Merge pull request #693 from GoogleCloudPlatform/keras3-dev

Rebasing fraudfinder onto Keras3-dev

Author: takumiohym

Makefile

@@ -18,7 +18,7 @@ kernels: \
  adk_kernel \
  object_detection_kernel \
  pytorch_kfp_kernel \
- lit_kernel
+ tf_privacy_kernel
 
 .PHONY: clean
 clean:
@@ -54,10 +54,9 @@ object_detection_kernel:
 pytorch_kfp_kernel:
 	./kernels/pytorch_kfp.sh
 
-.PHONY: lit_kernel
-lit_kernel:
-	./kernels/lit.sh
-
+.PHONY: tf_privacy_kernel
+tf_privacy_kernel:
+	./kernels/tf_privacy.sh
 
 .PHONY: tests
 tests:

kernels/lit.sh renamed to kernels/tf_privacy.sh

@@ -1,12 +1,12 @@
 #!/bin/bash
 #
-# To build the kernel:  ./kernels/lit.sh
-# To remove the kernel: ./kernels/lit.sh remove
+# To build the kernel:  ./kernels/tf_privacy.sh
+# To remove the kernel: ./kernels/tf_privacy.sh remove
 #
 # This scripts will create a ipython kernel named $ENVNAME
 
-ENVNAME=lit_kernel
-KERNEL_DISPLAY_NAME="LIT kernel"
+ENVNAME=tf_privacy_kernel
+KERNEL_DISPLAY_NAME="TF Privacy Kernel"
 
 source ~/.bashrc
 
@@ -24,11 +24,15 @@ conda activate $ENVNAME
 # Install packages using a pip local to the conda environment.
 conda install -q pip
 pip install -q ipykernel
+
 # Adds the conda kernel.
 DL_ANACONDA_ENV_HOME="${DL_ANACONDA_HOME}/envs/$ENVNAME"
 python -m ipykernel install --prefix "${DL_ANACONDA_ENV_HOME}" --name $ENVNAME --display-name "$KERNEL_DISPLAY_NAME"
 rm -rf "${DL_ANACONDA_ENV_HOME}/share/jupyter/kernels/python3"
 
-pip install -q tensorflow==2.14.1 lit-nlp keras_nlp
+# Install packages
+pip install tensorflow==2.14.0
+pip install --no-deps tensorflow-privacy==0.8.12 dp_accounting==0.4.3 tensorflow_probability==0.22.0
+pip install numpy==1.23.5 scipy~=1.9 dm-tree==0.1.8 attrs==25.3.0 cloudpickle==3.1.1 scikit-learn==1.*
 
 conda deactivate