Skip to content

Instantly share code, notes, and snippets.

@iamtimdavis

iamtimdavis/tf_lite_command_line

Last active September 16, 2019 04:58
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save iamtimdavis/43f06a12d7fe3a6075c2607a7f84f370 to your computer and use it in GitHub Desktop.
Save iamtimdavis/43f06a12d7fe3a6075c2607a7f84f370 to your computer and use it in GitHub Desktop.
Download ZIP
Enabling post-training quantization
Raw
tf_lite_command_line
converter=tf.contrib.lite.TocoConverter.from_saved_model(saved_model_dir)
converter.post_training_quantize=True
tflite_quantized_model=converter.convert()
open(“quantized_model.tflite”, “wb”).write(tflite_quantized_model)
@Akhtar303nu
Copy link

Hi I am trying same code but its genrate error
I saved_model_dir which contain mars-small128.pb but it genrate error
OSError: SavedModel file does not exist at: model_data/saved_model.pb/{saved_model.pbtxt|saved_model.pb}
Thanks

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment