Tf2 compatibility #17
Closed
Conversation
This file contains 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
Add enum34 requirement version 0.1.1 PiperOrigin-RevId: 204213584
PiperOrigin-RevId: 204219076
PiperOrigin-RevId: 204380049
PiperOrigin-RevId: 204923387
Make zip explicitly into a list for compatibility with Python 3. PiperOrigin-RevId: 204927989
PiperOrigin-RevId: 204999443
PiperOrigin-RevId: 211175333
PiperOrigin-RevId: 214970812
PiperOrigin-RevId: 215710578
PiperOrigin-RevId: 218168789
This prevents breakage in an upcoming CL that moves the `text_summary` module and hides the PLUGIN_NAME constant. As a replacement, we use the `text_pb()` method from TensorBoard's own summary API that is the equivalent of the text summary op but generates the Summary proto directly without any op execution. The tag needs to be manually overridden because text_pb() currently adds a "/text_summary" suffix, but we're planning on removing that in the near future, so I have a TODO to myself to remove the override then. PiperOrigin-RevId: 219666504
Mainly, this removes reliance on the text_summary module. Once all TB changes are in external TF we can update to rely on the newer (cleaner) API for text pb generation. PiperOrigin-RevId: 220354498
As a first-time user of gin, it seemed like gin might automatically detect and parse my `config.gin` file. But it appears that we're actually required to call `gin.parse_config_file`, so I think it would help to explicitly include an example of that in the documentation of basic usage.
[docs] Include example of `gin.parse_config_file`
PiperOrigin-RevId: 220398335
PiperOrigin-RevId: 221546376
PiperOrigin-RevId: 221822490
PiperOrigin-RevId: 222579310
PiperOrigin-RevId: 224664959
This allows to use constants_from_enum in a similar way as external_configurable. PiperOrigin-RevId: 225354908
… read command-line parameters. PiperOrigin-RevId: 226252299
PiperOrigin-RevId: 227744657
PiperOrigin-RevId: 228899251
PiperOrigin-RevId: 230403051
PiperOrigin-RevId: 231480235
PiperOrigin-RevId: 232975361
PiperOrigin-RevId: 234173797
PiperOrigin-RevId: 234233017
PiperOrigin-RevId: 234611371
PiperOrigin-RevId: 234665504
|
Looks like you guys have got on top of this separately. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Attempt to make things tf 1.x/2.0 compatible as discussed here.
I've ported the original tests to run using 1.x and 2.0 (where possible), though the 2.0 external configurables/utils still need tests. I'm unsure how much of the new API gin wants to cover, whether it wants to all be under
kerasor not, or how they want to namespace v1 vs v2 etc.I've tried to mimic tensorflows
tf.compat.v1/v2setup, but maybe this makes things more ugly than necessary.There's some dirty monkey patching of
tf.ioto support 1.12 which can disappear if the version constraint is lifted.ConfigSaverHook has been re-implemented as a keras
Callback. Apart from that, this is mostly just a text substitution at this point - not much effort has been made to make it work "the tensorflow 2 way".