Closed ramon-astudillo closed 5 years ago
@pschydlo found one error https://github.com/LxMLS/lxmls-toolkit/pull/136 though problem persists
There are two problems causing the failed CI issue:
torch.distributions.categorical
.torch==0.3.1
dependency was also specified inside tox.ini
, thus failing the tests.For reasons I haven't investigated further, running tests locally with tox does not cause this error, because tox install torch 0.4.0. However, if I have torch 0.3.1, install pytest, and then run pytest tests/test_sequence_models_deep_learning.py
, I get the exact same error as in TravisCI.
These are the fixes I propose (I'll submit a PR soon):
requirements.txt
, removing redundant dependencies specifications (which led to this problem).from torch.distributions import Categorical
instead of from torch.distributions.categorical import Categorical
. This can also be done to rollback the "ugly" fix @ramon-astudillo mentions in #137.OK, it's even worse. .travis.yml
holds yet another dependency on torch 0.3.1, and it directly calls pytest instead of tox. I'm going to fix it and push a new commit to #137.
@kepler solved one of the errors https://github.com/LxMLS/lxmls-toolkit/issues/128
there are other errors in the deep learning day. Maybe consequence of the RL day updates
https://travis-ci.org/LxMLS/lxmls-toolkit/jobs/543686960