{
"cells": [
{
"cell_type": "markdown",
"metadata": {
"id": "Tce3stUlHN0L"
},
"source": [
"##### Copyright 2020 The TensorFlow Authors."
]
},
{
"cell_type": "code",
"execution_count": 1,
"metadata": {
"cellView": "form",
"execution": {
"iopub.execute_input": "2024-03-09T12:31:59.193368Z",
"iopub.status.busy": "2024-03-09T12:31:59.192676Z",
"iopub.status.idle": "2024-03-09T12:31:59.197157Z",
"shell.execute_reply": "2024-03-09T12:31:59.196389Z"
},
"id": "tuOe1ymfHZPu"
},
"outputs": [],
"source": [
"#@title Licensed under the Apache License, Version 2.0 (the \"License\");\n",
"# you may not use this file except in compliance with the License.\n",
"# You may obtain a copy of the License at\n",
"#\n",
"# https://www.apache.org/licenses/LICENSE-2.0\n",
"#\n",
"# Unless required by applicable law or agreed to in writing, software\n",
"# distributed under the License is distributed on an \"AS IS\" BASIS,\n",
"# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n",
"# See the License for the specific language governing permissions and\n",
"# limitations under the License."
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "qFdPvlXBOdUN"
},
"source": [
"# Quantization aware training in Keras example"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "MfBg1C5NB3X0"
},
"source": [
""
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "Bjmi3qZeu_xk"
},
"source": [
"## Overview\n",
"\n",
"Welcome to an end-to-end example for *quantization aware training*.\n",
"\n",
"### Other pages\n",
"For an introduction to what quantization aware training is and to determine if you should use it (including what's supported), see the [overview page](https://www.tensorflow.org/model_optimization/guide/quantization/training.md).\n",
"\n",
"To quickly find the APIs you need for your use case (beyond fully-quantizing a model with 8-bits), see the\n",
"[comprehensive guide](https://www.tensorflow.org/model_optimization/guide/quantization/training_comprehensive_guide.md).\n",
"\n",
"### Summary\n",
"\n",
"In this tutorial, you will:\n",
"\n",
"1. Train a `keras` model for MNIST from scratch.\n",
"2. Fine tune the model by applying the quantization aware training API, see the accuracy, and\n",
" export a quantization aware model.\n",
"3. Use the model to create an actually quantized model for the TFLite\n",
" backend.\n",
"4. See the persistence of accuracy in\n",
" TFLite and a 4x smaller model. To see the latency benefits on mobile, try out the TFLite examples [in the TFLite app repository](https://www.tensorflow.org/lite/models)."
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "yEAZYXvZU_XG"
},
"source": [
"## Setup"
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {
"execution": {
"iopub.execute_input": "2024-03-09T12:31:59.201392Z",
"iopub.status.busy": "2024-03-09T12:31:59.200848Z",
"iopub.status.idle": "2024-03-09T12:32:03.950622Z",
"shell.execute_reply": "2024-03-09T12:32:03.949464Z"
},
"id": "zN4yVFK5-0Bf"
},
"outputs": [],
"source": [
"! pip install -q tensorflow\n",
"! pip install -q tensorflow-model-optimization\n"
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {
"execution": {
"iopub.execute_input": "2024-03-09T12:32:03.955616Z",
"iopub.status.busy": "2024-03-09T12:32:03.954858Z",
"iopub.status.idle": "2024-03-09T12:32:06.870862Z",
"shell.execute_reply": "2024-03-09T12:32:06.869882Z"
},
"id": "yJwIonXEVJo6"
},
"outputs": [],
"source": [
"import tempfile\n",
"import os\n",
"\n",
"import tensorflow as tf\n",
"\n",
"from tensorflow_model_optimization.python.core.keras.compat import keras"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "psViY5PRDurp"
},
"source": [
"## Train a model for MNIST without quantization aware training"
]
},
{
"cell_type": "code",
"execution_count": 4,
"metadata": {
"execution": {
"iopub.execute_input": "2024-03-09T12:32:06.875930Z",
"iopub.status.busy": "2024-03-09T12:32:06.875058Z",
"iopub.status.idle": "2024-03-09T12:32:29.629450Z",
"shell.execute_reply": "2024-03-09T12:32:29.628395Z"
},
"id": "pbY-KGMPvbW9"
},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"2024-03-09 12:32:07.505187: E external/local_xla/xla/stream_executor/cuda/cuda_driver.cc:282] failed call to cuInit: CUDA_ERROR_NO_DEVICE: no CUDA-capable device is detected\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\r",
" 1/1688 [..............................] - ETA: 6:35:23 - loss: 2.2706 - accuracy: 0.0312"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 12/1688 [..............................] - ETA: 8s - loss: 2.1713 - accuracy: 0.2604 "
