tensorflow/model-optimization

Issues

• Fail to fuse ReLU6 after QAT by using keras.layers.Activation('relu6')

  #1030 opened 2 years ago
  1

• 'Pruning in Keras' Example: Unable to fine-tune pruned model on GPU (TF 2.9.2)

  #1026 opened 2 years ago
  5

• Model Pruning with Yolo Object Detection Model

  #1018 opened 2 years ago
  1

• Pruning error for transfer learning models from TF/Keras API

  #1017 opened 2 years ago
  2

• Post training quantize TTS model fastspeech2 and failed

  #1010 opened 2 years ago
  5

• Input and resource quantization

  #1003 opened 2 years ago
  1

• Post-training integer quantization for SSDMobileNet - poor detection accuracy

  #1002 opened 2 years ago
  7

• bilinear upsampling layer cannot be 8-bit quantized correctly

  #1000 opened 2 years ago
  3

• about the Quantize layer when trans model

  #997 opened 2 years ago
  2

• QAT for subclass inside the subclass

  #996 opened 2 years ago
  2

• 1

  #995 opened 2 years ago
  0

• Allow quantization of tied weights

  #994 opened 2 years ago
  2

• Quantization aware training for a Transfer Learning MobileNet model

  #991 opened 2 years ago
  5

• Cannot joblib serialize pruned models

  #990 opened 2 years ago
  0

• Non-uniform sparsity layer-wise with PolynomialDecay scheduler

  #987 opened 2 years ago
  2

• Trying to quantise MobileNetv3 small Exception encountered when calling layer "tf.__operators__.add_137"

  #980 opened 3 years ago
  8

• PCQAT not working if Conv2D has kernel size 1x1

  #979 opened 3 years ago
  8

• How to extract original model's weight from trained QAT model?

  #977 opened 3 years ago
  9

• How to extract original model's weight from trained QAT model?

  #976 opened 3 years ago
  1

• Strange behaviour pruning with small datasets

  #975 opened 3 years ago
  4

• Full Int8 QAT not working

  #974 opened 3 years ago
  13

• Pruning only works for small batch sizes

  #973 opened 3 years ago
  4

• Yamnet clustering AttributeError: Exception encountered when calling layer "tf.__operators__.add" (type TFOpLambda).

  #972 opened 3 years ago
  1

• How to apply tfmot to a non tf.keras.Model type model..

  #968 opened 3 years ago
  4

• Unable to use strip_pruning for subclass model

  #965 opened 3 years ago
  7

• QAT model saving bug: Unable to save function b'__inference_separable_conv2d_layer_call_fn_961'

  #964 opened 3 years ago
  15

• Stripping Quantized Model

  #958 opened 3 years ago
  1

• Support Quantizing a tf.keras Model inside another tf.keras Model

  #957 opened 3 years ago
  0

• Quantize naive !!!

  #956 opened 3 years ago
  0

• Unable to prune/quantize multiple layers at the same time

  #955 opened 3 years ago
  0

• Unable to quantize to 4-bits

  #950 opened 3 years ago
  0

• prune_low_magnitude fail when wrapped with GRU layer, but success with conv2d and dense

  #944 opened 3 years ago
  0

• QAT support for LayerNormalization

  #942 opened 3 years ago
  0

• Unsupported operations when applying tfmot

  #941 opened 3 years ago
  1

• Not able to cluster Conv1DTranspose layer

  #940 opened 3 years ago
  3

• [clustering] Possible wrong call of the centroids initializer from the ClusterWeights wrapper

  #939 opened 3 years ago
  1

• LSTMPeephole package removed: Pypi not updated

  #938 opened 3 years ago
  0

• `prune_low_magnitude` can only prune an object of the following types: tf.keras.models.Sequential, tf.keras functional model, tf.keras.layers.Layer, list of tf.keras.layers.Layer. You passed an object of type: KerasTensor.

  #928 opened 3 years ago
  3

• Support for tensorflow hub layers

  #924 opened 3 years ago
  1

• GELU - Keras Layer

  #922 opened 3 years ago
  0

• Should the output of `Softmax` be quantized in the Default 8-bit Registry?

  #918 opened 3 years ago
  1

• TFLite: Maxpool layers (ops) don't behave as intended (VGG16 example)

  #911 opened 3 years ago
  5

• Learnable Scale Quantizer?

  #903 opened 3 years ago
  1

• Activity Regularizer not working with `tf.function` in TF2.7

  #902 opened 3 years ago
  1

• the API doc of prune_low_magnitude seems to be incomplete

  #901 opened 3 years ago
  1

• Model input names are not preserved during TFLite conversion when inference_input_type is tf.int8

  #889 opened 3 years ago
  4

• Pruning without training

  #888 opened 3 years ago
  2

• Weight miss-match when saving a QAT model with a particular layer pattern

  #887 opened 3 years ago
  3

• How to use default n bit in QAT ?

  #885 opened 3 years ago
  3

• QAT with trainable=False does not work as expected.

  #881 opened 3 years ago
  7