profile
viewpoint

mari-linhares/tensorflow-workshop 46

Code samples, slides and materials for TensorFlow Workshops

MarkDaoust/mvn 10

Multivariate Normal Distributions, in Python

MarkDaoust/probability 1

Probabilistic reasoning and statistical analysis in TensorFlow

MarkDaoust/addons 0

Useful extra functionality for TensorFlow 2.0 maintained by SIG-addons

MarkDaoust/agents 0

TF-Agents is a library for Reinforcement Learning in TensorFlow

MarkDaoust/altair 0

Declarative statistical visualization library for Python

MarkDaoust/beam 0

Apache Beam

MarkDaoust/dcgan-oreilly 0

Notebook for O'Reilly's "Deep Convolutional Generative Adversarial Networks"

pull request commenttensorflow/docs

Update images.ipynb

Darn square images...

kaimo455

comment created time in 6 days

PR closed tensorflow/docs

Reviewers
zh-CN: wmt17_translate.md cla: yes
+213 -0

1 comment

1 changed file

Adrian-Yan

pr closed time in 7 days

pull request commenttensorflow/docs

zh-CN: wmt17_translate.md

Hi we've moved all our translations to the docs-l10n repo, please submit this PR there.

https://github.com/tensorflow/docs-l10n

Thanks.

Adrian-Yan

comment created time in 7 days

delete branch tensorflow/docs

delete branch : 2.1

delete time in 8 days

create barnchtensorflow/docs

branch : r2.1

created branch time in 8 days

issue commenttensorflow/tensorflow

Documentaton issue: tf.split num_or_size_splits parameter

Thanks @yongtang!

bersbersbers

comment created time in 8 days

pull request commenttensorflow/docs-l10n

[JA] Proofreading on GitHub Actions

LGTM.

sfujiwara

comment created time in 8 days

issue closedtensorflow/tensorflow

what-if-tool link broken in tensorflow.org Tools page

#38076 URL(s) with the issue: https://www.tensorflow.org/resources/tools#

Please provide a link to the documentation entry, for example: https://www.tensorflow.org/resources/tools#

Description of issue (what needs changing):

Link to what-if-tool is broken, as what-if moved to new repo.

Clear description

Current broken link, when someone clicks on "Get Started" link at https://www.tensorflow.org/resources/tools

Broken link -> https://github.com/tensorflow/tensorboard/blob/master/tensorboard/plugins/interactive_inference/What_If_Tool_Notebook_Usage.ipynb

New Correct link should be -> https://github.com/PAIR-code/what-if-tool/blob/master/What_If_Tool_Notebook_Usage.ipynb

For example, why should someone use this method? How is it useful?

Correct links

https://github.com/PAIR-code/what-if-tool/blob/master/What_If_Tool_Notebook_Usage.ipynb

I wanted to submit a pull-request for this

closed time in 8 days

bikashkumars

issue commenttensorflow/tensorflow

what-if-tool link broken in tensorflow.org Tools page

Hi @bikashkumars, thanks for reporting this.

I've converted this to an internal bug assigned to the page owner (that's a page I can't fix).

I'm closing this as we have the internal bug now.

Thanks.

bikashkumars

comment created time in 8 days

pull request commenttensorflow/io

Added Visual Studio Code tutorial, and updated README

Yes, docs/_toc.yaml makes the first section, tutorials/_toc.yaml makes the second section in the left-nav here:

https://www.tensorflow.org/io/overview

vlasenkoalexey

comment created time in 8 days

pull request commenttensorflow/quantum

Uoft data

With regards to the open license, how do we arrange that? Do I need to sign something similar to the CLA?

If you're the author, I believe it's sufficient to just include a "LICENESE.txt" with the data download. But if it's going in the repo, then it should be covered by the CLA and top level "LICENSE.txt" file.

therooler

comment created time in 11 days

PR closed tensorflow/examples

Reviewers
Adding example for Transfer Learning cla: yes

!wget --no-check-certificate
https://storage.googleapis.com/mledu-datasets/inception_v3_weights_tf_dim_ordering_tf_kernels_notop.h5
-O /tmp/inception_v3_weights_tf_dim_ordering_tf_kernels_notop.h5

+22 -0

1 comment

1 changed file

officialsiddharthchauhan

pr closed time in 11 days

pull request commenttensorflow/examples

Adding example for Transfer Learning

Hi, Thanks for taking the time to make a PR.

This example is incomplete, and the proposed topic is already well covered by existing tutorials.

So we will not be accepting this PR.

officialsiddharthchauhan

comment created time in 11 days

issue commenttensorflow/tensorflow

tf.signal.rfft documentation refers to Tcomplex as an argument

@rryan

It looks like this is caused by the these two wrappers:

https://github.com/tensorflow/tensorflow/blob/v2.2.0/tensorflow/python/ops/signal/fft_ops.py#L114-L140

https://github.com/tensorflow/tensorflow/blob/v2.2.0/tensorflow/python/ops/signal/fft_ops.py#L145-L172

They hide the Tcomplex/Treal argument to the wrapped function, but docstring is copied over.

arvoelke

comment created time in 13 days

push eventMarkDaoust/docs

Mark Daoust

commit sha 43c730d9fe7f20f4044f3284cd851124c66b8edb

Add 3d-tensors

view details

push time in 14 days

create barnchMarkDaoust/docs

branch : tensor

created branch time in 14 days

pull request commenttensorflow/docs

fix resolve minor typo

No problem. Thanks for trying.

Gomyo

comment created time in 15 days

PR closed tensorflow/docs

Reviewers
fix resolve minor typo cla: yes

Hello :) I found a minor typo during tutorial.

It occurs NameError!

+1 -1

2 comments

1 changed file

Gomyo

pr closed time in 15 days

pull request commenttensorflow/docs

fix resolve minor typo

Hi, thanks for taking the time to send a PR.

I don't see a problem with this notebook: 1) We have tests for this notebook, it is still passing. 2) I ran it in colab, it works. 3) That label argument refers to the function argument label defined just above:

def pack(features, label):
  return tf.stack(list(features.values()), axis=-1), label
Gomyo

comment created time in 15 days

pull request commenttensorflow/docs

Definition of underfitting

PS: I'm actually an employee, not sure whether I should sign

If you register your github account with go/github the google-bots will recognize you after that (if you configure git to use your google email)

PedroGFonseca

comment created time in 15 days

push eventMarkDaoust/docs

Jake Tae

commit sha 56a14b06447227fd9163ae355b2f9407ac322fc2

Delete unnecessary max_length function As pointed out in #28371, the `max_length` function is unnecessary since `tf.keras.preprocessing.sequence.pad_sequences` called within `tokenize` already pads sequences by `maxlen`, which defaults to the length of the longest sequence. Therefore, the maximum length can simply be obtained by examining its shape.

view details

Mark Daoust

commit sha 33f62e328c3a6ca7e14823cd03c483b7a344e70e

Fix imports. PiperOrigin-RevId: 306719877

view details

Yash Katariya

commit sha 8b2e63bbeb1209e13fa8698e4d858ba0c6eae377

Remove tf_inspect and use inspect.signature. PiperOrigin-RevId: 306866145

view details

Copybara-Service

commit sha bb12b62f766b3c810fd12428b0eefdf2a8a1a86c

Merge pull request #1477 from jaketae:patch-1 PiperOrigin-RevId: 306878316

view details

Yash Katariya

commit sha b1f4c5b19264318b11fe02c0a17835c4f753b4ed

Check for singleton values when filtering in `ignore_typing` PiperOrigin-RevId: 306935162

view details

Yash Katariya

commit sha 8e3693077be200596935778743d75ce7318a5905

Check for callable when creating a module page PiperOrigin-RevId: 306935202

view details

Jake Tae

commit sha 221c971bb8f33ef951130ce90be0d26408335e36

fix: resolve minor typo and punctuation issue

view details

Jake Tae

commit sha 3091affe4f65d5af5c61a46b83956491d5097445

fix: resolve comment typo

view details

Mark Daoust

commit sha a70508620e84ed4cecf759636d2d49569aa39555

Trim redundant comments.

