GitLab

PiperOrigin-RevId: 334062074

* factor make_divisible function and move round_filters to nn_layers

* modify SqueezeExcitation to add two additional parameter: divisible_by and gating_activation

* modify the InvertedBottleneckBlock to include 1. use_depthwise, 2. use_residual, 3. regularize_depthwise additional boolean flag; Add control for depthwise activation and regularizer; remove expand_ratio from SqueezeExcitation

* add Conv2DBNBlock definition

* add mobilenet v2, v3 implementation

* add mobilenet v1

* put mobilenet_base into class body

* fix a type hint error

* the invertedbottlenetblock is different for mobilenet and efficientnet. Made necessary changes to cope both.

* add target_backbone while call invertedbottleneckblock

* add relu6 and hard_sigmoid

* add test for mobilenet

* add mobilenet to factory

* fix some typo; link the reference to the architectures

* remove future import
Co-authored-by: Shixin Luo <luoshixin@google.com>

PiperOrigin-RevId: 333979065

PiperOrigin-RevId: 333949668

PiperOrigin-RevId: 333878584

PiperOrigin-RevId: 333840074

This is config is created for auto assignment issues for tf-dev-support

PiperOrigin-RevId: 333784610

PiperOrigin-RevId: 333772508

(object detection/keypoint prediction tasks) tf.lite compatible.

PiperOrigin-RevId: 333766281

PiperOrigin-RevId: 333644426

PiperOrigin-RevId: 333591143

Add norm_first to Transformer Scaffold; Add an option in gated_feedword to disable the output layer_norm.

PiperOrigin-RevId: 333591020

PiperOrigin-RevId: 333576935

ComputeMetrics - for ease of code reading.

PiperOrigin-RevId: 333575115

PiperOrigin-RevId: 333559599

PiperOrigin-RevId: 333314981

PiperOrigin-RevId: 333306806

PiperOrigin-RevId: 333183470

PiperOrigin-RevId: 333163906

PiperOrigin-RevId: 333019996

PiperOrigin-RevId: 333014322

PiperOrigin-RevId: 332992539

PiperOrigin-RevId: 332953126

PiperOrigin-RevId: 332918217

PiperOrigin-RevId: 332891157

PiperOrigin-RevId: 332806032

Remove return_all_encoder_outputs argument from keras_nlp.BertEncoder as it is unused and we will always return all encoder outputs in the returned dict.

PiperOrigin-RevId: 332802761

PiperOrigin-RevId: 332781597

Migrate all TF-NLP tasks to consume models with dictionary outputs and encoders with dictionary outputs.

PiperOrigin-RevId: 332740616

PiperOrigin-RevId: 332576047

PiperOrigin-RevId: 332557462

With kernel=[3, 3] in all entries of MOBILENETV1_CONV_DEFS, the latent index error (likely due to copy-paste) did not cause problems.
Fix is needed to ensure proper padding with non-square kernels.

Currently, a variable named base_name is not defined.
So, occur error while extract from downloaded model file.
Signed-off-by: MyungSung Kwak <yesmung@gmail.com>

The previous `base_url` led to 404 errors. This commit uses the correct base URL.