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 24/1688 [..............................] - ETA: 7s - loss: 2.0470 - accuracy: 0.3893"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 37/1688 [..............................] - ETA: 7s - loss: 1.8777 - accuracy: 0.4966"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 50/1688 [..............................] - ETA: 7s - loss: 1.6933 - accuracy: 0.5650"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 63/1688 [>.............................] - ETA: 6s - loss: 1.5353 - accuracy: 0.6146"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 76/1688 [>.............................] - ETA: 6s - loss: 1.3968 - accuracy: 0.6460"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 89/1688 [>.............................] - ETA: 6s - loss: 1.2811 - accuracy: 0.6766"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 102/1688 [>.............................] - ETA: 6s - loss: 1.1856 - accuracy: 0.6976"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 115/1688 [=>............................] - ETA: 6s - loss: 1.1141 - accuracy: 0.7128"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 128/1688 [=>............................] - ETA: 6s - loss: 1.0524 - accuracy: 0.7288"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 142/1688 [=>............................] - ETA: 6s - loss: 0.9920 - accuracy: 0.7434"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 155/1688 [=>............................] - ETA: 6s - loss: 0.9508 - accuracy: 0.7534"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 168/1688 [=>............................] - ETA: 6s - loss: 0.9071 - accuracy: 0.7645"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 181/1688 [==>...........................] - ETA: 6s - loss: 0.8690 - accuracy: 0.7735"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 194/1688 [==>...........................] - ETA: 6s - loss: 0.8369 - accuracy: 0.7811"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 207/1688 [==>...........................] - ETA: 5s - loss: 0.8076 - accuracy: 0.7874"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 221/1688 [==>...........................] - ETA: 5s - loss: 0.7765 - accuracy: 0.7943"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 235/1688 [===>..........................] - ETA: 5s - loss: 0.7552 - accuracy: 0.7975"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 249/1688 [===>..........................] - ETA: 5s - loss: 0.7323 - accuracy: 0.8026"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 263/1688 [===>..........................] - ETA: 5s - loss: 0.7151 - accuracy: 0.8064"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 276/1688 [===>..........................] - ETA: 5s - loss: 0.7019 - accuracy: 0.8096"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 289/1688 [====>.........................] - ETA: 5s - loss: 0.6864 - accuracy: 0.8130"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 302/1688 [====>.........................] - ETA: 5s - loss: 0.6757 - accuracy: 0.8154"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 316/1688 [====>.........................] - ETA: 5s - loss: 0.6610 - accuracy: 0.8187"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 330/1688 [====>.........................] - ETA: 5s - loss: 0.6494 - accuracy: 0.8214"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 343/1688 [=====>........................] - ETA: 5s - loss: 0.6413 - accuracy: 0.8230"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 357/1688 [=====>........................] - ETA: 5s - loss: 0.6301 - accuracy: 0.8258"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 371/1688 [=====>........................] - ETA: 5s - loss: 0.6166 - accuracy: 0.8293"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 384/1688 [=====>........................] - ETA: 5s - loss: 0.6101 - accuracy: 0.8312"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 397/1688 [======>.......................] - ETA: 5s - loss: 0.6002 - accuracy: 0.8333"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 410/1688 [======>.......................] - ETA: 5s - loss: 0.5912 - accuracy: 0.8354"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 423/1688 [======>.......................] - ETA: 4s - loss: 0.5820 - accuracy: 0.8377"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 436/1688 [======>.......................] - ETA: 4s - loss: 0.5750 - accuracy: 0.8399"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 449/1688 [======>.......................] - ETA: 4s - loss: 0.5675 - accuracy: 0.8421"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 463/1688 [=======>......................] - ETA: 4s - loss: 0.5595 - accuracy: 0.8437"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 476/1688 [=======>......................] - ETA: 4s - loss: 0.5519 - accuracy: 0.8460"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 490/1688 [=======>......................] - ETA: 4s - loss: 0.5458 - accuracy: 0.8477"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 504/1688 [=======>......................] - ETA: 4s - loss: 0.5383 - accuracy: 0.8504"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 518/1688 [========>.....................] - ETA: 4s - loss: 0.5332 - accuracy: 0.8514"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 532/1688 [========>.....................] - ETA: 4s - loss: 0.5260 - accuracy: 0.8534"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 546/1688 [========>.....................] - ETA: 4s - loss: 0.5200 - accuracy: 0.8551"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 560/1688 [========>.....................] - ETA: 4s - loss: 0.5142 - accuracy: 0.8565"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 573/1688 [=========>....................] - ETA: 4s - loss: 0.5088 - accuracy: 0.8580"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 586/1688 [=========>....................] - ETA: 4s - loss: 0.5029 - accuracy: 0.8597"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 600/1688 [=========>....................] - ETA: 4s - loss: 0.4966 - accuracy: 0.8609"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 613/1688 [=========>....................] - ETA: 4s - loss: 0.4918 - accuracy: 0.8619"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 627/1688 [==========>...................] - ETA: 4s - loss: 0.4863 - accuracy: 0.8631"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 641/1688 [==========>...................] - ETA: 4s - loss: 0.4813 - accuracy: 0.8642"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 655/1688 [==========>...................] - ETA: 4s - loss: 0.4774 - accuracy: 0.8652"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 668/1688 [==========>...................] - ETA: 3s - loss: 0.4750 - accuracy: 0.8660"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 681/1688 [===========>..................] - ETA: 3s - loss: 0.4713 - accuracy: 0.8671"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 694/1688 [===========>..................] - ETA: 3s - loss: 0.4671 - accuracy: 0.8686"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 707/1688 [===========>..................] - ETA: 3s - loss: 0.4635 - accuracy: 0.8697"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 720/1688 [===========>..................] - ETA: 3s - loss: 0.4595 - accuracy: 0.8708"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 733/1688 [============>.................] - ETA: 3s - loss: 0.4567 - accuracy: 0.8716"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 746/1688 [============>.................] - ETA: 3s - loss: 0.4543 - accuracy: 0.8724"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 759/1688 [============>.................] - ETA: 3s - loss: 0.4511 - accuracy: 0.8731"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 773/1688 [============>.................] - ETA: 3s - loss: 0.4488 - accuracy: 0.8737"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 787/1688 [============>.................] - ETA: 3s - loss: 0.4460 - accuracy: 0.8742"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 801/1688 [=============>................] - ETA: 3s - loss: 0.4416 - accuracy: 0.8755"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 815/1688 [=============>................] - ETA: 3s - loss: 0.4390 - accuracy: 0.8763"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 829/1688 [=============>................] - ETA: 3s - loss: 0.4355 - accuracy: 0.8773"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 843/1688 [=============>................] - ETA: 3s - loss: 0.4329 - accuracy: 0.8782"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 856/1688 [==============>...............] - ETA: 3s - loss: 0.4300 - accuracy: 0.8791"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 870/1688 [==============>...............] - ETA: 3s - loss: 0.4280 - accuracy: 0.8796"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 884/1688 [==============>...............] - ETA: 3s - loss: 0.4248 - accuracy: 0.8805"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 898/1688 [==============>...............] - ETA: 3s - loss: 0.4225 - accuracy: 0.8810"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 912/1688 [===============>..............] - ETA: 3s - loss: 0.4191 - accuracy: 0.8818"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 926/1688 [===============>..............] - ETA: 2s - loss: 0.4167 - accuracy: 0.8824"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 940/1688 [===============>..............] - ETA: 2s - loss: 0.4135 - accuracy: 0.8832"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 954/1688 [===============>..............] - ETA: 2s - loss: 0.4117 - accuracy: 0.8840"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 967/1688 [================>.............] - ETA: 2s - loss: 0.4095 - accuracy: 0.8846"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 981/1688 [================>.............] - ETA: 2s - loss: 0.4072 - accuracy: 0.8851"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
" 994/1688 [================>.............] - ETA: 2s - loss: 0.4049 - accuracy: 0.8857"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1008/1688 [================>.............] - ETA: 2s - loss: 0.4033 - accuracy: 0.8860"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1022/1688 [=================>............] - ETA: 2s - loss: 0.4017 - accuracy: 0.8864"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1036/1688 [=================>............] - ETA: 2s - loss: 0.3994 - accuracy: 0.8870"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1049/1688 [=================>............] - ETA: 2s - loss: 0.3970 - accuracy: 0.8877"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1063/1688 [=================>............] - ETA: 2s - loss: 0.3942 - accuracy: 0.8883"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1077/1688 [==================>...........] - ETA: 2s - loss: 0.3916 - accuracy: 0.8891"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1091/1688 [==================>...........] - ETA: 2s - loss: 0.3896 - accuracy: 0.8895"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1104/1688 [==================>...........] - ETA: 2s - loss: 0.3868 - accuracy: 0.8902"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1118/1688 [==================>...........] - ETA: 2s - loss: 0.3851 - accuracy: 0.8907"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1132/1688 [===================>..........] - ETA: 2s - loss: 0.3831 - accuracy: 0.8913"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1146/1688 [===================>..........] - ETA: 2s - loss: 0.3816 - accuracy: 0.8919"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1160/1688 [===================>..........] - ETA: 2s - loss: 0.3800 - accuracy: 0.8923"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1174/1688 [===================>..........] - ETA: 1s - loss: 0.3784 - accuracy: 0.8928"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1188/1688 [====================>.........] - ETA: 1s - loss: 0.3764 - accuracy: 0.8934"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1201/1688 [====================>.........] - ETA: 1s - loss: 0.3746 - accuracy: 0.8938"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1215/1688 [====================>.........] - ETA: 1s - loss: 0.3731 - accuracy: 0.8943"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1229/1688 [====================>.........] - ETA: 1s - loss: 0.3713 - accuracy: 0.8948"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1242/1688 [=====================>........] - ETA: 1s - loss: 0.3698 - accuracy: 0.8951"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1256/1688 [=====================>........] - ETA: 1s - loss: 0.3686 - accuracy: 0.8956"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1270/1688 [=====================>........] - ETA: 1s - loss: 0.3671 - accuracy: 0.8960"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1284/1688 [=====================>........] - ETA: 1s - loss: 0.3659 - accuracy: 0.8964"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1298/1688 [======================>.......] - ETA: 1s - loss: 0.3641 - accuracy: 0.8969"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1311/1688 [======================>.......] - ETA: 1s - loss: 0.3626 - accuracy: 0.8974"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1325/1688 [======================>.......] - ETA: 1s - loss: 0.3608 - accuracy: 0.8980"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1339/1688 [======================>.......] - ETA: 1s - loss: 0.3593 - accuracy: 0.8984"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1352/1688 [=======================>......] - ETA: 1s - loss: 0.3575 - accuracy: 0.8989"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1366/1688 [=======================>......] - ETA: 1s - loss: 0.3549 - accuracy: 0.8997"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1380/1688 [=======================>......] - ETA: 1s - loss: 0.3528 - accuracy: 0.9003"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1394/1688 [=======================>......] - ETA: 1s - loss: 0.3514 - accuracy: 0.9007"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1408/1688 [========================>.....] - ETA: 1s - loss: 0.3502 - accuracy: 0.9011"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1422/1688 [========================>.....] - ETA: 1s - loss: 0.3481 - accuracy: 0.9019"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1436/1688 [========================>.....] - ETA: 0s - loss: 0.3463 - accuracy: 0.9024"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1450/1688 [========================>.....] - ETA: 0s - loss: 0.3451 - accuracy: 0.9027"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1463/1688 [=========================>....] - ETA: 0s - loss: 0.3437 - accuracy: 0.9031"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1476/1688 [=========================>....] - ETA: 0s - loss: 0.3421 - accuracy: 0.9035"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1489/1688 [=========================>....] - ETA: 0s - loss: 0.3412 - accuracy: 0.9038"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1502/1688 [=========================>....] - ETA: 0s - loss: 0.3400 - accuracy: 0.9040"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1515/1688 [=========================>....] - ETA: 0s - loss: 0.3386 - accuracy: 0.9045"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1528/1688 [==========================>...] - ETA: 0s - loss: 0.3380 - accuracy: 0.9046"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1541/1688 [==========================>...] - ETA: 0s - loss: 0.3365 - accuracy: 0.9050"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1555/1688 [==========================>...] - ETA: 0s - loss: 0.3357 - accuracy: 0.9054"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1569/1688 [==========================>...] - ETA: 0s - loss: 0.3345 - accuracy: 0.9057"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1583/1688 [===========================>..] - ETA: 0s - loss: 0.3331 - accuracy: 0.9060"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1597/1688 [===========================>..] - ETA: 0s - loss: 0.3321 - accuracy: 0.9063"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1610/1688 [===========================>..] - ETA: 0s - loss: 0.3306 - accuracy: 0.9068"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1623/1688 [===========================>..] - ETA: 0s - loss: 0.3292 - accuracy: 0.9072"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1637/1688 [============================>.] - ETA: 0s - loss: 0.3280 - accuracy: 0.9075"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1651/1688 [============================>.] - ETA: 0s - loss: 0.3265 - accuracy: 0.9079"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1665/1688 [============================>.] - ETA: 0s - loss: 0.3251 - accuracy: 0.9083"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1679/1688 [============================>.] - ETA: 0s - loss: 0.3241 - accuracy: 0.9086"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"1688/1688 [==============================] - 21s 4ms/step - loss: 0.3232 - accuracy: 0.9089 - val_loss: 0.1358 - val_accuracy: 0.9618\n"
]
},
{
"data": {
"text/plain": [
""
]
},
"execution_count": 4,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# Load MNIST dataset\n",
"mnist = tf.keras.datasets.mnist\n",
"(train_images, train_labels), (test_images, test_labels) = mnist.load_data()\n",
"\n",
"# Normalize the input image so that each pixel value is between 0 to 1.\n",
"train_images = train_images / 255.0\n",
"test_images = test_images / 255.0\n",
"\n",
"# Define the model architecture.\n",
"model = keras.Sequential([\n",
" keras.layers.InputLayer(input_shape=(28, 28)),\n",
" keras.layers.Reshape(target_shape=(28, 28, 1)),\n",
" keras.layers.Conv2D(filters=12, kernel_size=(3, 3), activation='relu'),\n",
" keras.layers.MaxPooling2D(pool_size=(2, 2)),\n",
" keras.layers.Flatten(),\n",
" keras.layers.Dense(10)\n",
"])\n",
"\n",
"# Train the digit classification model\n",
"model.compile(optimizer='adam',\n",
" loss=keras.losses.SparseCategoricalCrossentropy(from_logits=True),\n",
" metrics=['accuracy'])\n",
"\n",
"model.fit(\n",
" train_images,\n",
" train_labels,\n",
" epochs=1,\n",
" validation_split=0.1,\n",
")"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "K8747K9OE72P"
},
"source": [
"## Clone and fine-tune pre-trained model with quantization aware training\n"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "F19k7ExXF_h2"
},
"source": [
"### Define the model"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "JsZROpNYMWQ0"
},
"source": [
"You will apply quantization aware training to the whole model and see this in the model summary. All layers are now prefixed by \"quant\".\n",
"\n",
"Note that the resulting model is quantization aware but not quantized (e.g. the weights are float32 instead of int8). The sections after show how to create a quantized model from the quantization aware one.\n",
"\n",
"In the [comprehensive guide](https://www.tensorflow.org/model_optimization/guide/quantization/training_comprehensive_guide.md), you can see how to quantize some layers for model accuracy improvements."
