Skip to content
This repository was archived by the owner on Jul 7, 2023. It is now read-only.

v1.5.6#686

Merged
lukaszkaiser merged 69 commits intotensorflow:masterfrom
rsepassi:push
Apr 5, 2018
Merged

v1.5.6#686
lukaszkaiser merged 69 commits intotensorflow:masterfrom
rsepassi:push

Conversation

@rsepassi
Copy link
Contributor

@rsepassi rsepassi commented Apr 5, 2018

No description provided.

Lukasz Kaiser and others added 30 commits April 5, 2018 10:09
Allow sampling from autoencoders and add an ordered discrete one.
f79f361 
PiperOrigin-RevId: 188739584
Bug fix, blacklist image_summary metric for TPU.
312bf3d 
PiperOrigin-RevId: 188768118
Fix exporter to work with image generation problems.
40213ef 
PiperOrigin-RevId: 188768263
Fix how num_samples is set in interactive decoding.
0004ed8 
PiperOrigin-RevId: 188826463
Set seed for decoder so that we can reproduce results for image gener…
0f44a32 
…ation problems

PiperOrigin-RevId: 188926627
Fix self.class_labels call in Text2ClassProblem
9b83219 
PiperOrigin-RevId: 189255171
Make t2t_avg_all work and add new papers.
9c3c29a 
PiperOrigin-RevId: 189274197
Add module for within block attention.
d5b0e28 
PiperOrigin-RevId: 189515902
Fix transformer decoding when using attention other than dot_product.
8739822 
PiperOrigin-RevId: 189602759
@royaurko
Use residual vector quantization
8d726c6 
PiperOrigin-RevId: 189621426
Add bipolar ReLU and ELU and use them in autoencoders, adding a resid…
57ffb37 
…ual one.

PiperOrigin-RevId: 189674996
@royaurko
Small bug in update for residual vq
4999347 
PiperOrigin-RevId: 189693410
@nshazeer
Make adafactor not crash for sparse updates (just call the dense code).
7008853 
PiperOrigin-RevId: 189754369
Fix bug in t2t_model.py where a random metric is returned rather than…
11b34e8 
… the full list of metrics.

PiperOrigin-RevId: 189756073
Add T2T paper to README.
4a7bdea 
PiperOrigin-RevId: 189852789
@brian-barnes
allow user to pass an additional feature batch_prediction_key throu…
5e53cd3 
…gh model_fn

PiperOrigin-RevId: 189870117
Separate latent model input embedding from the autoencoder, make it l…
f9e9aa7 
…arger to train better.

PiperOrigin-RevId: 189973354
Adds a new (local) modality: SigmoidClassSymbolModality for performin…
21526ac 
…g binary (sigmoid_cross_entropy_with_logits) classification.

PiperOrigin-RevId: 189977484
@royaurko
Separate latent model input embedding from the autoencoder, make it l…
954010f 
…arger to train better.

PiperOrigin-RevId: 189985804
@royaurko
Make a few fixes to use dvq without ema.
3306a31 
PiperOrigin-RevId: 189989463
Separate latent model input embedding from the autoencoder.
aece44e 
PiperOrigin-RevId: 189999191
Adding video problem as a first step to creating Video modality
00c5dec 
PiperOrigin-RevId: 190003725
Add multiscale imagenet problem
daab2b3 
PiperOrigin-RevId: 190004869
Add CelebA for multiple resolutions.
41faa82 
PiperOrigin-RevId: 190007134
@nshazeer
Fix to lm1b problems - go back to building vocabulary based on first
2f907f9 
PiperOrigin-RevId: 190016885
Enable bfloat16 for Transformer model.
f768cde 
PiperOrigin-RevId: 190074885
@royaurko
Compute losses before updating the residuals
1873a4c 
PiperOrigin-RevId: 190106063
Extend multi-resolution data generators to accept hyperparameters.
9d6135c 
PiperOrigin-RevId: 190106996
remove dp
da7e46c 
PiperOrigin-RevId: 190116862
Disable Travis export and serving test because of a TF Serving bug
121d1d4 
PiperOrigin-RevId: 190120427
Ryan Sepassi and others added 24 commits April 5, 2018 10:16
Use six.moves for xrange
a0c2f6c 
PiperOrigin-RevId: 190546387
@royaurko
Simplify code by removing slo which we are not using any more.
5390eae 
PiperOrigin-RevId: 190674800
Enable setting OOV token in Text2X problems
7e25c82 
PiperOrigin-RevId: 190699170
Fix loss averaging for moe models
feaadca 
PiperOrigin-RevId: 190789535
Encode UTF string features for both Python 2 and Python 3.
37296f8 
PiperOrigin-RevId: 190856578
@royaurko
Modify residual quantization to update the codebooks in sequential or…
14f49d6 
…der.

PiperOrigin-RevId: 190888091
Disable broken beam vs. fast decode_length test. Add TODO to fix.
354c9d1 
PiperOrigin-RevId: 190948289
Add iterator to read from multiple files
286f1fd 
PiperOrigin-RevId: 190956348
@royaurko
Get rid of pbn masking which is not needed
d6e0396 
PiperOrigin-RevId: 190966538
@aidangomez
Add ScalarSummary support for TPU
f9dda47 
PiperOrigin-RevId: 190971885
@nshazeer
Parameter quantization simulation experiments.
0567881 
PiperOrigin-RevId: 191000535
Internal merges
e9a329e 
PiperOrigin-RevId: 191126281
@brian-barnes
bumping Cloud ML Engine runtime version to 1.5
2d8f4b6 
PiperOrigin-RevId: 191237772
@royaurko
Clean up soft EM and use tf.exp instead of tf.nn.softmax
71c553f 
PiperOrigin-RevId: 191332800
Fix Squad problem and add SquadConcat with question and context conca…
111129e 
…tenated in inputs

PiperOrigin-RevId: 191349174
Fix variable scoping for transformer slow decoding.
f4e9961 
PiperOrigin-RevId: 191465623
Enable additional reserved tokens in Text2TextProblem. Add QUESTION_S…
edb9ca0 
…EPARATOR as a reserved token for QuestionAndContext2TextProblem. Update SquadConcat to use it.

PiperOrigin-RevId: 191501377
Add stacked autoencoders and a larger discrete autoencoder config.
3a71eef 
PiperOrigin-RevId: 191519145
@brian-barnes
support prediction requests to CMLE and adding serving_utils to decou…
66e644f 
…ple serving logic from query.py

PiperOrigin-RevId: 191520955
Make the basic_conv_gen model run on the gym problem.
39c46ff 
PiperOrigin-RevId: 191670194
Internal
c5518b8 
PiperOrigin-RevId: 191698639
Internal merge tensorflow#685
bca81be 
PiperOrigin-RevId: 191758046
Add forgotten unicode punctuation normalization to get_ende_bleu.
fc9335c 
PiperOrigin-RevId: 191758943
Update comment on shape in SymbolModality
b39d152 
PiperOrigin-RevId: 191759697
@googlebot
Copy link

googlebot commented Apr 5, 2018

We found a Contributor License Agreement for you (the sender of this pull request), but were unable to find agreements for all the commit author(s) or Co-authors. If you authored these, maybe you used a different email address in the git commits than was used to sign the CLA (login here to double check)? If these were authored by someone else, then they will need to sign a CLA as well, and confirm that they're okay with these being contributed to Google.
In order to pass this check, please resolve this problem and have the pull request author add another comment and the bot will run again. If the bot doesn't comment, it means it doesn't think anything has changed.

Lukasz Kaiser added 3 commits April 5, 2018 12:01
lukaszkaiser
lukaszkaiser approved these changes Apr 5, 2018
View reviewed changes
Copy link
Contributor

@lukaszkaiser lukaszkaiser left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks!

@lukaszkaiser lukaszkaiser merged commit 9053259 into tensorflow:master Apr 5, 2018
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.

Reviewers

1 more reviewer

@lukaszkaiser lukaszkaiser lukaszkaiser approved these changes

Reviewers whose approvals may not affect merge requirements

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

8 participants

@rsepassi @googlebot @lukaszkaiser @ndvbd @royaurko @nshazeer @brian-barnes @aidangomez