8000 Commits · hueck/tensor2tensor · GitHub

More Web Proxy on the site http://driver.im/

Commits

Commits on Jun 22, 2021

Internal

T2T Team
authored and
copybara-github
committed

Commits on May 11, 2021

A more generic packing op for seqio

0x0539
authored and
copybara-github
committed

Commits on May 3, 2021

A more generic packing op

0x0539
authored and
copybara-github
committed

Commits on Mar 5, 2021

[Mesh-TF] Add is_training as an arg to mtf.dropout

afrozenator
authored and
copybara-github
committed

Commits on Feb 25, 2021

Don't overspecify WeightNorm input_spec, match input_spec of wrapped

T2T Team
authored and
copybara-github
committed

Commits on Feb 10, 2021

Internal change

tjgq
authored and
copybara-github
committed

Commits on Nov 16, 2020

Add Seq2Edits (go/seq2edits-paper) to T2T.

T2T Team
authored and
copybara-github
committed

Commits on Sep 29, 2020

Removing layers cache to avoid issues with TF2 check on initialization.

tgale96
authored and
copybara-github
committed

Commits on Jul 25, 2020

Internal change.

Dmitry Tsarkov
authored and
copybara-github
committed

Commits on Jul 18, 2020

Merge of PR #1834

martinpopel
authored and
copybara-github
committed
Purely Google refactor

hsudhof
authored and
copybara-github
committed
add a backtranslation-based en-cs NMT problem and hparams (#1834)
martinpopel
authored

Commits on Jul 3, 2020

Fixes typo.

T2T Team
authored and
copybara-github
committed

Commits on Jun 27, 2020

Use separate markers for cached weights/state to improve clarity, correct state corruption bug and move and add comments to a few other functions for clarity.

Lukasz Kaiser
authored and
copybara-github
committed

Commits on Jun 18, 2020

Fix a bug that causes a crash when decoding for language models because initially targets is empty and only "infer_targets" is populated.

royaurko
authored and
copybara-github
committed

Commits on Jun 17, 2020

Bump T2T to 1.15.7

afrozenator
authored and
copybara-github
committed

Commits on Jun 16, 2020

Commits on Jun 12, 2020

Allow action vectors and images as input for training Emily. When using images, action visualization is added to tf summary.

T2T Team
authored and
copybara-github
committed

Commits on Jun 2, 2020

[T2T] Bump version to 1.15.6

afrozenator
authored and
copybara-github
committed

Commits on Jun 1, 2020

Fixing feature encoder for tf.string variable length features.

eli7
authored and
copybara-github
committed

Commits on May 29, 2020

adding hparam to make encoder self-attention optional.

eli7
authored and
copybara-github
committed

Commits on May 20, 2020

Add basic support for TF2 modeling.

rjpower
authored and
copybara-github
committed

Commits on May 14, 2020

Documentation update.

w-hat
authored and
copybara-github
committed

Commits on May 5, 2020

Automated refactoring to make code Python 3 compatible.

T2T Team
authored and
copybara-github
committed

Commits on Apr 29, 2020

Simplify implementation of sample_temperature_per_example and make it work with dynamic shapes.

T2T Team
authored and
copybara-github
committed

Commits on Apr 25, 2020

Automated refactoring to make code Python 3 compatible.

T2T Team
authored and
copybara-github
committed

Commits on Apr 23, 2020

Remove functional_ops based sru

jaingaurav
authored and
copybara-github
committed

Commits on Apr 22, 2020

Minor nits in README.md

afrozenator
authored and
copybara-github
committed

Commits on Apr 18, 2020

Commits on Apr 17, 2020

Set an older gym version to correct problems with OSS TF/gym interaction and segfaults.

Lukasz Kaiser
authored and
copybara-github
committed

0