]
},
{
"cell_type": "code",
"execution_count": 5,
"metadata": {
"execution": {
"iopub.execute_input": "2024-03-09T12:32:29.634089Z",
"iopub.status.busy": "2024-03-09T12:32:29.633282Z",
"iopub.status.idle": "2024-03-09T12:32:31.854178Z",
"shell.execute_reply": "2024-03-09T12:32:31.853416Z"
},
"id": "oq6blGjgFDCW"
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Model: \"sequential\"\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"_________________________________________________________________\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
" Layer (type) Output Shape Param # \n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"=================================================================\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
" quantize_layer (QuantizeLa (None, 28, 28) 3 \n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
" yer) \n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
" \n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
" quant_reshape (QuantizeWra (None, 28, 28, 1) 1 \n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
" pperV2) \n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
" \n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
" quant_conv2d (QuantizeWrap (None, 26, 26, 12) 147 \n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
" perV2) \n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
" \n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
" quant_max_pooling2d (Quant (None, 13, 13, 12) 1 \n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
" izeWrapperV2) \n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
" \n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
" quant_flatten (QuantizeWra (None, 2028) 1 \n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
" pperV2) \n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
" \n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
" quant_dense (QuantizeWrapp (None, 10) 20295 \n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
" erV2) \n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
" \n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"=================================================================\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Total params: 20448 (79.88 KB)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Trainable params: 20410 (79.73 KB)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Non-trainable params: 38 (152.00 Byte)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"_________________________________________________________________\n"
]
}
],
"source": [
"import tensorflow_model_optimization as tfmot\n",
"\n",
"quantize_model = tfmot.quantization.keras.quantize_model\n",
"\n",
"# q_aware stands for for quantization aware.\n",
"q_aware_model = quantize_model(model)\n",
"\n",
"# `quantize_model` requires a recompile.\n",
"q_aware_model.compile(optimizer='adam',\n",
" loss=keras.losses.SparseCategoricalCrossentropy(from_logits=True),\n",
" metrics=['accuracy'])\n",
"\n",
"q_aware_model.summary()"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "uDr2ijwpGCI-"
},
"source": [
"### Train and evaluate the model against baseline"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "XUBEn94hXYB1"
},
"source": [
"To demonstrate fine tuning after training the model for just an epoch, fine tune with quantization aware training on a subset of the training data."
]
},
{
"cell_type": "code",
"execution_count": 6,
"metadata": {
"execution": {
"iopub.execute_input": "2024-03-09T12:32:31.865418Z",
"iopub.status.busy": "2024-03-09T12:32:31.864822Z",
"iopub.status.idle": "2024-03-09T12:32:34.082690Z",
"shell.execute_reply": "2024-03-09T12:32:34.081881Z"
},
"id": "_PHDGJryE31X"
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"\r",
"1/2 [==============>...............] - ETA: 1s - loss: 0.1939 - accuracy: 0.9500"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
"2/2 [==============================] - 2s 480ms/step - loss: 0.1749 - accuracy: 0.9500 - val_loss: 0.1863 - val_accuracy: 0.9700\n"
]
},
{
"data": {
"text/plain": [
""
]
},
"execution_count": 6,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"train_images_subset = train_images[0:1000] # out of 60000\n",
"train_labels_subset = train_labels[0:1000]\n",
"\n",
"q_aware_model.fit(train_images_subset, train_labels_subset,\n",
" batch_size=500, epochs=1, validation_split=0.1)"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "-byC2lYlMkfN"
},
"source": [
"For this example, there is minimal to no loss in test accuracy after quantization aware training, compared to the baseline."
]
},
{
"cell_type": "code",
"execution_count": 7,
"metadata": {
"execution": {
"iopub.execute_input": "2024-03-09T12:32:34.086899Z",
"iopub.status.busy": "2024-03-09T12:32:34.086145Z",
"iopub.status.idle": "2024-03-09T12:32:35.485822Z",
"shell.execute_reply": "2024-03-09T12:32:35.485046Z"
},
"id": "6bMFTKSSHyyZ"
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Baseline test accuracy: 0.9550999999046326\n",
"Quant test accuracy: 0.9584000110626221\n"
]
}
],
"source": [
"_, baseline_model_accuracy = model.evaluate(\n",
" test_images, test_labels, verbose=0)\n",
"\n",
"_, q_aware_model_accuracy = q_aware_model.evaluate(\n",
" test_images, test_labels, verbose=0)\n",
"\n",
"print('Baseline test accuracy:', baseline_model_accuracy)\n",
"print('Quant test accuracy:', q_aware_model_accuracy)"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "2IepmUPSITn6"
},
"source": [
"## Create quantized model for TFLite backend"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "1FgNP4rbOLH8"
},
"source": [
"After this, you have an actually quantized model with int8 weights and uint8 activations."
]
},
{
"cell_type": "code",
"execution_count": 8,
"metadata": {
"execution": {
"iopub.execute_input": "2024-03-09T12:32:35.490094Z",
"iopub.status.busy": "2024-03-09T12:32:35.489388Z",
"iopub.status.idle": "2024-03-09T12:32:37.835766Z",
"shell.execute_reply": "2024-03-09T12:32:37.834624Z"
},
"id": "w7fztWsAOHTz"
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"INFO:tensorflow:Assets written to: /tmpfs/tmp/tmp3vwylslo/assets\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"INFO:tensorflow:Assets written to: /tmpfs/tmp/tmp3vwylslo/assets\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/lite/python/convert.py:964: UserWarning: Statistics for quantized inputs were expected, but not specified; continuing anyway.\n",
" warnings.warn(\n",
"WARNING: All log messages before absl::InitializeLog() is called are written to STDERR\n",
"W0000 00:00:1709987557.593385 26473 tf_tfl_flatbuffer_helpers.cc:390] Ignored output_format.\n",
"W0000 00:00:1709987557.593449 26473 tf_tfl_flatbuffer_helpers.cc:393] Ignored drop_control_dependency.\n"
]
}
],
"source": [
"converter = tf.lite.TFLiteConverter.from_keras_model(q_aware_model)\n",
"converter.optimizations = [tf.lite.Optimize.DEFAULT]\n",
"\n",
"quantized_tflite_model = converter.convert()"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "BEYsyYVqNgeY"
},
"source": [
"## See persistence of accuracy from TF to TFLite"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "saadXD4JQsBK"
},
"source": [
"Define a helper function to evaluate the TF Lite model on the test dataset."
]
},
{
"cell_type": "code",
"execution_count": 9,
"metadata": {
"execution": {
"iopub.execute_input": "2024-03-09T12:32:37.840381Z",
"iopub.status.busy": "2024-03-09T12:32:37.839618Z",
"iopub.status.idle": "2024-03-09T12:32:37.846491Z",
"shell.execute_reply": "2024-03-09T12:32:37.845754Z"
},
"id": "b8yBouuGNqls"
},
"outputs": [],
"source": [
"import numpy as np\n",
"\n",
"def evaluate_model(interpreter):\n",
" input_index = interpreter.get_input_details()[0][\"index\"]\n",
" output_index = interpreter.get_output_details()[0][\"index\"]\n",
"\n",
" # Run predictions on every image in the \"test\" dataset.\n",
" prediction_digits = []\n",
" for i, test_image in enumerate(test_images):\n",
" if i % 1000 == 0:\n",
" print('Evaluated on {n} results so far.'.format(n=i))\n",
" # Pre-processing: add batch dimension and convert to float32 to match with\n",
" # the model's input data format.\n",
" test_image = np.expand_dims(test_image, axis=0).astype(np.float32)\n",
" interpreter.set_tensor(input_index, test_image)\n",
"\n",
" # Run inference.\n",
" interpreter.invoke()\n",
"\n",
" # Post-processing: remove batch dimension and find the digit with highest\n",
" # probability.\n",
" output = interpreter.tensor(output_index)\n",
" digit = np.argmax(output()[0])\n",
" prediction_digits.append(digit)\n",
"\n",
" print('\\n')\n",
" # Compare prediction results with ground truth labels to calculate accuracy.\n",
" prediction_digits = np.array(prediction_digits)\n",
" accuracy = (prediction_digits == test_labels).mean()\n",
" return accuracy"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "TuEFS4CIQvUw"
},
"source": [
"You evaluate the quantized model and see that the accuracy from TensorFlow persists to the TFLite backend."
]
},
{
"cell_type": "code",
"execution_count": 10,
"metadata": {
"execution": {
"iopub.execute_input": "2024-03-09T12:32:37.850006Z",
"iopub.status.busy": "2024-03-09T12:32:37.849531Z",
"iopub.status.idle": "2024-03-09T12:32:38.600645Z",
"shell.execute_reply": "2024-03-09T12:32:38.599821Z"
},
"id": "VqQTyqz4NsWd"
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Evaluated on 0 results so far.\n",
"Evaluated on 1000 results so far.\n",
"Evaluated on 2000 results so far.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"INFO: Created TensorFlow Lite XNNPACK delegate for CPU.\n",
"WARNING: Attempting to use a delegate that only supports static-sized tensors with a graph that has dynamic-sized tensors (tensor#12 is a dynamic-sized tensor).\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Evaluated on 3000 results so far.\n",
"Evaluated on 4000 results so far.\n",
"Evaluated on 5000 results so far.\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Evaluated on 6000 results so far.\n",
"Evaluated on 7000 results so far.\n",
"Evaluated on 8000 results so far.\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Evaluated on 9000 results so far.\n",
"\n",
"\n",
"Quant TFLite test_accuracy: 0.9584\n",
"Quant TF test accuracy: 0.9584000110626221\n"
]
}
],
"source": [
"interpreter = tf.lite.Interpreter(model_content=quantized_tflite_model)\n",
"interpreter.allocate_tensors()\n",
"\n",
"test_accuracy = evaluate_model(interpreter)\n",
"\n",
"print('Quant TFLite test_accuracy:', test_accuracy)\n",
"print('Quant TF test accuracy:', q_aware_model_accuracy)"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "z8D7WnFF5DZR"
},
"source": [
"## See 4x smaller model from quantization"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "I1c2IecBRCdQ"
},
"source": [
"You create a float TFLite model and then see that the quantized TFLite model\n",
"is 4x smaller."
