Commits

Branch selector

 master

User selector

yuanliangzhe

Datepicker

All time

Commit History

Commits on Aug 3, 2022

• Internal change

  Show description for e800e56
  

  yuanliangzhe

  authored and

  tensorflower-gardener

  committed
  e800e56

Commits on Jul 19, 2022

• Register ViT architecture hyperparams.

  Show description for 7c917ef
  

  yuanliangzhe

  authored and

  tensorflower-gardener

  committed
  7c917ef

Commits on Jul 18, 2022

• Support 'none' classifier type to ViT. Also rename 'classifier' to 'pooler' for better naming.

  Show description for 866b476
  

  yuanliangzhe

  authored and

  tensorflower-gardener

  committed
  866b476

• Support ViT dynamic positional embedding interpolation. This is useful to finetune a model pre-trained on a different resolution.

  Show description for 27faf02
  

  yuanliangzhe

  authored and

  tensorflower-gardener

  committed
  27faf02

Commits on Jul 16, 2022

• Enable to construct decoder only in DETRTransformer and refactor DETR model. Also fix a typo in DetectionTask.

  Show description for 87495b0
  

  yuanliangzhe

  authored and

  tensorflower-gardener

  committed
  87495b0

Commits on Jul 12, 2022

• Add freeze_backbone flag for image_classification.

  Show description for 2068563
  

  yuanliangzhe

  authored and

  tensorflower-gardener

  committed
  2068563

Commits on May 25, 2022

• Add class index in the object detection evaluation logging info.

  Show description for 2bd76d1
  

  yuanliangzhe

  authored and

  TF Object Detection Team

  committed
  2bd76d1

Commits on May 23, 2022

• Add horizontal_flip_box in box_ops.

  Show description for 473f34a
  

  yuanliangzhe

  authored and

  tensorflower-gardener

  committed
  473f34a

Commits on Apr 28, 2022

• #movinet #vcc Add option global_average_pooling to Movinet. Allow model to output time-dim preserved features.

  Show description for 97b9757
  

  yuanliangzhe

  authored and

  tensorflower-gardener

  committed
  97b9757

Commits on Apr 27, 2022

• Propagate zero_centering_image flag in video_input.

  Show description for e75624b
  

  yuanliangzhe

  authored and

  tensorflower-gardener

  committed
  e75624b

Commits on Mar 23, 2022

• Refactor resnet_3d.

  Show description for 48c6be9
  

  yuanliangzhe

  authored and

  tensorflower-gardener

  committed
  48c6be9

Commits on Mar 16, 2022

• Start a project folder in preparing for ConST-CL open-source. (https://arxiv.org/abs/2112.05181)

  Show description for 77bf83b
  

  yuanliangzhe

  authored and

  tensorflower-gardener

  committed
  77bf83b

Commits on Feb 4, 2022

• #movinet Add tool to quantize movinet stream model. Optionally save representative dataset for post-training quantization.

  Show description for e5e8bf3
  

  yuanliangzhe

  authored and

  tensorflower-gardener

  committed
  e5e8bf3

Commits on Feb 1, 2022

• #movinet Explicitly add classifier activation option in yaml configs.

  Show description for d545771
  

  yuanliangzhe

  authored and

  tensorflower-gardener

  committed
  d545771

• #movinet Move export_saved_model.py to movinet/tools/ and add support for customized classifier activation.

  Show description for 685da94
  

  yuanliangzhe

  authored and

  tensorflower-gardener

  committed
  685da94

Commits on Jan 31, 2022

• #movinet Change 'reshape->conv->reshape->bn->activation' pattern to 'reshape->conv->bn->activation->reshape' in the MobileConv2D layer in movinet.

  Show description for 98d7c8e
  

  yuanliangzhe

  authored and

  tensorflower-gardener

  committed
  98d7c8e

Commits on Jan 28, 2022

• #movinet Support 'none' squeeze and excitation layers in Movinet.

  Show description for 441e5ae
  

  yuanliangzhe

  authored and

  tensorflower-gardener

  committed
  441e5ae

Commits on Jan 19, 2022

• Remove internal dependencies from OSS.

  Show description for f4c0fc2
  

  yuanliangzhe

  authored and

  tensorflower-gardener

  committed
  f4c0fc2

Commits on Jan 18, 2022

• internal change.

  Show description for f13be76
  

  yuanliangzhe

  authored and

  tensorflower-gardener

  committed
  f13be76

Commits on Jan 10, 2022

• Fix the audio dimensions for eval input.

  Show description for 1080693
  

  yuanliangzhe

  authored and

  tensorflower-gardener

  committed
  1080693

Commits on Jan 7, 2022

• #movinet Add se_type option in tools/convert_3d_2plus1d.py

  Show description for 98a558b
  

  yuanliangzhe

  authored and

  tensorflower-gardener

  committed
  98a558b

Commits on Dec 13, 2021

• Internal change

  Show description for fd431be
  

  yuanliangzhe

  authored and

  tensorflower-gardener

  committed
  fd431be

• #movinet Change the flag 'causal = True' by default in tools/convert_3d_2plus1d.py because in most cases the script is used for the stream models.

  Show description for 03847a7
  

  yuanliangzhe

  authored and

  tensorflower-gardener

  committed
  03847a7

• #movinet set NormActivation.activation = None by default and gauge its use case.

  Show description for dcde67c
  

  yuanliangzhe

  authored and

  tensorflower-gardener

  committed
  dcde67c

• #movinet Fix a typo.

  Show description for 8e8aac5
  

  yuanliangzhe

  authored and

  tensorflower-gardener

  committed
  8e8aac5

Commits on Dec 9, 2021

• #movinet Update two flags in readme demo script.

  Show description for d7858c2
  

  yuanliangzhe

  authored and

  tensorflower-gardener

  committed
  d7858c2

Commits on Jul 19, 2021

• Internal change

  Show description for f93bea8
  

  yuanliangzhe

  authored and

  tensorflower-gardener

  committed
  f93bea8

• Internal changes.

  Show description for 04c1e39
  

  yuanliangzhe

  authored and

  tensorflower-gardener

  committed
  04c1e39

Commits on Apr 14, 2021

• Fix the typo-ed name "remove_unecessary_ema" -> "remove_unnecessary_ema" and also re-implement it with deep dictionary copy.

  Show description for bf2d135
  

  yuanliangzhe

  authored and

  TF Object Detection Team

  committed
  bf2d135

Commits on Dec 6, 2020

• Internal change.

  Show description for 8ffa944
  

  yuanliangzhe

  authored and

  tensorflower-gardener

  committed
  8ffa944

Commits on Oct 23, 2020

• Fix a bug of slicing boxes in _postprocess_keypoints_for_class_and_image and switch to use python assert in _get_shape.

  Show description for f57fa41
  

  yuanliangzhe

  authored and

  TF Object Detection Team

  committed
  f57fa41

Commits on Sep 26, 2020

• Support to use separable_conv in CenterNet task head.

  Show description for 2ebe7c3
  

  yuanliangzhe

  authored and

  TF Object Detection Team

  committed
  2ebe7c3

Commits on Jun 11, 2020

• Make quantizable_separable_conv2d initializer explicitly configurable. (otherwise we could use slim.arg_scope to config the initializer) (tensorflow#8661)

  Show description for ffb5bb6
  yuanliangzhe

  authored
  ffb5bb6

Commits on Mar 3, 2020

• Merged commit includes the following changes: (tensorflow#8235)

  Show description for 4a0b3e4
  

  yuanliangzhe

  and

  dreamdragon

  authored
  4a0b3e4