This repository was archived by the owner on Jul 7, 2023. It is now read-only.
transformer_fast_decode, beam search: take an optional cache and return it #999
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
Note: I have added tests but I cannot run them because pytest claims "no test dir found testing here". |
…rn it Some models, eg. semantic parsing models with copying mechanisms, want to use the output of Transformer for multiple predictions. One way to do so is to modify the symbols_to_logits_fn to generate the additional predictions and save it in the cache dictionary. To do so, though, fast_decode() must allow an externally supplied cache, and must return it to the caller after the loop.
gcampax
force-pushed
the
wip/fast-decode
branch
from
608defc to
5ea09f3
Compare
lukaszkaiser
approved these changes
Jan 10, 2019
|
I'm terribly sorry that this PR got forgotten for such a long time! It's a great help indeed and was perfectly fine to merge from the start. Thanks again for doing this! |
tensorflow-copybara
pushed a commit
that referenced
this pull request
Jan 10, 2019
PiperOrigin-RevId: 228622817
kpe
pushed a commit
to kpe/tensor2tensor
that referenced
this pull request
Mar 2, 2019
…rn it (tensorflow#999) Some models, eg. semantic parsing models with copying mechanisms, want to use the output of Transformer for multiple predictions. One way to do so is to modify the symbols_to_logits_fn to generate the additional predictions and save it in the cache dictionary. To do so, though, fast_decode() must allow an externally supplied cache, and must return it to the caller after the loop.
kpe
pushed a commit
to kpe/tensor2tensor
that referenced
this pull request
Mar 2, 2019
PiperOrigin-RevId: 228622817
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Some models, eg. semantic parsing models with copying mechanisms,
want to use the output of Transformer for multiple predictions.
One way to do so is to modify the symbols_to_logits_fn to generate
the additional predictions and save them in the cache dictionary.
To do so, though, fast_decode() must allow an externally supplied cache,
and must return it to the caller after the loop.