]
},
{
"cell_type": "code",
"execution_count": 11,
"metadata": {
"execution": {
"iopub.execute_input": "2024-03-09T12:32:38.604480Z",
"iopub.status.busy": "2024-03-09T12:32:38.604162Z",
"iopub.status.idle": "2024-03-09T12:32:39.608037Z",
"shell.execute_reply": "2024-03-09T12:32:39.607219Z"
},
"id": "jy_Lgfh8VkyX"
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"INFO:tensorflow:Assets written to: /tmpfs/tmp/tmppztp93bk/assets\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"INFO:tensorflow:Assets written to: /tmpfs/tmp/tmppztp93bk/assets\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Float model in Mb: 0.08068466186523438\n",
"Quantized model in Mb: 0.0236053466796875\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"W0000 00:00:1709987559.464849 26473 tf_tfl_flatbuffer_helpers.cc:390] Ignored output_format.\n",
"W0000 00:00:1709987559.464884 26473 tf_tfl_flatbuffer_helpers.cc:393] Ignored drop_control_dependency.\n"
]
}
],
"source": [
"# Create float TFLite model.\n",
"float_converter = tf.lite.TFLiteConverter.from_keras_model(model)\n",
"float_tflite_model = float_converter.convert()\n",
"\n",
"# Measure sizes of models.\n",
"_, float_file = tempfile.mkstemp('.tflite')\n",
"_, quant_file = tempfile.mkstemp('.tflite')\n",
"\n",
"with open(quant_file, 'wb') as f:\n",
" f.write(quantized_tflite_model)\n",
"\n",
"with open(float_file, 'wb') as f:\n",
" f.write(float_tflite_model)\n",
"\n",
"print(\"Float model in Mb:\", os.path.getsize(float_file) / float(2**20))\n",
"print(\"Quantized model in Mb:\", os.path.getsize(quant_file) / float(2**20))"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "0O5xuci-SonI"
},
"source": [
"## Conclusion"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "O2I7xmyMW5QY"
},
"source": [
"In this tutorial, you saw how to create quantization aware models with the TensorFlow Model Optimization Toolkit API and then quantized models for the TFLite backend.\n",
"\n",
"You saw a 4x model size compression benefit for a model for MNIST, with minimal accuracy\n",
"difference. To see the latency benefits on mobile, try out the TFLite examples [in the TFLite app repository](https://www.tensorflow.org/lite/models).\n",
"\n",
"We encourage you to try this new capability, which can be particularly important for deployment in resource-constrained environments. \n"
]
}
],
"metadata": {
"accelerator": "GPU",
"colab": {
"collapsed_sections": [],
"name": "training_example.ipynb",
"toc_visible": true
},
"kernelspec": {
"display_name": "Python 3",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.9.18"
}
},
"nbformat": 4,
"nbformat_minor": 0
}
![](\"https://www.tensorflow.org/images/tf_logo_32px.png\"){kind=logo}
View on TensorFlow.org\n",
" ![](\"https://www.tensorflow.org/images/colab_logo_32px.png\"){kind=logo}
Run in Google Colab\n",
" ![](\"https://www.tensorflow.org/images/GitHub-Mark-32px.png\")
View source on GitHub\n",
" ![](\"https://www.tensorflow.org/images/download_logo_32px.png\"){kind=logo}
Download notebook\n",
"