However, if I use the default docker base image, I got below error messages;
So I added "faiss" to the requirements.txt, then I got below error messages;
And when I use another base image #nsml: nsml/default_ml:cuda9_torch1.0,
the session runs well though there's an ERROR msg like below;
(Just to get rid of the ERROR msg, I added "faiss-gpu" to the requirements.txt, but the creation of the session failed with a failure of pip install faiss-gpu.)
+This ERROR msg somehow disappeared now.
Could you let me know why the NSML default docker image doesn't work for me? (Using another base image worked, but I'm just curious!)
I pulled baseline repository and ran a session.
However, if I use the default docker base image, I got below error messages;
So I added "faiss" to the requirements.txt, then I got below error messages;
And when I use another base image #nsml: nsml/default_ml:cuda9_torch1.0, the session runs well though there's an ERROR msg like below;
(Just to get rid of the ERROR msg, I added "faiss-gpu" to the requirements.txt, but the creation of the session failed with a failure of pip install faiss-gpu.) +This ERROR msg somehow disappeared now.