codes about annotations

[liesun1994]

I feel puzzled when reading lamtram code , so is there any detailed annotations when lamtram will be updated next time?

[neubig]

I understand your puzzlement. As lamtram developed very organically with the progress of neural machine translation, there were some design decisions that don't make sense given what modern architectures look like. We're in the progress of creating a major overhaul that may be done sometime soon, which should be significantly easier to understand. I'll announce it here when we're done.

…

On Wed, Mar 1, 2017 at 8:46 AM, kun wang ***@***.***> wrote: I feel puzzled when reading lamtram code , so is there any detailed annotations when lamtram will be updated next time? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#23>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAYWG-Taq2bQEvrw0aU7c-HzxBFeiN7Dks5rhXbTgaJpZM4MPrEh> .

[liesun1994]

thx. it's a nice framework. hope it become a widely spread framework. and of course be easy to understand. Aha

[neubig]

We've made public xnmt, a translation framework based on DyNet that is meant to replace lamtram at some point:
https://github.com/neulab/xnmt/

There are still a number of features in lamtram that aren't implemented in xnmt yet, but we're planning on getting most of them implemented soon. If you're interested, take a look!