keras-surgeon
keras-surgeon copied to clipboard
BenWhetton
Pruning and other network surgery for trained Keras models.
Hello I have been using this library for a bit and have a slightly edited library that works for at least tensorflow 2.5 https://github.com/Woodyet/Pix2Pix_Auto_Prune/tree/master/Pix2Pix_Auto_Prune/Tensorflow2.5.0/kerassurgeon Most of the updates are lifted...
When running inside of tensorflow 2.4, I am getting the following error: `TypeError: Keras symbolic inputs/outputs do not implement __len__. You may be trying to pass Keras symbolic inputs/outputs to...
when I deal with my conv1D not 2D netwrok it report this mistake: "AttributeError: 'NoneType' object has no attribute 'op'"
Once a model contains BatchNorm layers, keras-surgeon fails. ### Minimal code to reproduce the error: ```python3 #%% import tensorflow as tf from kerassurgeon.operations import delete_channels model = tf.keras.Sequential([ tf.keras.layers.Conv2D(16, kernel_size=(3,3),...
Version 0.1.3 which was just released should fix the "infinite loop" repeatable bug. There are definitely opportunities to improve the memory usage but I don't think this ticket needs to...
model_new_4 = delete_layer(model, layer_6) ValueError: Input 0 is incompatible with layer dense_5: expected axis -1 of input shape to have value 18496 but got shape (None, 10368)
ValueError: "SeparableConv2D" layers are currently unsupported. How can i implemented the channels delete of "SeparableConv2D" layers?
Group Normalization is a normalization layer that divides channels into groups and normalizes the features within each group. It is very useful when batch size is very small. Use "pip...
I am not sure if I am using this correctly but I am trying to experiment with Kerassurgeon to delete and add layers. Currently me and my team are struggling...
Hi, I tried manual filter pruning with Keras and after pruning there is a drastic change in the accuracy of the model when evaluated. Later to validate the model i...
Metadata
Owner
Metadata
Pruning and other network surgery for trained Keras models.