Add documentation for RNN transducer#1171
Conversation
| @@ -0,0 +1,21 @@ | |||
| .. role:: hidden | |||
| :class: hidden-section | |||
|
|
|||
There was a problem hiding this comment.
I think it's a good idea to note that prototype is nightly-only feature at the top of all the pages for prototype. So that there is a less chance of receiving the report like this
There was a problem hiding this comment.
I've labelled the issue as prototype and linked to the "source of truth" definition of prototype, which says "not part of release". Do you mean to say that you feel this may not be clear enough?
There was a problem hiding this comment.
Yes, I think we cannot be too careful here.
I suggest add .. note:: or .. warning:: and tell users to use nightly build in a more straightforward manner.
Reading the classification, I feel like it is not even included in nightly build, which is another binary distribution.
Prototype: These features are typically not available as part of binary distributions like PyPI or Conda, except sometimes behind run-time flags, and are at an early stage for feedback and testing.
b492ec5 to
75cf8e2
Compare
mthrok
left a comment
There was a problem hiding this comment.
Approving so that you are not blocked, can merge it and see how it is reflected at pytorch.org/audio
After the merge, you can run the nightly job manually from #306 and check how it looks on pytorch.org/audio. If you see an issue, feel free to open a new PR.
Add prototype RNN transducer loss to the documentation. This will need to be guarded off from releases.
cc #1137