view details

Taycir Yahmed

commit sha d5d25cfd47e51ca62c6458fff9786e2de0d1a71c

typo Typos in notebook : site/en/tutorials/keras/overfit_and_underfit.ipynb

view details

Copybara-Service

commit sha 20d60b9cfca3fd6c7fa41d54a3d50e378f094f4f

Merge pull request #1546 from jaketae:patch-1 PiperOrigin-RevId: 307069920

view details

Mark Daoust

commit sha 1395bb8349912713996ac46d5ea9122c1541965d

Strip TODOs PiperOrigin-RevId: 307392766

view details

Edward Loper

commit sha 1fb5dc8a53046130957c9e81d8b2f74d74984ab4

Functions returned by the `get_concrete_function` method of `tf.Function` objects can now be called with arguments consistent with the original arguments or type specs passed to `get_concrete_function`. In particular: * If a composite tensor (such as RaggedTensor or SparseTensor) was passed to `get_concrete_function`, then the returned function will accept a composite tensor of the same type for that argument. * If a nested structure (such as a list or dict) was passed to `get_concrete_function`, then the returned function will accept a value with the same nesting structure. Each tensor or composite tensor value must have the same type as was used in the original argument; and each non-Tensor value (such as bools or ints) must be equal value that was used in the original argument. * If a non-tensor value (such as a bool or int) was passed to `get_concrete_function`, then the returned function no longer deletes that argument; instead, it updates the argument's default value to the value that was passed to `get_concrete_function`. Passing in any other value will raise an exception. * Arguments are not renamed based on `TensorSpec.name`. For backwards compatibility, the functions returned by `get_concrete_function` will continue to accept arguments with the existing calling conventions (where nested structures and composite tensors are flattened; non-tensor arguments are deleted; suffixes are automatically added to disambiguate arguments with the same name; and TensorSpec.name is used to rename arguments). However, the preferred calling convention is the one that is consistent with the original arguments or type specs passed to `get_concrete_function`. PiperOrigin-RevId: 307398918

view details

Mark Daoust

commit sha dd6c6e4cfa705ad963031c801ad64b349c30157b

Centralize object type checking. Package MemberInfo and MethodInfo. PiperOrigin-RevId: 307405801

view details

Yash Katariya

commit sha 310fde53cfb7046abbb8926b79740f6c975ee791

Internal PiperOrigin-RevId: 307467935

view details

Taycir Yahmed

commit sha 80c03e3ef4464b6790d76f96e4d69f1103608e0c

Merge pull request #1 from tensorflow/master .

view details

Mark Daoust

commit sha 60fab890661147935246747065154778561e3d6b

Fix format string. PiperOrigin-RevId: 307673136

view details

Edward Loper

commit sha 529ba4346b8fc5e830e762a2f0ee87b3c345c0c9

Update the concrete_function guide to describe the fact that concrete functions can now be called with the same calling convention as polymorphic functions. (New calling convention for concrete functions was added in PR f39aab309). Also updated the guide to discuss composite tensor and nested argument values. PiperOrigin-RevId: 307691942

view details

Mark Daoust

commit sha dc61d8f3895f923ac1865bad7a24bfac74be3386

Or test set.

view details

Copybara-Service

commit sha 7b289f32e2f5b23c00ba2be6d3eb4effa5595b6d

Merge pull request #1549 from tensorflow:MarkDaoust-patch-1 PiperOrigin-RevId: 307838824

view details

push time in 15 days

pull request commenttensorflow/tensorflow

Adding Instruction to install tf using Conda

We don't own/test/support TensorFlow with conda, so it's hard for us to include this in the official docs.

Right @lamberta?

shubham15gupta09

comment created time in 16 days

pull request commenttensorflow/docs

tensorflow-2.2.0 seems to work well with CUDA 10.2

See line #427

This is in a section called. "Tested Build configurations". It might work with plenty of other configurations, but be run a lot of tests on something before we make it official, so we can't accept this PR.

Thanks for taking the time.

WurmD

comment created time in 16 days

PR closed tensorflow/docs

Reviewers
tensorflow-2.2.0 seems to work well with CUDA 10.2 cla: yes

TF 2.2 seems to work well with CUDA 10.2 in my server.

Should we make that official by updating the docs?

+1 -1

1 comment

1 changed file

WurmD

pr closed time in 16 days

PR opened tensorflow/tensorboard

Freshen notebooks.
  • Update TensorFlow versions to 2.2
  • Remove GPU from graphs.ipynb (runs in <10s on CPU)
  • Refresh output on graphs.ipynb
  • Remove nightly installs from profiling.ipynb
+47 -48

0 comment

4 changed files

pr created time in 17 days

push eventMarkDaoust/tensorboard

Mark Daoust

commit sha 78909f1039725d46c1e64056c2d18d3e9eef9e25

graphs.ipynb: refresh output

view details

Mark Daoust

commit sha 121b19e1d0ec67a9395a179976ef26de25d877d2

profiling: remove outdated installs

view details

push time in 17 days

push eventMarkDaoust/tensorboard

Mark Daoust

commit sha e073886e8db16bac6fe2f1c6c93e63a80bde2ace

Remove GPU from 'graphs.ipynb'

view details

push time in 17 days

push eventMarkDaoust/tensorboard

Mark Daoust

commit sha 187d0eb457647aa3c4029a8437c3cf9b7b0c6062

update tensorflow versions, remove __future__

view details

push time in 17 days

push eventMarkDaoust/tensorboard

William Chargin

commit sha 7e65cb9818386209f6943de206bf47a851d050de

Use `rel="noopener"` for `target="_blank"` links (#2516) Summary: Links with `target="_blank"` open in a new tab, but by default that new tab can control the page that opened it via the `window.opener` object. Using `rel="noopener"` closes this security hole. Test Plan: No automated test for now; we tend to catch these in code review, so this is a backfill. Manual verification: ``` $ git grep -o noopener '*.html' | wc -l 20 $ git grep -o _blank '*.html' | wc -l 20 ``` wchargin-branch: rel-noopener

view details

Stephan Lee

commit sha 089e2917b562227388567722506deca5a2a9e3d5

polymer2: fixed plugin dropdown UI issue (#2517) Now that we have real ShadowDom, the input within the paper-input has to be styled through mulitple layers of webcomponents, correctly. Normally, we would be using ShadyCSS (CSS mixin did not make it to the spec) and a mixin API exposed by appropriate WebComponents to achieve that but it is not viable in this case. Specifically, paper-dropdown-menu allows styling of paper-input using a mixin but in order to style input within the paper-input, we need to use another mixin. ShadyCSS explicitly says the nested mixin is not supported. Instead of transforming with CSS, we decided to fix the issue by pre-processing the tabName in JavaScript. Also, we fixed the wrong CSS selector after paper-menu -> paper-listbox change. See: https://github.com/webcomponents/polyfills/tree/master/packages/shadycss#known-issues

view details

William Chargin

commit sha 2fe5301a587f5e3a0ac5dd5fedfea7e3b10a5d55

scalars: omit empty runs from `/tags` route (#2520) Summary: This is not a user-facing change, because the runs selector is populated by the separate `/data/runs` endpoint. As we begin to consider generic data providers (cf. #2490), it seems superfluous that a “list scalars” call to a storage layer might be required to include _all_ run names in the response even if only a few have relevant data and the rest will be ignored, anyway. Test Plan: Tests updated. Manually confirmed that runs like `text_demo` continue to appear in the scalar dashboard’s run selector even though they have no scalar data. wchargin-branch: scalars-omit-empty

view details

Stephan Lee

commit sha d6875a2bb890f0ec2d03b74035ed11c6ef0c22b4

Wrap rollup_bundle around tf_js_binary for interop (#2521) In order to sync the latest change that includes rollup_bundle into Google, we need to make a wrapper that change some behavior internally. Extraneous changes are auto formatter (buildifier plugin) acting up.

view details

HarikrishnanBalagopal

commit sha a8203ba8390ec9adc146f37c74c0ce2657d66c77

Fixed Typo: 'RestNet' -> 'ResNet' (#2526)

view details

Stephan Lee

commit sha c3a63e1478b8ecaf54449432c9ed0da87869e5d2

ui: fixed graph causing dropdown-menu bug (#2523) repro case: https://next.plnkr.co/edit/4MCqtk5HAo3oTOKa Unclear exactly why it behaves as it does. We can posit that it is a bug with how ShadyCSS handles the mixin and applies.

