{"id":68749,"date":"2025-06-02T18:30:47","date_gmt":"2025-06-02T10:30:47","guid":{"rendered":"https:\/\/fwq.ai\/blog\/?p=68749"},"modified":"2025-06-02T18:30:47","modified_gmt":"2025-06-02T10:30:47","slug":"%e8%bd%bb%e9%87%8f%e7%ba%a7ai%e6%a8%a1%e5%9e%8b%e4%bc%98%e5%8c%96%e4%b8%8e%e9%83%a8%e7%bd%b2%e5%ae%9e%e8%b7%b5%ef%bc%9a%e8%be%b9%e7%bc%98%e8%ae%be%e5%a4%87%e4%b8%8a%e7%9a%84%e6%99%ba%e8%83%bd%e4%b9%8b","status":"publish","type":"post","link":"https:\/\/fwq.ai\/blog\/68749\/","title":{"rendered":"\u8f7b\u91cf\u7ea7AI\u6a21\u578b\u4f18\u5316\u4e0e\u90e8\u7f72\u5b9e\u8df5\uff1a\u8fb9\u7f18\u8bbe\u5907\u4e0a\u7684\u667a\u80fd\u4e4b\u9053"},"content":{"rendered":"<p>&nbsp;<\/p>\n<h2>\u524d\u8a00<\/h2>\n<p>\u968f\u7740AI\u6280\u672f\u5728\u8fb9\u7f18\u8ba1\u7b97\u9886\u57df\u7684\u5e7f\u6cdb\u5e94\u7528\uff0c\u8f7b\u91cf\u7ea7AI\u6a21\u578b\u7684\u4f18\u5316\u4e0e\u90e8\u7f72\u6b63\u9010\u6e10\u6210\u4e3a\u70ed\u95e8\u8bfe\u9898\u3002\u7279\u522b\u662f\u5728\u8ba1\u7b97\u8d44\u6e90\u53d7\u9650\u7684\u73af\u5883\u4e2d\uff0c\u5982\u4f55\u5728\u4fdd\u8bc1\u6027\u80fd\u7684\u540c\u65f6\u5b9e\u73b0\u9ad8\u6548\u63a8\u7406\uff0c\u662f\u6280\u672f\u4eba\u5458\u548c\u4f01\u4e1a\u7528\u6237\u5173\u6ce8\u7684\u7126\u70b9\u3002\u672c\u6587\u5c06\u6df1\u5165\u5256\u6790\u8f7b\u91cf\u7ea7AI\u6a21\u578b\u7684\u4f18\u5316\u65b9\u6cd5\u4e0e\u5b9e\u6218\u90e8\u7f72\u6848\u4f8b\uff0c\u5e76\u63a8\u8350\u9002\u5408\u8fd0\u884c\u8fd9\u4e9b\u4efb\u52a1\u7684\u9ad8\u6027\u80fd<strong>\u7f8e\u56fd\u670d\u52a1\u5668<\/strong>\u89e3\u51b3\u65b9\u6848\u3002<\/p>\n<h2>\u4e3a\u4ec0\u4e48\u9700\u8981\u8f7b\u91cf\u7ea7AI\u6a21\u578b<\/h2>\n<h3>\u8fb9\u7f18\u8ba1\u7b97\u7684\u73b0\u5b9e\u9700\u6c42<\/h3>\n<p>\u5728\u7269\u8054\u7f51\u8bbe\u5907\u3001\u5de5\u4e1a\u73b0\u573a\u63a7\u5236\u5668\u4ee5\u53ca\u5b89\u9632\u7ec8\u7aef\u4e2d\uff0cAI\u6a21\u578b\u7684\u90e8\u7f72\u9762\u4e34\u591a\u4e2a\u6311\u6218\u3002\u8fd9\u4e9b\u8bbe\u5907\u666e\u904d\u5b58\u5728\u5982\u4e0b\u9650\u5236\uff1a<\/p>\n<ul>\n<li><strong>\u8ba1\u7b97\u80fd\u529b\u6709\u9650<\/strong>\uff1a\u5982ARM Cortex-M MCU\uff0c\u4ec5\u5177\u5907\u57fa\u7840\u5904\u7406\u6027\u80fd\uff1b<\/li>\n<li><strong>\u5185\u5b58\u8d44\u6e90\u53d7\u9650<\/strong>\uff1aRAM\u5e38\u5728\u51e0\u767eKB\u5230\u6570MB\u4e4b\u95f4\uff1b<\/li>\n<li><strong>\u529f\u8017\u4e25\u683c\u63a7\u5236<\/strong>\uff1a\u9700\u6ee1\u8db3\u7535\u6c60\u7eed\u822a\u9700\u6c42\uff1b<\/li>\n<li><strong>\u5b9e\u65f6\u54cd\u5e94\u8981\u6c42\u9ad8<\/strong>\uff1a\u9002\u7528\u4e8e\u89c6\u9891\u76d1\u63a7\u3001\u673a\u5668\u4eba\u63a7\u5236\u7b49\u5b9e\u65f6\u5e94\u7528\u573a\u666f\u3002<\/li>\n<\/ul>\n<p>\u56e0\u6b64\uff0c\u4f20\u7edf\u7684AI\u6a21\u578b\u5f80\u5f80\u4e0d\u9002\u5408\u76f4\u63a5\u8fc1\u79fb\u5230\u6b64\u7c7b\u8bbe\u5907\u4e0a\uff0c\u9700\u8981\u4e13\u95e8\u7684\u4f18\u5316\u4e0e\u88c1\u526a\u3002<\/p>\n<h3>\u90e8\u7f72AI\u6a21\u578b\u7684\u6838\u5fc3\u74f6\u9888<\/h3>\n<p>\u5728\u53d7\u9