Skip to content

Conversation

@jamesr66a
Copy link
Collaborator

Put these in test_jit for two reason:

  1. They'll get run under CI
  2. We have existing machinery here to test traced models and their gradients

These tests revealed many issues, which I've created issues for and will shortly create a meta-issue to track.

@jamesr66a jamesr66a changed the title End-to-end example-based robustness testing for hybrid frontend [JIT] End-to-end example-based robustness testing for hybrid frontend Jun 13, 2018
Copy link
Contributor

@apaszke apaszke left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM. Some code is unnecessary and could be removed.

test/test_jit.py Outdated
init.orthogonal_(self.conv4.weight)

net = Net(upscale_factor=4)
net._initialize_weights()

This comment was marked as off-topic.

test/test_jit.py Outdated
self.affine2 = nn.Linear(128, 2)

self.saved_log_probs = []
self.rewards = []

This comment was marked as off-topic.

@zdevito
Copy link
Contributor

zdevito commented Jun 14, 2018

@pytorchbot retest this please

1 similar comment
@zdevito
Copy link
Contributor

zdevito commented Jun 14, 2018

@pytorchbot retest this please

@jamesr66a
Copy link
Collaborator Author

@zdevito i think that one fusion test was indeed flaky; i don't see the failure anymore. I'm going to go ahead and merge

@jamesr66a jamesr66a merged commit 34c9d16 into pytorch:master Jun 14, 2018
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants