New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Latest tensorflow has no compatible CUDA versions #694
Comments
`make generate ` Ref #694 Signed-off-by: Ben Firshman <ben@firshman.co.uk>
And update compatibility matrices with `make generate` Ref #694 Signed-off-by: Ben Firshman <ben@firshman.co.uk>
And update compatibility matrices with `make generate` Ref #694 Signed-off-by: Ben Firshman <ben@firshman.co.uk>
Thanks for the report! Cog 0.4.1 has been pushed with a couple of fixes for this. We now support Tensorflow 2.9.0 (but not 2.9.1 because that doesn't seem to be picked up by our version compatibility scraper yet). #697 We now also support unsupported tensorflow versions so the cog.yaml you posted above will now work. #696 You'll need to use CUDA 11.6 though, because 11.7 isn't supported by the NVIDIA Docker base images yet. Upgrade instructions are here: https://github.com/replicate/cog#upgrade |
Seems to be broken then. Would appreciate help if you were willing to do a PR. :) #700 |
I'm trying to wrap a model I made with Keras, but I can't get
cog build
to work because apparently the version of tensorflow I'm using:tensorflow==2.9.1
(the latest version) has no compatible CUDA versions. I can't override it manually either by specifying a version in my cog.yaml file, same error either way. Why is this? I know for a fact my 11.7 installation works fine with this version of Tensorflow.The text was updated successfully, but these errors were encountered: