Tensorboard Embedding Projection Example

How to make embeddings projection on TensorBoard

Install docker

Follow these steps

Install nvidia-container-toolkit or nvidia-docker2

Follow these steps

Install and run

The command below will build the docker image and run other installation steps

make install

Build docker image

This is already performed if you previously have run make install

docker build . -t mnist_projection

Enter container

docker run -it \
-v ${PWD}/projections:/projections/ \
-v ${PWD}/keras_datasets:/root/.keras/datasets \
-p 6006:6006 \
--rm --gpus all mnist_projection bash

Train model

python -m mnist_train \
    --output_dir /projections\
    --batch_size 16 \
    --epochs 5

Extract and Visualize Embeddings

python -m mnist_project_embeddings \
    --output_dir /projections/<timestamp>/ \
    --ckpt_path /projections/<timestamp>/model.hdf5 \
    --layer_name model_dense_1

Visualize embeddings

tensorboard --logdir /projections/<timestamp>/tensorboard/projector/ --port 6006

Enter localhost:6006 at your browser