blocks-examples
blocks-examples copied to clipboard
mila-iqia
Proposed README
Would you guys kindly consider filling up the XXX and accepting this PR? Since this arguably the most important experiment in this repo, I believe it deservers a little extra information. What do you think?
I'll try to rerun the experiments and fill it up myself if you are interested. A TODO for this PR should also be a CLI to generate sample translations.
Failure is unrelated to what I did. Maybe the Travis script needs an update.
We need this change in Theano for travis to continue to work with conda:
https://github.com/Theano/Theano/pull/3597/files
This was fixed in Blocks by @dwf with https://github.com/mila-udem/blocks/commit/e185f3d26c7432f9ab5cb59c3b7f037092a5b9d1, but maybe not in blocks-examples?
On Wed, Nov 11, 2015 at 5:07 PM, Frédéric Bastien [email protected] wrote:
We need this change in Theano for travis to continue to work with conda:
https://github.com/Theano/Theano/pull/3597/files
— Reply to this email directly or view it on GitHub https://github.com/mila-udem/blocks-examples/pull/56#issuecomment-155924738 .
Bah, yes, I got blocks and blocks-extras I think. Need to do it in fuel and blocks-examples.
btw, I got out of memory in a 5Gb GPU. Which gpus you guys used for that experiment?
@EderSantana I got a similar OOM error
Error allocating 74400000 bytes of device memory (out of memory). Driver report 34045952 bytes free and 12079136768 bytes total
So, it looks like 75 MB of GPU memory is what you may need and your GPU might have been using all its memory. I tried by closing other GPU processes and it worked. I used K80 for this.
@anuragranj would you consider adding the numbers you got here? https://github.com/EderSantana/blocks-examples/blob/patch-1/machine_translation/README.md
Just edit it with Github itself.