Update CUDA version to 10.2 #1373
Conversation
|
Deploy preview for pytorch-tutorials-preview ready! Built with commit b01eac6 https://deploy-preview-1373--pytorch-tutorials-preview.netlify.app |
16c57b3 to
d2fe576
Compare
.circleci/config.yml
Outdated
| @@ -122,7 +122,7 @@ pytorch_tutorial_build_defaults: &pytorch_tutorial_build_defaults | |||
| command: | | |||
| set -e | |||
|
|
|||
| export pyTorchDockerImageTag=291 | |||
| export pyTorchDockerImageTag=9de29bef4a5dc0dd1dd19428d83e5a66a44a1ed2-d5ed57569b35aaa56850259a988130c0625addf2 | |||
There was a problem hiding this comment.
Typically the base images that we utilize for CI are just a commit SHA from $(git rev-parse HEAD:.circleci/docker) so they shouldn't include a -${WORKFLOW_ID} which is what I'm seeing here.
Basically any tag with a hyphen is reaped after 2 weeks in our build system by the ECR garbage collection script.
| export pyTorchDockerImageTag=9de29bef4a5dc0dd1dd19428d83e5a66a44a1ed2-d5ed57569b35aaa56850259a988130c0625addf2 | |
| export pyTorchDockerImageTag=9de29bef4a5dc0dd1dd19428d83e5a66a44a1ed2 |
|
Looks like we're really close on this. I'm rerunning the two failed workers. I think these are probably transient errors. I'll look deeper if they fail again. |
|
@malfet Looks like a windows issue. Can you look at it. |
|
Since this passes all but the two windows tests, I'm going to deploy this to allow us to start to test 1.8. Thanks! |
* use more modern python for devcontainer * update to cuda10.2 for circleci builds Co-authored-by: Brian Johnson <[email protected]>
* use more modern python for devcontainer * update to cuda10.2 for circleci builds Co-authored-by: Brian Johnson <[email protected]>
No description provided.