experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Update quantization docs to use TFLiteConverter.from_saved_model() API instead of .from_keras_model() API

Browse Source 
PiperOrigin-RevId: 317251205
Change-Id: Ia8166decfa76327e3fd44871b194ffcae0f049f8
This commit is contained in:
Meghna Natraj 2020-06-18 22:33:44 -07:00 committed by TensorFlower Gardener
parent 158d4be42d
commit 539e9cb3a2
1 changed files with 4 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
tensorflow/lite/g3doc/performance/post_training_quantization.md
View File

@ -34,7 +34,7 @@ weights from floating point to integer, which has 8-bits of precision:
<pre>
import tensorflow as tf
converter = tf.lite.TFLiteConverter.from_keras_model(keras_model)
converter = tf.lite.TFLiteConverter.from_saved_model(saved_model_dir)
<b>converter.optimizations = [tf.lite.Optimize.DEFAULT]</b>
tflite_quant_model = converter.convert()
</pre>
@ -68,7 +68,7 @@ the following steps:
<pre>
import tensorflow as tf
converter = tf.lite.TFLiteConverter.from_keras_model(keras_model)
converter = tf.lite.TFLiteConverter.from_saved_model(saved_model_dir)
<b>converter.optimizations = [tf.lite.Optimize.DEFAULT]
def representative_dataset_gen():
for _ in range(num_calibration_steps):
@ -96,7 +96,7 @@ the following steps:
<pre>
import tensorflow as tf
converter = tf.lite.TFLiteConverter.from_keras_model(keras_model)
converter = tf.lite.TFLiteConverter.from_saved_model(saved_model_dir)
converter.optimizations = [tf.lite.Optimize.DEFAULT]
def representative_dataset_gen():
for _ in range(num_calibration_steps):
@ -120,7 +120,7 @@ quantization of weights, use the following steps:
<pre>
import tensorflow as tf
converter = tf.lite.TFLiteConverter.from_keras_model(keras_model)
converter = tf.lite.TFLiteConverter.from_saved_model(saved_model_dir)
<b>converter.optimizations = [tf.lite.Optimize.DEFAULT]
converter.target_spec.supported_types = [tf.float16]</b>
tflite_quant_model = converter.convert()