view details

Stephan Lee

commit sha 0e9a000a1ef484762e743335a6b5754154bd9cdd

ng: remove tf_web_libray target (#2524) This change removes the tf_web_library target from the tf_ng_tensorboard to unblock google sync process. Generally speaking, tf_web_library is an impediment in general to our building/development cycle so we should reconsider using the tf_web_library for angular source at all. We are currently thinking that Angular source can depend on Polymer but Polymer cannot depend on Angular.

view details

Shanqing Cai

commit sha 7e9a96faac61a7ebc5167f6b944a825ecb4cf6f3

[graph plugin] Prevent health pills from showing up by default; Fix layout and default (#2528) * Motivation for features / changes * Towards fixing https://github.com/tensorflow/tensorboard/issues/2518 * Fixes https://github.com/tensorflow/tensorboard/issues/2527 * Technical description of changes * Check not only the `nodeNamesToHealthPills` variable is truthy, but also that the number of its keys is greater than 0. * Correct layout of the "show health pill" toggle. * Screenshots of UI changes ![image](https://user-images.githubusercontent.com/16824702/62734080-16601600-b9f6-11e9-84b4-466e8d46054e.png) The health pills no longer show up by default. The layout is fixed.

view details

Shanqing Cai

commit sha 2195fd39b59fb694e358edafeb6df5a259f405d4

[tensor-widget] Render basic tensor info (#2515) * Technical description of changes * Add tensor-widget-impl.ts that will hold the core logic that renders the tensor widget. * Add the style sheet for TensorWidget: tensor-widget.css * Add code that renders the header section and the basic info subsection of it. * The basic info subsection includes the following fields: * Name (optional) * Tensor DType * Tensor Shape * Add string-utilts.ts to handle the formatting of long tensor names. * Add shape-utils.ts. Right now it only has a trivial helper function to format shape strings. * Add unit tests for string-utils.ts and shape-utils.ts. * Add four tensor widget panels in demo.html. * Screenshots of UI changes ![image](https://user-images.githubusercontent.com/16824702/62636305-4aa8d900-b907-11e9-9de3-b4d2a726b5c1.png)

view details

William Chargin

commit sha afbdf876423189a7e7db792515d7e9c817707f9c

Make scalars plugin tests mostly TF 2.x–compatible (#2522) Summary: These tests were disabled for two reasons: they used `tf.placeholder`, and they used `tf.contrib.summary` for DB mode tests. This patch revises the implementation of the non–DB mode tests to avoid using placeholders, and enables those tests in all versions of TensorFlow. The DB mode tests still cannot be run in TF 2.x because `create_db_writer` is not exported under any public APIs. Test Plan: Tests pass in current `tf-nightly` and `tf-nightly-2.0-preview`. wchargin-branch: scalars-plugin-tests-tf2x

view details

William Chargin

commit sha 1c9412d1846f92473bbe742fdcd63363bd65b5f5

build: add `.bazelrc` to disable Python toolchains (#2535) Summary: This expands our forward-compatibility window to at least 0.29.0rc4 by opting out of the `--incompatible_use_python_toolchains` flag whose default was flipped in 0.27.0 (see #2355). Fixes #2534. Test Plan: On Bazel 0.29.0rc4, verify that `//tensorboard` fetches, builds, and runs correctly and that all tests pass, without the need to manually specify any flags to Bazel. wchargin-branch: bazelrc-toolchains

view details

Qiumin Xu

commit sha 9759e27ca841539faa5ae5e2506fb13588c47560

Improve memory viewer and tf-op-profile tool. (#2525) * Improve memory viewer and tf-op-profile tool. Add internal changes that have been made that are not in Github repository. * fix formatting * address comments * fix comment

view details

Stephan Lee

commit sha 2eb607e2823e3c3e48c772ef641ac8ad5f03a8df

infra: fix tomservo and nightly build (#2539) With the symlink_node_modules (default is True), the macro created a symlink from $PWD/node_modules into bazel internal directory (e.g., wherever bazel-tensorboard/external/npm/node_modules leads to). Because we rm -rf the source directory, the node_modules folder gets removed, creating a broken symlink. This change makes sure npm install does not create such symlink but instead creates every folder under the bazel internal directory.

view details

William Chargin

commit sha bb17d5f34ff92cafda1a369ba7d1176f2c2b1cf7

Fix dynamic plugin bootstrap sequence in Firefox (#2540) Summary: We initialize dynamic plugins by creating a new iframe and injecting into it a `script` element with `import("path/to/index.js")`. This does not work in Firefox, throwing what appears to be an internal error: <https://bugzilla.mozilla.org/show_bug.cgi?id=1573262> @stephanwlee had the idea to try wrapping the contents of the inner frame’s script with `setTimeout(…, 0)`, which worked! It turns out that we have to move the `baseURI` initialization into that deferred context, too. Fixes #2536. Test Plan: Launch TensorBoard and navigate to `/#projector`. Confirm that the projector plugin loads properly in both Chrome 76 and Firefox 68. The vanilla JS `example_basic` plugin also works in both now, too. Co-authored-by: Stephan Lee <stephanwlee@gmail.com> wchargin-branch: dynamic-plugin-firefox

view details

Scott Wegner

commit sha 57461efe5ec223d30865f821325a19d2f1ae2de0

Fix links for example ops (tf.matmul, tf.nn.relu) in README.md (#2545)

view details

Stephan Lee

commit sha 1dac2e4d080edc830174ed08bb6e606e295faeb2

graph: provide built-binary for legacy target (#2544) Internal build requires the built binary.

view details

William Chargin

commit sha 062c1641b9aacdb2a7acc1bdb86ef2b131197e9d

data: add partial generic data provider base class (#2529) Summary: This commit begins an initiative to enable TensorBoard to consume data from multiple sources using a single API. Like TensorBoard’s existing experimental “DB mode”, this opens the door to addressing fundamental performance limitations. Unlike DB mode, this is not specific to any particular database implementation. In the short term, we’ll provide one implementation of these APIs via a thin layer on top of the existing event multiplexer system. We will migrate plugins to read data using the new APIs instead of from the multiplexer directly. At that point, we can begin looking into providing alternate, more efficient implementations: e.g., an SQLite-backed store whose state persists across TensorBoard invocations. For now, we include only the API calls for listing and reading scalar data. We’ll expand the API surface as we proceed. Test Plan: Unit tests included (there’s not much to test). wchargin-branch: data-provider

view details

William Chargin

commit sha 3655a3b3e6d44cb2f1ddb761fc5098f9680f8866

data: add multiplexer provider implementation (#2530) Summary: This commit implements the new generic data provider APIs on top of the existing `plugin_event_multiplexer.EventMultiplexer` store. Test Plan: Unit tests included. wchargin-branch: multiplexer-provider

view details

William Chargin

commit sha 8cb63105e8fafcdfc75fea1ac61fcbc4ec47c67c

data: add `--generic_data` flag and `TBContext` field (#2531) Summary: This patch adds a `--generic_data` flag that defaults to `auto`; in the `auto` state, each plugin determines whether its usage of the generic data APIs is stable enough. The flag can also be set to `true` for testing and dogfooding, or to `false` as an escape hatch in case some bugs make it into the plugins deemed stable. Setting the flag to `false` also skips constructing the data provider object itself. wchargin-branch: generic-data-flag

view details

William Chargin

commit sha bf2274710900c008c8c1347603138f8977ae5751

data: add scalars plugin integration (#2532) Summary: This patch teaches the scalars plugin about the new generic data APIs. These are guarded behind a feature flag (`--generic_data=true`), which is intended to have no user-visible effect. Test Plan: Existing unit tests expanded, reusing the test cases from the existing multiplexer tests. For manual testing, add logging to verify when the new APIs are being used: ```diff diff --git a/tensorboard/plugins/scalar/scalars_plugin.py b/tensorboard/plugins/scalar/scalars_plugin.py index 41c3db3c..cd086f57 100644 --- a/tensorboard/plugins/scalar/scalars_plugin.py +++ b/tensorboard/plugins/scalar/scalars_plugin.py @@ -150,6 +150,9 @@ class ScalarsPlugin(base_plugin.TBPlugin): def scalars_impl(self, tag, run, experiment, output_format): """Result of the form `(body, mime_type)`.""" if self._data_provider: + from tensorboard.util import tb_logging + tb_logging.get_logger().warn("Data provider: %r", self._data_provider) + # Downsample reads to 1000 scalars per time series, which is the # default size guidance for scalars under the multiplexer loading # logic. ``` Then, run `//tensorboard` pointing at a logdir with scalar data, with all three values for `--generic_data` (`false`, `auto`, `true`). Verify that TensorBoard works correctly in all cases, and that the logging indicates that it is using the new APIs only with `--generic_data=true`. wchargin-branch: generic-scalars

view details

push time in 17 days

create barnchMarkDaoust/tensorboard

branch : notebooks

created branch time in 17 days

create barnchtensorflow/docs

branch : 2.1

created branch time in 19 days

PR closed tensorflow/tensorflow

Reviewers
Pull request cla: no

Pull request

+104982 -48506

3 comments

2550 changed files

sushovan14

pr closed time in 20 days

pull request commenttensorflow/tensorflow

Pull request

Hi @sushovan14,

It looks like you made some mistake with branches or merges and a bunch of the history is getting duplicated.

Please make a new PR with only your commits in it.

sushovan14

comment created time in 20 days

pull request commenttensorflow/quantum

Uoft data

Yes, if the data is clearly open-licensed, then we can put it in one of our buckets (downloads.tensorflow.org, or TFQ could make their own.)

15 kb of data.

If that's the whole thing I'd be tempted to just put it right in the git-repository and distribute it directly with TFQ.

therooler

comment created time in 21 days

pull request commenttensorflow/addons

Don't hide utils from the API docs generator

Would it make sense to import that utils.types in the root init.py so you don't document the rest of utils too?

Or switch to that explicit package thing and add import tensorflow_addons.utils.types to utils?

Can you post a diff of the output TOC either way?

yashk2810

comment created time in 21 days

pull request commenttensorflow/docs

Fixing Oxford pets dataset download error in tutorials/images/segmentation.ipynb

The tfds team is working on a fix, so let's hold for now.

logiccodes

comment created time in 22 days

pull request commenttensorflow/examples

Fixed forgotten activation function

There are a couple of tricky things about the activation=sigmoid approach, and I'd rather not encourage it. But maybe I want too far when I removed them all.

As I was working on the course,

If the course material mentions these specifically, then we should fix it, and we'll just remember not to use this pattern in the future.

guruyaya

comment created time in 22 days

push eventtensorflow/quantum

Mark Daoust

commit sha dd8610cd6c326bb2715d99b7cfaffb24b42cc19f

Fix gradients.ipynb

view details

Mark Daoust

commit sha 0202da77c6bc7565724bb523a25ec065d4874106

Merge pull request #231 from tensorflow/MarkDaoust-patch-1 Fix typo in gradients.ipynb

view details

push time in 22 days

PR merged tensorflow/quantum

Reviewers
Fix typo in gradients.ipynb
+1 -1

1 comment

1 changed file

MarkDaoust

pr closed time in 22 days

PR opened tensorflow/quantum

Fix typo in gradients.ipynb
+1 -1

0 comment

1 changed file

pr created time in 22 days

create barnchtensorflow/quantum

branch : MarkDaoust-patch-1

created branch time in 22 days

pull request commenttensorflow/tensorflow

Fix incorrect usage of tf.broadcast_weights in docstring

Sorry about that Yong. Thank you too.

yongtang

comment created time in 22 days

PR opened tensorflow/quantum

pin cirq versions
+602 -662

0 comment

5 changed files

pr created time in 22 days

create barnchMarkDaoust/quantum

branch : pin_cirq

created branch time in 22 days

push eventMarkDaoust/quantum

Mark Daoust

commit sha 1a16097ebb0a1d1bdcbf9bcf9b1343c84533ce4f

no-restart (#146) * no-restart * remove restart instructions

view details

Mark Daoust

commit sha 22cf3058ae5bfb4f0ca7ed70cb691bd7be650e61

rm api_docs/, these are on tensorflow.org now. (#147)

view details

Mark Daoust

commit sha b75e063b8baf0224273926e44a7ff5e533070ef9

Reduce mnist runtime (#155) Uncomment remove_contradicting

view details

MichaelBroughton

commit sha c40af6bdcf8bc642e43bbeb00bcbe3906742128a

Hello update (#156) * Modified Hello Many worlds tutorial to verify outputs. * metadata fix. * last of metadata. * Billy feedback. * break things down. * removed the. * Update docs/tutorials/hello_many_worlds.ipynb Co-Authored-By: Billy Lamberta <blamb@google.com> * Removed blank cell at end. Co-authored-by: Michael Broughton <mbbrough@google.com> Co-authored-by: Billy Lamberta <blamb@google.com>

view details

MichaelBroughton

commit sha 3b934bf3ae623a8996c040ecc3e18f592c54dfdd

Added ability to sample from state vector in qsim. (#153) * Added ability to sample from state vector in qsim. * downgraded random values to floats. Co-authored-by: Michael Broughton <mbbrough@google.com>

view details

abhinavsp0730

commit sha 953cf0d93930f2fedafcb1dcaa804cc00aadf722

Update barren_plateaus.ipynb (#141) * Update barren_plateaus.ipynb changed this ```try: %tensorflow_version 2.x except Exception: pass ``` to ```try: # %tensorflow_version only works in Colab. %tensorflow_version 2.x except Exception: pass ``` * Revert "Update barren_plateaus.ipynb" This reverts commit db9e2e5bd32b5c4f40b69a9b5de9274e5163cf9f. * Update barren_plateaus.ipynb * Update barren_plateaus.ipynb * Update gradients.ipynb * Update hello_many_worlds.ipynb * Update mnist.ipynb * Update qcnn.ipynb * Update gradients.ipynb * Update hello_many_worlds.ipynb * Update mnist.ipynb * Update qcnn.ipynb Co-authored-by: MichaelBroughton <MichaelBroughton@users.noreply.github.com>

view details

MichaelBroughton

commit sha a0ef76b80e33e64b6035e3a31322958fd2e8e945

Move sample (#158) * Moved sample into statespace_test. * cleanup. Co-authored-by: Michael Broughton <mbbrough@google.com>

view details

MichaelBroughton

commit sha 396ffa15917caeda89f842bfd4abd6359cadd463

Added flag to enable _USE_MATH_DEFINES on windows Windows builds were flaking because they were unable to see the `M_PI` symbol. This will add the `_USE_MATH_DEFINES` flag into our builds which should hopefully fix the issue.

view details

Jae H. Yoo

commit sha d5c02073e011ffa8e69eb55d9791cc2fce6b76df

Merge pull request #159 from tensorflow/windows_math Added flag to enable _USE_MATH_DEFINES on windows

view details

MichaelBroughton

commit sha 7ab25b12d7f8e5838c0e1aa033e94e9d8ce562ee

Added Stack Overflow tag to readme.md (#163) * Added Stack Overflow tag to readme.md * added more words

view details

MichaelBroughton

commit sha 4fbd426f0a2d7180d0ea3c0eac5ec3bf4642d289

Added z basis transformation to circuit_parser. (#162) Co-authored-by: Michael Broughton <mbbrough@google.com>

view details

MichaelBroughton

commit sha fe81a992cb94596a7796c3db05a720fa8388dcc1

Change from multiprocessing to pathos for windows compatability. (#161) Co-authored-by: Michael Broughton <mbbrough@google.com>

view details

MichaelBroughton

commit sha 260ad7e4631ed10109eab9fc34f242cfc1aac3af

Some small util ops cleanup. (#165) * Some small util ops cleanup. * missed some tstrings. Co-authored-by: Michael Broughton <mbbrough@google.com>

view details

MichaelBroughton

commit sha 3461d74a6f5cf6f1b8c0309c80140e7b78310961

Modified layers to use .shape instead of len when dealing with Tensors. (#168) Co-authored-by: Michael Broughton <mbbrough@google.com>

view details

MichaelBroughton

commit sha 32bcf5ff19b0fc0289abd9c9f40f94b0767a045b

Added symbol properties to high level layers. (#167) Co-authored-by: Michael Broughton <mbbrough@google.com>

view details

Antonio Martinez

commit sha 723e460e1ef2afecfbb545a85b54519a405d3642

State space tests (#172) * Updated test names * Added Valid() test * Initialization tested * Initializer test passing * Test * Only test in base cc file * Remove unneeded file * Aded back file * Added GetType tests * Added NULL checks to avoid seg faults * Clone tested * Added amplitude tests * Improved docstring wording * SetStateZero tested * CopyFrom tested * tested GetRealInnerProduct * Reorganized * formatting * initial 1 qubit apply test * Added slow 1 gate test * Updated ApplyGate1 test * new headers * Build update * Update tested * extended update tests * Generalized update test * test return * Initial ComputeExpectation test passing * Corrected test endianess * Phase flip tests passing * Formatting * Addressed review comments

view details

Antonio Martinez

commit sha 650309fbd2ebffd57fd6105b2fe49199a97d2ffd

replaced tf version (#174)

view details

MichaelBroughton

commit sha c483dcde96bceca6fe1dd0d9dad3bc4e8fb49c34

Fixed bias in sample. (#175) Co-authored-by: Michael Broughton <mbbrough@google.com>

view details

MichaelBroughton

commit sha 37d7d0d661950b0904123678aba43b65e9a86e47

Fixed small oob error in sampling. (#178) Co-authored-by: Michael Broughton <mbbrough@google.com>

view details

MichaelBroughton

commit sha 46aa11bf64664174d5d725b4aec22da7f0d8b7b8

Added Symbol values function to PQC (#177) * Added symbol_values function to PQC. * strange format. Co-authored-by: Michael Broughton <mbbrough@google.com>

view details

push time in 22 days

push eventtiancheng2000/quantum

MichaelBroughton

commit sha 18e9cb70a14d6cb64891222f7042561f6a94ba3d

Fixed formatting causing autograph problem. (#207) Co-authored-by: Michael Broughton <mbbrough@google.com>

view details

MichaelBroughton

commit sha c1c23dbff2b81cd4ae0d5ce021be04af42e87c69

Add Unitary calculation to qsim (#205) * initial work on unitary function. * all tests written and passing. * experiment with actions clang-format version? * fix clang-format version to 6? * A Feedback. Co-authored-by: Michael Broughton <mbbrough@google.com>

view details

MichaelBroughton

commit sha 7ff60baee7f7da0d2e09a5fa51f84c5ed951f35e

Pinned version of google-api-client. (#216) Co-authored-by: Michael Broughton <mbbrough@google.com>

view details

MichaelBroughton

commit sha dee7653864f2ab0ed2cb38e9be9f1c9593e0439f

Added support for Cirq.I. (#209) * Added support for Cirq.I. * shortened I gate reference. Co-authored-by: Michael Broughton <mbbrough@google.com>

view details

MichaelBroughton

commit sha c55b0af391f2245c52198c289e23f01664130cbd

Added python version support. (#214) Co-authored-by: Michael Broughton <mbbrough@google.com>

view details

MichaelBroughton

commit sha 42d04622e67dd5f122ef9c423e7678e7a8c98d50

Added tfq.calculate_unitary op. (#215) * Added tfq.calculate_unitary op. * format. * missed import check. * Jae Feedback. * added missing import check. * missed import.

view details

Michael Broughton

commit sha 482239db53c84de53e942e84443d6ce96cc6014a

Fix windows compiler warnings.

view details

Jae H. Yoo

commit sha 856e198227ad3e4777528cf1b90c09d0f6dd2f4c

Merge pull request #218 from tensorflow/win_fix Fix windows compiler warnings.

view details

Jae H. Yoo

commit sha 6efc39a2ab73693d3417d40516935788188306b3

Change size_t to ptrdiff_t (#219)

view details

Antonio Martinez

commit sha ddf4b18a87ff405729e148117b3f3573842d7be4

Input check refactor (#220) * initial files * Renamed files * Updating the tests * initial compile passing * input error tests * finished tests * updated docstring * Added rank checks * Refactored sample layer * refactored state layer * build fix, added rank test * refactored sampled expectation * removed shape checks * refactored expectation * formatting * removed lint * Factored out operator expansion * factored out operator * Renamed expansion, cut superfluous checks * update docstrings * Added tests * Formatting * Removed lint * reverted to original expansion logic * formatting * comment

view details

Mark Daoust

commit sha 9ddffdf62f367086c7d251ec11f9cce14f77e5b7

Merge branch 'master' into patch-1

view details

push time in 22 days

push eventtiancheng2000/quantum

MichaelBroughton

commit sha 18e9cb70a14d6cb64891222f7042561f6a94ba3d

Fixed formatting causing autograph problem. (#207) Co-authored-by: Michael Broughton <mbbrough@google.com>

view details

MichaelBroughton

commit sha c1c23dbff2b81cd4ae0d5ce021be04af42e87c69

Add Unitary calculation to qsim (#205) * initial work on unitary function. * all tests written and passing. * experiment with actions clang-format version? * fix clang-format version to 6? * A Feedback. Co-authored-by: Michael Broughton <mbbrough@google.com>

view details

MichaelBroughton

commit sha 7ff60baee7f7da0d2e09a5fa51f84c5ed951f35e

Pinned version of google-api-client. (#216) Co-authored-by: Michael Broughton <mbbrough@google.com>

view details

MichaelBroughton

commit sha dee7653864f2ab0ed2cb38e9be9f1c9593e0439f

Added support for Cirq.I. (#209) * Added support for Cirq.I. * shortened I gate reference. Co-authored-by: Michael Broughton <mbbrough@google.com>

view details

MichaelBroughton

commit sha c55b0af391f2245c52198c289e23f01664130cbd

Added python version support. (#214) Co-authored-by: Michael Broughton <mbbrough@google.com>

view details

MichaelBroughton

commit sha 42d04622e67dd5f122ef9c423e7678e7a8c98d50

Added tfq.calculate_unitary op. (#215) * Added tfq.calculate_unitary op. * format. * missed import check. * Jae Feedback. * added missing import check. * missed import.

view details

Michael Broughton

commit sha 482239db53c84de53e942e84443d6ce96cc6014a

Fix windows compiler warnings.

view details

Jae H. Yoo

commit sha 856e198227ad3e4777528cf1b90c09d0f6dd2f4c

Merge pull request #218 from tensorflow/win_fix Fix windows compiler warnings.

view details

Jae H. Yoo

commit sha 6efc39a2ab73693d3417d40516935788188306b3

Change size_t to ptrdiff_t (#219)

view details

Antonio Martinez

commit sha ddf4b18a87ff405729e148117b3f3573842d7be4

Input check refactor (#220) * initial files * Renamed files * Updating the tests * initial compile passing * input error tests * finished tests * updated docstring * Added rank checks * Refactored sample layer * refactored state layer * build fix, added rank test * refactored sampled expectation * removed shape checks * refactored expectation * formatting * removed lint * Factored out operator expansion * factored out operator * Renamed expansion, cut superfluous checks * update docstrings * Added tests * Formatting * Removed lint * reverted to original expansion logic * formatting * comment

view details

Mark Daoust

commit sha 7af428e957edc4be22c9f464965e19eed3ccdd41

Merge branch 'master' into patch-2

view details

push time in 22 days

push eventtiancheng2000/quantum

Mark Daoust

commit sha 428a69e81694d47d807866b48bf67d517d1a3e73

capture output so it doesn't print the arrays.

view details

push time in 22 days

push eventtiancheng2000/quantum

Mark Daoust

commit sha 4d3058a3736ac6f322bf5851c838aa7c866ff7ec

Don't use the binary-non-contradictory data.

view details

push time in 22 days

pull request commenttensorflow/quantum

y_train_hinge must be paired with y_train_nocon

This change looks correct to me, I'm not sure how it got out of sync, but I'm running it once to confirm that everything works.

tiancheng2000

comment created time in 22 days

pull request commenttensorflow/quantum

y_train_hinge must be paired with y_train_nocon

@MichaelBroughton

I had to pin cirq==0.7 to make this work, otherwise (cirq==0.8) import tfq fails with

/usr/local/lib/python3.6/dist-packages/tensorflow_quantum/core/serialize/serializer.py in _eigen_gate_serializer(gate_type, serialized_id)
    112             serialized_name="exponent",
    113             serialized_type=float,
--> 114             gate_getter=lambda x: _symbol_extractor(x.exponent)),
    115         cirq.google.SerializingArg(
    116             serialized_name="exponent_scalar",

TypeError: __init__() got an unexpected keyword argument 'gate_getter'
tiancheng2000

comment created time in 22 days

push eventtiancheng2000/quantum

Mark Daoust

commit sha 7ce603f9477035091b8fe531cb26c342733e2650

pin cirq==0.7

view details

push time in 22 days

PR closed tensorflow/tensorflow

Reviewers
Added TANH op and test to micro cla: yes size:XL

see previous pull request: https://github.com/tensorflow/tensorflow/pull/39055

+60754 -45658

2 comments

1628 changed files

yair-ehrenwald

pr closed time in 23 days

pull request commenttensorflow/tensorflow

Added TANH op and test to micro

Files changed 1,628

This still doesn't look right.

Can you try remaking this PR with just your changes?

Update your master branch, create your pr branch from there, apply your changes, and submit that.

It's impossible to review like this. I've never seen this error before:

Commits that don’t share a common merge base with the base branch are not supported.
You can view the full commit page here instead.
yair-ehrenwald

comment created time in 23 days

Pull request review commenttensorflow/quantum

Uoft data

+# Copyright 2020 The TensorFlow Quantum Authors. All Rights Reserved.+#+# Licensed under the Apache License, Version 2.0 (the "License");+# you may not use this file except in compliance with the License.+# You may obtain a copy of the License at+#+#     http://www.apache.org/licenses/LICENSE-2.0+#+# Unless required by applicable law or agreed to in writing, software+# distributed under the License is distributed on an "AS IS" BASIS,+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.+# See the License for the specific language governing permissions and+# limitations under the License.+# ==============================================================================+"""Quantum datasets for several quantum many-body spin systems in the form of+parameterized circuits"""+import os+import numpy as np+import sympy+import scipy.sparse as sps+import cirq+++class SpinSystem(object):+    """Abstract spin system class containing information about the quantum+    data+    """++    def __init__(self, g: float, res_e: float, fidelity: float,+                 gs_energy: float, parameters: np.ndarray, path: str):+        """Instantiate this spin system.++        Args:+            g: Order parameter+            res_e: Residual energy between the exact diagonalization ground+            state and the circuit state.+            fidelity: Fidelity between the exact diagonalization ground state+            and the circuit state.+            gs_energy: Exact diagonalization  round state energy+            parameters: Numpy array of shape (M,depth) where M is the required+             number of parameters per layer.+            path: Path to the data.+        """+        self.g = g+        self.res_e = res_e+        self.fidelity = fidelity+        self.gs_energy = gs_energy+        self.parameters = parameters+        self.path = path++    def __lt__(self, other):+        return self.g < other.g++    def hamiltonian(self):+        """Load the hamiltonian into a scipy sparse CSR matrix."""+        return sps.load_npz(self.path + '/H.npz')++    def ground_state(self):+        """Load the ground state wave function into a complex numpy array."""+        return np.load(self.path + '/groundstate.npy')+++def unique_name():+    """+    Generator to generate an infinite number of unique names.++    Yields:+        Strings of the form 'theta_<integer>'++    """+    num = 0+    while True:+        yield 'theta_' + str(num)+        num += 1+++class SpinSystemDataset(object):+    """Abstract class that handles a single instance of the system for a+    single set of order parameters.+    """++    def __init__(self, nspins: int, system_name: str):+        """Instantiate this data set.++        Args:+            nspins: The number of spins in the system.+            system_name: Name of the system we want to load.+        """++        # TODO: How should downloading and storing the data be handled?+        data_path = '/tmp/data/' + system_name + "_{}/".format(int(nspins))

Prefer tempfile.mkdtemp(), it works on all platforms, "/tmp/" doesn't work on windows, for example.

Also note that temp directories are often cleared by reboots. Fo it you're expecting these to be big downloads you may want an option to set the save directory to somewhere durable, and/or a print out telling the iser where the data is.

therooler

comment created time in a month

delete branch tensorflow/addons

delete branch : MarkDaoust-patch-1

delete time in a month

issue commenttensorflow/tensorflow

examples/get_started/regression/dnn_regression.py failed with ValueError (nightly build)

Thanks for closing.

Sorry about the confusion. I've deleted these example as they were no longer relevant.

wdirons

comment created time in a month

pull request commenttensorflow/addons

Add test to verify tutorial table of contents

We'd like to automate the generation of the tutorial TOC, but need to know where we can insert that so that docs team will pick it up.

build_docs.py is run when importing the "API rule", which is usually always run at the same times as the "narrative rule"

Exactly, it's two separate tasks: "build the api-reference", and "import the docs directory", so there isn't really any hook you can use here. Automating it would have to be done on our side.

Obviously it's not too complicated if we have a good way to get a title and choose an order (alphabetical?).

Subsites often forget to update their TOC. So we either need more testing or an auto-toc.

seanpmorgan

comment created time in a month

PR opened tensorflow/addons

Add title.

Add a title to average_optimizers_callback.ipynb

+3 -3

0 comment

1 changed file

pr created time in a month

create barnchtensorflow/addons

branch : MarkDaoust-patch-1

created branch time in a month

push eventtensorflow/examples

Ashar

commit sha d17191018ff36b06c36d7679d233e74e22ff3367

Fix loss typo (#209)

view details

push time in a month

PR merged tensorflow/examples

Reviewers
Fix loss typo cla: yes

The comment mentions binary cross-entropy loss whereas the loss being used is categorical cross entropy

+1 -1

5 comments

1 changed file

ashar-7

pr closed time in a month

pull request commenttensorflow/examples

Fix loss typo

Thanks!

ashar-7

comment created time in a month

issue closedtensorflow/tensorflow

Could we have more helpful error messages?

Description of issue (what needs changing):

Tensorflow gives many errors, and most of them aren't very helpful. Something like "module tensorflow has no attribute reset_graph." Can we change the error messages so they are more constructive? In this situation, the issue was partially solved by downgrading to tensorflow 1.12. It would be helpful if instead of the "reset_graph" error message, we could get a message more like: "this version of tensorflow is incompatible with the current project. Please downgrade to tensorflow 1.12 using: pip install tensorflow==1.12

Clear description

For example, why should someone use this method? How is it useful? To keep from tearing their own hair out.

closed time in a month

Tylersuard

issue commenttensorflow/tensorflow

Could we have more helpful error messages?

Hi @Tylersuard,

We understand that TensorFlow errors can be opaque at times. We are working on it, in general.

But this specific error is a hard one to fix. It's a python AttributeError, and there's no way to fix it without converting TensorFlow modules to some custom class with different errors depending on which non-existent attribute the user looks up... which would be messy and slow. So I don't think we can fix this.

We've been advertising for over a year that TensorFlow 2 would not be compatible with TensorFlow 1 code. It's something users need to be aware of, but there's no good way to fix this.

Tylersuard

comment created time in a month

issue commenttensorflow/tensorflow

Replacement for experimental_run_tf_function after removal from tf.keras.Model.compile

I'm not familiar with all, the issues here.

But, have you seen that train_step is now a overloadable in keras models: https://twitter.com/fchollet/status/1250622989541838848

That lets you affect how the model calculates the gradients. Does that help this use case?

tgaddair

comment created time in a month

pull request commenttensorflow/docs

Fix beginner's doc validation set issue

Can I delete this branch?

Yes, when a pr is closed you can do whatever you like with the branch.

oke-aditya

comment created time in a month

PR closed tensorflow/docs

Reviewers
Fix beginner's doc validation set issue cla: yes
  • This fixes the issue #38735.

File changed:- beginner.ipynb.

  • Replaced the validation set to test set. (This created the confusion).
  • Also updated the history object. So that in next tutorials people get comfortable with plotting it.
+10 -4

3 comments

1 changed file

oke-aditya

pr closed time in a month

pull request commenttensorflow/docs

Fix beginner's doc validation set issue

Hi @oke-aditya,

Thanks for taking the time to make a PR.

We're trying to keep this as short as possible, so I'd rather not mention features that aren't necessary (History).

(Also we try to avoid using "we" in these docs. )

For tensorflow/tensorflow#38735 I'd rather just say "validation set or test set": #1549.

Thanks again. I'm closing this, please don't take it the wrong way.

oke-aditya

comment created time in a month

PR opened tensorflow/docs

Reviewers
Or test set.

Simler fix for #38735.

Fixes tensorflow/tesorflow#38735 See also #1548

+1 -1

0 comment

1 changed file

pr created time in a month

create barnchtensorflow/docs

branch : MarkDaoust-patch-1

created branch time in a month

push eventoke-aditya/docs

Mark Daoust

commit sha 683c45dbf5a50ce785edb26441b5e87863041010

Update beginner.ipynb

view details

push time in a month

issue commenttensorflow/docs-l10n

[ko][zh-cn] How to get chinese/korean fonts to work in matplotlib + Colab?

@rickiepark

Haesun are you familiar with the required setup for Korean?

MarkDaoust

comment created time in a month

push eventMarkDaoust/hub

TensorFlow Hub Authors

commit sha 45c42c370d6f9b81fba9ddd2a3e7e7868f7bd3f6

Add BigGan and Image Classification to toc. PiperOrigin-RevId: 223985761

view details

TensorFlow Hub Authors

commit sha f9f69b43690b975622c714ab84bb0fcf426c9268

Move Community upper-tab to right. PiperOrigin-RevId: 224193154

view details

TensorFlow Hub Authors

commit sha e3cfa4ccb7efbff1f7844037502143b97b684192

Replace `tf.estimator.inputs` with `tf.compat.v1.estimator.inputs` PiperOrigin-RevId: 224856629

view details

TensorFlow Hub Authors

commit sha 6318be522c1b27a1fc538b0120548ba1d4e784fb

Remove outdated pip installs. PiperOrigin-RevId: 226230174

view details

TensorFlow Hub Authors

commit sha f174589de6dea7b461b91c5811d1964d03e09fa4

use packages@tensorflow.org as author email for python package PiperOrigin-RevId: 227688933

view details

TensorFlow Hub Authors

commit sha 2b0d428ac124a826edd2b061b264a5edcab6e723

Remove broken links to list of image modules. This fixes https://github.com/tensorflow/hub/issues/217 PiperOrigin-RevId: 229161101

view details

TensorFlow Hub Authors

commit sha b8c1f486d54313dba14f98d6d78e9fd152d8bdba

Changing modules link to tfhub.dev fixing links to available modules to point to https://tfhub.dev instead of the (now missing) static /docs/modules folder Original pull request 205 by kmader, edited by arnoegw. PiperOrigin-RevId: 229520683

view details

TensorFlow Hub Authors

commit sha 16a3e7ade4e1c69cd70878c43d32adee17cf3129

Split py_binary into py_binary and py_library to avoid having py_binary in deps. PiperOrigin-RevId: 230681447

view details

TensorFlow Hub Authors

commit sha dc53bdc31a0310623aa4871ca81206aade944c26

Add pointer to hub_with_keras doc PiperOrigin-RevId: 232336490

view details

TensorFlow Hub Authors

commit sha d9b489b2e696b84eaf7460c19c16bcd610a81e90

Switch flags and logging from tf with absl. PiperOrigin-RevId: 233558731

view details

TensorFlow Hub Authors

commit sha d9e6efb0e9fc3bc670df1427879e257504c5c599

Add native_module.get_unsupported_collections(). PiperOrigin-RevId: 233584508

view details

TensorFlow Hub Authors

commit sha 61113ef8361f7a9bd526dfbb1cd0f7a48dc2c8b5

Colab demonstrating basic usage of multilingual universal encoder (universal-encoder-xling-many) module. PiperOrigin-RevId: 233990556

view details

TensorFlow Hub Authors

commit sha f3e3811561beed6e7e4a18066932674aa88149a3

Enable using TF Hub library against TF 2.0 in V1 compat mode. PiperOrigin-RevId: 234569263

view details

TensorFlow Hub Authors

commit sha d3de17ac3f7f82336a3812f58729676a46be22d7

Implement hub.load() API for loading TF Hub modules in TF 2.0 PiperOrigin-RevId: 234950981

view details

TensorFlow Hub Authors

commit sha 66c202e4269b72f317f2a0632e2e51fac1e17517

Implementing a proper test for hub.load() and use load_v2() for version where the symbol is available. PiperOrigin-RevId: 235196181

view details

TensorFlow Hub Authors

commit sha 30346d1704a21463eea640e66edea9346607dc37

Add hub.KerasLayer for use with the future, TF2-compatible kind of Hub modules. PiperOrigin-RevId: 235517604

view details

TensorFlow Hub Authors

commit sha 655afe415391bd11ac353512098da5db1d9c6939

Add dev summit announcement banner to all projects. PiperOrigin-RevId: 236066165

view details

TensorFlow Hub Authors

commit sha 93c5ce3141031953cedc21513c64d2c4109a4e1e

Improve error messaging when loading V1 module throught V2 API. PiperOrigin-RevId: 236121366

view details

TensorFlow Hub Authors

commit sha fbbbb374e8359a74f75ec244814fdea100e2cfa7

Make the "output_shape" argument of KerasLayer optional. This argument exists to work around tf functionality and should eventually not be needed. Additionally having it with a default "None" limits the ability to in the future distinguish it from not provided and actually being None. PiperOrigin-RevId: 236127216

view details

junxnone

commit sha 835becd6749372138acce86bc2dc07bb81572e84

Add the png format support Signed-off-by: junxnone <junchen0102@gmail.com> PiperOrigin-RevId: 236286179

view details

push time in a month

issue openedtensorflow/docs-l10n

[ko][zh-cn] How to get chinese/korean fonts to work in matplot + Colab?

The default Matplotlib setup in Colab doesn't include Chinese or Korean fonts, so these characters don't render.

I believe this is one of the reasons we have not been translating figure text.

I can get the browser to render this text by outputting svg-text:

from IPython.display import set_matplotlib_formats
set_matplotlib_formats('svg')
matplotlib.rcParams['svg.fonttype'] = 'none'

But that messes up a lot of the formatting.

Some searching shows that it might just be a matter of installing the right fonts and adding them to the matplotlib.rc configuration, but I haven't found a setup that works yet.

Does anyone have experience setting this up?

created time in a month

pull request commenttensorflow/docs

Resolve minor typo and punctuation issue

Thanks.

jaketae

comment created time in a month

push eventjaketae/docs

Mark Daoust

commit sha a70508620e84ed4cecf759636d2d49569aa39555

Trim redundant comments.

view details

push time in a month

push eventMarkDaoust/docs

Mark Daoust

commit sha 701c9362c293fd438848e3559f40f28f80fcc553

test

view details

push time in a month

create barnchMarkDaoust/docs

branch : test-1

created branch time in a month

push eventMarkDaoust/docs

Rachin Kalakheti

commit sha 0be71f4da56a3760e40b4cad5c3cbf99c82dfc52

Added Tutorial on data augmentation using tf.image

view details

Rachin Kalakheti

commit sha 40f58d0aa0ebd384e5fb48b5c86719a128744c2d

Merge pull request #1 from akalakheti/akalakheti-data-augmentation Added Tutorial on data augmentation using tf.image

view details

Rachin Kalakheti

commit sha be6eed689f364e50a216c4dcba7f4d424037dffd

Delete Data_augmentation.ipynb

view details

Rachin Kalakheti

commit sha f154e78d724d055cf3d6553ffd7366a380f32cc2

Added Data_Augmentation.ipynb

view details

Billy Lamberta

commit sha a892fa0d897ec620e45a19e85ac180fd1724e46e

Lower case file names, step 1

view details

Billy Lamberta

commit sha f417522dc3928eec8c8e71954c05a8379634e95e

Lower case file names, step 2

view details

Billy Lamberta

commit sha 0559d1a1661db1515932262157cf29d5bd41cbd2

Initial formatting

view details

Mark Daoust

commit sha edb6351b6ab16b1f3b389e6f7cddf49348229db0

Switch image - It has a clear license - It download.tensorflow.org works with our test system

view details

Mark Daoust

commit sha c722144d6a30a2c6e4eac4b0c3a4e1f96e2fb36c

try %tensorflow_version

view details

Mark Daoust

commit sha 91bd2d02bc9926bd8d8e9252988d1a6141c04f48

fix json.

view details

Mark Daoust

commit sha a046c54774574ba442f202661bd74e10421b9d94

rotate

view details

Mark Daoust

commit sha 293078314338219182b1512485870c022e8f2cb3

Add toc entry, fit colab name

view details

Rachin Kalakheti

commit sha d2a1f54331267f3ecb427d81cbe9aed364628508

Update _toc.yaml

view details

Rachin Kalakheti

commit sha a588a90c060ba17c309d30b72de7d02c49b19fcb

Update data_augmentation.ipynb

view details

Rachin Kalakheti

commit sha b38713dc4256732fbf092d72e6b1fb632935a26e

Apply suggestions from code review Co-Authored-By: Billy Lamberta <blamb@google.com>

view details

Rachin Kalakheti

commit sha 48e5cf4f6b8c9eee94e8e09f915baed7380c2ad0

Updated colab link

view details

Rachin Kalakheti

commit sha 0fc38e1705b5882a56c24eb6fccda5ef8a6bb266

Suggested Changes done

view details

Rachin Kalakheti

commit sha 4d2f2f61cf18f6bff436dac4e2493ce7bef82fcd

Fixed JSON error

view details

Rachin Kalakheti

commit sha bfe426004bec7e8a7b1b8990eb138fb0ffcb499d

Fixed file name mistake

view details

Mark Daoust

commit sha f07bb0fa580237438ff8185da7f9d48407b9ee93

Skip extra whitespace. Tested: On TensorFlow. No unexpected changes. PiperOrigin-RevId: 297721712

view details

push time in a month

PR closed tensorflow/docs

Reviewers
Update word_embeddings.ipynb cla: yes
+2 -2

3 comments

1 changed file

Eshan-Agarwal

pr closed time in a month

pull request commenttensorflow/docs

Update word_embeddings.ipynb

Yeah, everything in compat.v1 is effectively deprecated too.

I'm planning to delete this cell after the 2.2 launch (any day now).

It looks like there's no longer a supported way to get the nested shape structure from a tf.data.Dataset. The closest thing I can find is: tf.nest.map_structure(lambda s:s.shape.as_list(), dataset.element_spec).

If there's a spot to mention that it's in the docs for padded batch.

Eshan-Agarwal

comment created time in a month

pull request commenttensorflow/docs

Add SequenceExample tutorial

Hi @ADBalici, I see we don't have any examples of this API.

It's always felt like an over-compilcated way to do this sort of thing, and sees very little usage. (50x less than tf.Example)

There is nothing incorrect about your sample code, but we probably don't want to give this so much visibility.

I would be more comfortable just having quick, concrete doctest-samples added injected into to the docstrings for SequenceExample, and the surrounding APIs.

Also note that if we do move forward with this it will be important to have code that shows how to load the SequenceExamples out of the TFRecord file.

@rohan100jain, WDYT about this?

ADBalici

comment created time in a month

PR closed tensorflow/tensorflow

Reviewers
Update api_def_Cast.pbtxt API review awaiting review cla: yes size:S
+9 -0

5 comments

1 changed file

rohanreddych

pr closed time in a month

pull request commenttensorflow/tensorflow

Update api_def_Cast.pbtxt

Same as #37605. Please re-open or make a new PR if you have bandwidth to fix this.

Thanks.

rohanreddych

comment created time in a month

pull request commenttensorflow/tensorflow

Update api_def_Div.pbtxt

The example is in 8582a58 is clear on how to do this.

@rohanreddych If you're still interested in fixing this follow that pattern please. Feel free to re-open this or make a new PR if you get back to it.

But let's not just leave this open.

rohanreddych

comment created time in a month

PR closed tensorflow/tensorflow

Reviewers
Update api_def_Div.pbtxt API review awaiting review cla: yes size:XS
+8 -0

7 comments

1 changed file

rohanreddych

pr closed time in a month

PR closed tensorflow/docs

Reviewers
Added an image augmentation tutorial using tf.image awaiting-community-review awaiting-editorial-review awaiting-technical-review cla: yes
+357 -0

4 comments

1 changed file

Jake-Short

pr closed time in a month

pull request commenttensorflow/docs

Added an image augmentation tutorial using tf.image

Hi @Jake-Short ,

Thanks for submitting this, it looks like the other implementation got merged, and we don't have room for two so I'm closing.

Jake-Short

comment created time in a month

PR closed MarkDaoust/docs

test
+2 -0

0 comment

1 changed file

MarkDaoust

pr closed time in a month

PR closed tensorflow/examples

Reviewers
Add softmax cla: no
+1 -1

8 comments

1 changed file

xsaq

pr closed time in a month

pull request commenttensorflow/examples

Add softmax

It looks like you're using a different email address for your commits than what you signed the CLA with, so we can't move forward with this PR, unless you fix it. ANd I'm requesting a different change, so I'll gust fo fix that.

xsaq

comment created time in a month

PR closed tensorflow/docs

Reviewers
R2.0 cla: yes

https://www.tensorflow.org/tutorials/keras/classification#verify_predictions seem to have two wrong pictures.

This part has 3 pictures. The second one should be red. Bottom left corner of the third should also be red.

+647350 -62

4 comments

3546 changed files

DrRyanHuang

pr closed time in a month

pull request commenttensorflow/docs

R2.0

It looks like something went wrong when putting this CL together.

Did you base it on the wrong branch or something?

I'm closing this one, please try again.

DrRyanHuang

comment created time in a month

pull request commenttensorflow/docs

Adding tutorial for Grad-CAM Heatmap Visualization

Hi @Qwerty71,

Thanks for the submission. It looks like we've got a similar tutorial submission in #1413, Please see the comments on there.

Maybe you and @Mbah-Javis can work together on finalizing a single tutorial? We don't have room for 2-heatmap vis tutorials, and I think you each cover points missed by the other? WDYT?

Qwerty71

comment created time in a month

Pull request review commenttensorflow/docs

Tutoruial on activation heatmaps with tf.keras

 {+ "cells": [+  {

Please revert this file:

git checkout master -- tools/templates/notebook.ipynb
git add tools/templates/notebook.ipynb
git commit -m "revert notebook template"
Mbah-Javis

comment created time in a month

issue closedtensorflow/docs-l10n

Several notebooks failing on `keras.experimental.export_saved_model`

https://github.com/tensorflow/docs-l10n/search?q=%22keras.experimental.export_saved_model%22&unscoped_q=%22keras.experimental.export_saved_model%22

# Exportar el modelo a 'SavedModel'
keras.experimental.export_saved_model(model, 'path_to_saved_model')

# Recrea exactamente el mismo modelo
new_model = keras.experimental.load_from_saved_model('path_to_saved_model')

These should just be model.save('path_to_saved_model'), and keras.models.load_model('path_to_saved_model').

closed time in 2 months

MarkDaoust

issue closedtensorflow/docs-l10n

Generate unexpected notebook cell in Japanese Documentation

I found issues about after convert to docs that generated extra notebook cell.

  • https://www.tensorflow.org/tutorials/keras/classification in Japanesse version

image

  • Original: notebook
  • https://github.com/masa-ita/tf-docs/blob/b8cd622e042ce322345d5fa30858087ee1abcab1/site/ja/tutorials/keras/basic_classification.ipynb

closed time in 2 months

hurutoriya

issue commenttensorflow/docs-l10n

Generate unexpected notebook cell in Japanese Documentation

But soon enough, Colab will default to TF 2 and we can remove these blocks throughout.

Done.

hurutoriya

comment created time in 2 months

issue commenttensorflow/docs-l10n

[ru] Failing notebooks.

TypeError: padded_batch() missing 1 required positional argument: 'padded_shapes'

Tis will be fixed by 2.2, so can maybe just be ignored for now.

MarkDaoust

comment created time in 2 months

more