Nov 23 00:24:20 ERROR [0.077s]: test_tensor_sharing (jit.test_data_parallel.TestDataParallel)
Nov 23 00:24:20 ----------------------------------------------------------------------
Nov 23 00:24:20 Traceback (most recent call last):
Nov 23 00:24:20   File "/var/lib/jenkins/workspace/test/jit/test_data_parallel.py", line 118, in test_tensor_sharing
Nov 23 00:24:20     r0_forward = replica[0].forward(x)
Nov 23 00:24:20   File "/var/lib/jenkins/workspace/test/common_utils.py", line 82, in prof_meth_call
Nov 23 00:24:20     return prof_callable(meth_call, *args, **kwargs)
Nov 23 00:24:20   File "/var/lib/jenkins/workspace/test/common_utils.py", line 76, in prof_callable
Nov 23 00:24:20     return callable(*args, **kwargs)
Nov 23 00:24:20 RuntimeError: diff_view_meta->output_nr_ == 0 INTERNAL ASSERT FAILED at /var/lib/jenkins/workspace/torch/csrc/autograd/variable.cpp:326, please report a bug to PyTorch. 
Nov 23 00:24:20 The above operation failed in interpreter.
Nov 23 00:24:20 Traceback (most recent call last):
Nov 23 00:24:20   File "/opt/conda/lib/python3.6/site-packages/torch/nn/functional.py", line 1370
Nov 23 00:24:20     if input.dim() == 2 and bias is not None:
Nov 23 00:24:20         # fused op is marginally faster
Nov 23 00:24:20         ret = torch.addmm(bias, input, weight.t())
Nov 23 00:24:20               ~~~~~~~~~~~ <--- HERE
Nov 23 00:24:20     else:
Nov 23 00:24:20         output = input.matmul(weight.t())