I am getting the following error for the docker submission (ID: 9715908). Can you please help me regarding this? @vchung ** ERROR (nifti_image_write_hdr_img2): cannot open output file '/output/BraTS2021_00037.nii.gz'

Created by Srivathsa Pasumarthi subtleresearcher
@subtleresearcher , Great question. As far as Docker submissions go, you are set. You are given 2 total submissions for this phase however, so if you wish to further fine-tune the model, you are more than welcome to submit one more time before the deadline this Friday, Sept. 24th. We will consider the latest submission as the final one, and will run that model on the test data. Hope this clarifies things!
@vchung So can I consider that my Docker phase submission is completed? Or is there anything else left to do on my side? Please let me know. Thanks again for your help and patience. Really appreciate it.
Got it thanks.
@subtleresearcher , Apologies for the confusion. An update was sent out last month with respect to this phase, specifically: > When you submit your docker image, we will run them against 5 cases from the validation data to make sure your containers run, then we will run them against the unseen test set. Information pertaining to the test data is to not be shared with the participants, which is why we are using the validation data to help with debugging at this time.
Thank you @vchung for your help. In the scores CSV file, I was able to see only 5 cases. Are there only 5 cases in the test set? Or will the model be evaluated on more cases?
Hi @subtleresearcher , Unfortunately, I am still looking into your submissions, as I am still not able to reproduce the `nifti_image_write_hdr_img2` error when running the workflow locally. I did, however, locally score the output generated from the model of submission 9716123 and have annotated the submission as such. You may view your scores [here](https://www.synapse.org/#!Synapse:syn26227262). I will continue to look into your submission and will update you when a solution arises. Again, apologies for the inconvenience.
@vchung Is there any other solution or updates for this? Would really appreciate your help in any way.
I am getting the same error during resubmission also (Submission ID: 9716123) Entire log file content is here ----- Orig patch: (112, 112, 96). Model input patch: [112, 112, 96] Converting nii.gz to h5... 0%| | 0/1 [00:00 (4, 240, 240, 155), (240, 240, 155) img dirs [] 0 files are split to 0 training, 0 test 'all' 1 samples, num_modalities: 4, chosen: -1 Namespace(D_groupsize=1, D_pool_K=2, ablate_multihead=False, ablate_pos_embed_type=None, apply_attn_stage='early', attention_mode_dim=-1, attn_clip=500, aug_degree=0.5, backbone_type='i3d', base_initializer_range=0.02, batch_size=1, bb_feat_upsize=True, binarize=False, checkpoint_dir='runs/segtran', chosen_modality=-1, debug=False, device='cuda', ds_class='BratsSet', ds_split='all', eval_robustness=False, gpu='0', in_fpn_layers='34', in_fpn_scheme='AN', in_fpn_use_bn=False, inchan_to3_scheme='bridgeconv', input_dir='/input', input_patch_size=[112, 112, 96], input_scale=(1, 1, 1), iters='20000', job_name='brats-val', mid_type='shared', net='segtran', num_attractors=512, num_classes=4, num_modes=-1, num_translayers=1, only_first_linear_in_squeeze=True, orig_in_channels=4, orig_input_size=None, orig_patch_size=(112, 112, 96), out_fpn_layers='1234', out_fpn_scheme='AN', out_fpn_upsampleD_scheme='conv', pos_embed_every_layer=True, pos_in_attn_only=False, segtran_type='3d', task_name='brats', test_ds_name='val', test_interp=None, trans_output_type='private', translayer_compress_ratios=[1, 1], use_pretrained=True, use_squeezed_transformer=True, val=None, verbose_output=False, vis_mode=None, xyz_permute=None) 'args' orig in-feat: 1024, in-feat: 1024, out-feat: 1024, in-scheme: AN, out-scheme: AN, translayer_dims: [1024, 1024] Segtran Fusion Encoder with 1 trans-layers Fusion0-in-squeeze: pool_modes_feat=softmax, mid_type=shared, only_first_linear=True, trans_output_type=private Fusion0-in-squeeze in_feat_dim: 1024, feat_dim: 1024 Fusion0-squeeze-out: pool_modes_feat=softmax, mid_type=shared, only_first_linear=False, trans_output_type=private Fusion0-squeeze-out in_feat_dim: 1024, feat_dim: 1024 i3d created Loaded pretrained i3d model '/home/user/brats21/open_brats2020/src/models/segtran/networks/aj_i3d/aj_rgb_imagenet.pth' Initialize QK scheme: shared Initialize QK scheme: shared Model loaded from 'runs/segtran/iter_20000.pth' Voxels: [1, 2352, 1024]. Model DHW scales: 8x8x8. Total scales: [8.0, 8.0, 8.0] [[0. 0. 0. 0.] [0. 0. 0. 0.] [0. 0. 0. 0.]] BraTS2021_00037.h5: [[0. 0. 0. 0.] [0. 0. 0. 0.] [0. 0. 0. 0.]] 0%| | 0/1 [00:00 loading checkpoint /home/user/brats21/runs/attn_equinet/model_best.pth.tar => loaded checkpoint '/home/user/brats21/runs/attn_equinet/model_best.pth.tar' (epoch 200) IVDNet invalid literal for int() with base 10: '' using default group number ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ----- Creating FUSION_NET HD (Assymetric) network... ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Creating IVDNet => loading checkpoint /home/user/brats21/runs/ivdnet/model_best.pth.tar => loaded checkpoint '/home/user/brats21/runs/ivdnet/model_best.pth.tar' (epoch 200) Val dataset number of batch: 1 torch.Size([1, 4, 130, 170, 136]) torch.Size([1, 4, 130, 170, 136]) ext pred... (155, 240, 240, 4) Writing /output/BraTS2021_00037.nii.gz
@subtleresearcher , Apologies for the late reply! I looked into submission 9715908 but could not reproduce the error. I also ran your model from the submission locally and it worked as expected; many apologies for the troubles there. Can you see if submitting the model again will run through successfully?

Docker submission - cannot open output file page is loading…