Closed pchalasani closed 7 years ago
It looks like you installed nodejs directly (using apt-get/homebrew). Can you try installing nodejs using nvm
? This way we can make sure the version is correct and you won't need to use sudo
to install npm packages
I managed to get nodejs installed and also npm install -g deepforge
works as well ( I made it install in my home dir, so I don't need sudo
) on my Mac. But now I'm having problems with deepforge start
-- some brew link errors (permission issues I think). Frankly quite tiresome overall.
I would try npm install -g bower
(from here) and see if that fixes the original error.
On ubuntu, are you using NodeJS from the official repos or installing from elsewhere? The official repos contain a rather old version of NodeJS (it looks like v4.x).
Also, I would strongly recommend that you use nvm rather than a modified homebrew installation. Debugging brew linking errors with a customized installation of nodejs using homebrew could be tricky and I suspect a bit more involved than just using nvm
.
Otherwise, there are docker images for deepforge which would allow you to not have to worry about setting up a nodejs environment for the server. The instructions for starting deepforge using the docker images can be found in the docs.
Sorry for the trouble but thanks for the feedback! I can look into this some more and see what I can do to make the installation process go more smoothly (including updating the docs to officially recommend using nvm
).
Got it working finally on the mac -- used nvm
as above, and had to run deepforge start
a couple times, and finally it said it was running at localhost:8888
and I go there in chrome and I see the UI.
I already have RNN code in PyTorch, will this framework help me somehow keep track of my experiments and results, etc? (I guess I would need to install it on my AWS Ubuntu Deep Learning AMI, but wanted to know if it would be worth the trouble)
Deepforge currently doesn't support pytorch (just torch7 atm) but pytorch support is in the works (https://github.com/deepforge-dev/deepforge/issues/1024).
Of course, if you are using torch7 (in the future, pytorch), it should help manage experiments and results as they are available from the web ui as well as version controlled (both the code and the data) so past experiments are reproducible.
is this ready for use ?
On Mac OSX, I did
sudo npm install -g deepforge
On Ubuntu I get similar errors