| commit | db02fbd9bf4c6bcfaeb0d23794fad6137804e389 | [log] [tgz] |
|---|---|---|
| author | Aapo Kyrola <akyrola@fb.com> | Thu Aug 17 19:11:15 2017 -0700 |
| committer | Facebook Github Bot <facebook-github-bot@users.noreply.github.com> | Thu Aug 17 19:32:07 2017 -0700 |
| tree | f0caa7dd3190e1c1c5b51479a3cb85e16d7e8d52 | |
| parent | 5f612d974086b0fa7cca9576ec2815ccdf8a74b2 [diff] |
Fix stepworkspace sizing Summary: When forward-only mode, we need only 2 workspaces. Errornously we sized the length of the workspace vector to 2 if it was different than 2. But if it was longer (because the step workspaces was shared by an non-forward-only op), we end up deleting the workspaces. With RNN Executor, this is a problem, because it held a reference to the deleted workspaces. Without RNN executor, we just ended recrearing the nets. Reviewed By: jhcross Differential Revision: D5654534 fbshipit-source-id: 1e6276e63453831747fee6a85c5057f01b89fde5
Caffe2 is a lightweight, modular, and scalable deep learning framework. Building on the original Caffe, Caffe2 is designed with expression, speed, and modularity in mind.
Caffe2 research award competition request for proposals
Please use Github issues (https://github.com/caffe2/caffe2/issues) to ask questions, report bugs, and request new features.
Please participate in our survey (https://www.surveymonkey.com/r/caffe2). We will send you information about new releases and special developer events/webinars.
Caffe2 is released under the BSD 2-Clause license.