0000,   26.5000],

         [   0.0000,    0.0000,    0.0000,    1.0000]]], dtype=torch.float64)

	as_closest_canonical: tensor([False])

	spatial_shape: tensor([[218, 218,  89]], dtype=torch.int16)

	original_channel_dim: ['no_channel']

	filename_or_obj: ['/data/dan_blanaru/AMOS22_preprocessed/labelsTr/amos_0001.nii.gz']


Applied operations

[ [ { class: 'Orientation',

      extra_info: { 'original_affine': array([[  -2.       ,   -0.       ,    0.       ,  233.       ],

       [  -0.       ,    2.       ,   -0.       , -373.4296875],

       [   0.       ,    0.       ,    5.       ,   26.5      ],

       [   0.       ,    0.       ,    0.       ,    1.       ]])},

      id: 140476851341200,

      orig_size: (218, 218, 89)},

    { class: 'DivisiblePad',

      extra_info: {'padded': [(0, 0), (3, 3), (3, 3), (3, 4)]},

      id: 140476851341248,

      orig_size: (218, 218, 89)},

    { class: 'CropForeground',

      extra_info: { 'cropped': [0, 0, 0, 0, 0, 0],

                    'pad_info': { class: 'Pad',

                                  extra_info: {'padded': [(0, 0), (0, 0), (0, 0), (0, 0)]},

                                  id: 140476851340768,

                                  orig_size: (224, 224, 96)}},

      id: 140476851340672,

      orig_size: (224, 224, 96)}]]

Is batch?: True


Epoch 0:  20%|██        | 2/10 [00:01<00:05,  1.48it/s, loss=0.684, v_num=0]

How do I get rid of this?

The transforms are

 train_transforms = Compose(
            [
                LoadImaged(keys=["image", "label"]),
                EnsureChannelFirstd(keys=["image", "label"]),
                Orientationd(keys=["image", "label"], axcodes="RAS"),
                DivisiblePadd(["image", "label"], 16),
                # Spacingd(
                #     keys=["image", "label"],
                #     pixdim=(1.5, 1.5, 2.0),
                #     mode=("bilinear", "nearest"),
                # ),
                ScaleIntensityRanged(
                    keys=["image"], a_min=-57, a_max=164,
                    b_min=0.0, b_max=1.0, clip=True,
                ),
                CropForegroundd(keys=["image", "label"], source_key="image"),

and the training config is

net = Net()

# set up loggers and checkpoints
log_dir = os.path.join(root_dir, "logs")
print("LOGDIR:", log_dir, os.path.exists(log_dir))
tb_logger = pytorch_lightning.loggers.TensorBoardLogger(
    save_dir=log_dir
)
tb_logger.log_graph(net)
# initialise Lightning's trainer.
trainer = pytorch_lightning.Trainer(
    gpus=[0],
    max_epochs=30,
    logger=tb_logger,
    enable_checkpointing=False,
    num_sanity_val_steps=1,
    # log_every_n_steps=10,
)
# train
trainer.fit(net)

Also, the versions of relevant packages are:

monai==0.9.1
pytorch-lightning==1.7.7
tensorboard==2.10.1

[wyli]

thanks for reporting, that's improved since v0.9.1, could you please try monai==1.0.1?

[DanBlanaru]

yep, solved it, thank you very much!