Statically and dynamically inspect tool for TensorFlow models
Switch branches/tags
Nothing to show
Clone or download



tfmodel is the command-line tool to inspect TensorFlow models for statically and dynamically metrics.

Read more documentation in


Install with pip.

pip install tensorflow-model

Install with docker.

docker run -it -v /:/host tobegit3hub/tfmodel bash

Install from source.

git clone

cd ./tfmodel

python ./ install


Validate Model

User can validate the model to check if is the deployable TensorFlow SavedModel.

tfmodel validate ./model

Inspect Model

User can inspect the model to check the signature before deploying.

tfmodel inspect ./model

Benchmark Model

User can run benchmark locally and get the performance report with only model files.

tfmodel benchmark ./model

TensorBoard Model

User can export the TensorBoard event files without re-training the models.

tfmodel tensorboard ./model