650\u8bbe\u5907\u4e0a\u90e8\u7f72AI\u65f6\uff0c\u5f00\u53d1\u8005\u901a\u5e38\u9762\u4e34\u5982\u4e0b\u6311\u6218\uff1a<\/p>\n<ul>\n<li><strong>\u8fd0\u7b97\u74f6\u9888<\/strong>\uff1a\u7f3a\u5c11\u6d6e\u70b9\u5355\u5143\u6216AI\u52a0\u901f\u5668\uff1b<\/li>\n<li><strong>\u5185\u5b58\u4e0d\u8db3<\/strong>\uff1a\u6a21\u578b\u53c2\u6570\u96be\u4ee5\u5b8c\u5168\u8f7d\u5165\uff1b<\/li>\n<li><strong>\u7535\u6e90\u7ea6\u675f<\/strong>\uff1a\u5904\u7406\u5668\u9700\u7ef4\u6301\u4f4e\u529f\u8017\u8fd0\u884c\uff1b<\/li>\n<li><strong>\u7cfb\u7edf\u7a33\u5b9a\u6027<\/strong>\uff1a\u9700\u8981\u4fdd\u8bc1\u957f\u671f\u65e0\u5f02\u5e38\u8fd0\u884c\u3002<\/li>\n<\/ul>\n<p>\u6b64\u7c7b\u95ee\u9898\u4fc3\u4f7f\u6280\u672f\u4eba\u5458\u8f6c\u5411\u8f7b\u91cf\u7ea7AI\u6a21\u578b\u8bbe\u8ba1\uff0c\u5e76\u7ed3\u5408\u9002\u914d\u6027\u5f3a\u7684\u670d\u52a1\u5668\u8d44\u6e90\u8fdb\u884c\u63a8\u7406\u524d\u8bad\u7ec3\u548c\u6a21\u578b\u8f6c\u6362\uff0c\u63a8\u8350\u4f7f\u7528<strong>\u7f8e\u56fdvps<\/strong>\u8fdb\u884c\u6a21\u578b\u5f00\u53d1\u548c\u90e8\u7f72\u6d4b\u8bd5\u3002<\/p>\n<h2>\u8f7b\u91cf\u7ea7\u6a21\u578b\u4f18\u5316\u5173\u952e\u6280\u672f<\/h2>\n<p>\u4e3a\u4e86\u9002\u5e94\u8fb9\u7f18\u73af\u5883\uff0c\u4ee5\u4e0b\u4e09\u79cd\u4f18\u5316\u6280\u672f\u5c24\u4e3a\u5173\u952e\uff1a<\/p>\n<h3>1. \u91cf\u5316\u6280\u672f\uff08Quantization\uff09<\/h3>\n<p>\u5c06\u6d6e\u70b9\u6a21\u578b\u8f6c\u4e3a\u5b9a\u70b9\u683c\u5f0f\uff08\u5982INT8\u6216FP16\uff09\uff0c\u53ef\u663e\u8457\u964d\u4f4e\u6a21\u578b\u4f53\u79ef\u548c\u8fd0\u7b97\u8d1f\u8f7d\u3002\u4f8b\u5982\u4f7f\u7528TensorFlow Lite\u5b9e\u73b0\u5982\u4e0b\u4f18\u5316\uff1a<\/p>\n<pre><code class=\"language-python\">import tensorflow as tf\r\nmodel = tf.keras.models.load_model('model.h5')\r\nconverter = tf.lite.TFLiteConverter.from_keras_model(model)\r\nconverter.optimizations = [tf.lite.Optimize.DEFAULT]\r\nconverter.target_spec.supported_types = [tf.float16]\r\nquantized_model = converter.convert()\r\n<\/code><\/pre>\n<h3>2. \u526a\u679d\u6280\u672f\uff08Pruning\uff09<\/h3>\n<p>\u901a\u8fc7\u79fb\u9664\u795e\u7ecf\u7f51\u7edc\u4e2d\u7684\u5197\u4f59\u6743\u91cd\uff0c\u5b9e\u73b0\u6a21\u578b\u538b\u7f29\u3002\u4f8b\u5982\uff1a<\/p>\n<pre><code class=\"language-python\">import tensorflow_model_optimization as tfmot\r\npruning_params = {'pruning_schedule': tfmot.sparsity.keras.ConstantSparsity(0.5, 0)}\r\nmodel_for_pruning = tfmot.sparsity.keras.prune_low_magnitude(model, **pruning_params)\r\n<\/code><\/pre>\n<h3>3. \u77e5\u8bc6\u84b8\u998f\uff08Knowledge Distillation\uff09<\/h3>\n<p>\u901a\u8fc7\u8ba9\u8f7b\u91cf\u6a21\u578b\u201c\u6a21\u4eff\u201d\u590d\u6742\u6a21\u578b\u7684\u8f93\u51fa\u884c\u4e3a\uff0c\u5b9e\u73b0\u6027\u80fd\u8fc1\u79fb\u3002\u9002\u5408\u5728\u8bad\u7ec3\u9636\u6bb5\u4f7f\u7528PyTorch\u7b49\u6846\u67b6\u5b8c\u6210\u3002<\/p>\n<h2>\u5e94\u7528\u6848\u4f8b\u5206\u6790<\/h2>\n<h3>\u667a\u80fd\u6444\u50cf\u5934\u76ee\u6807\u68c0\u6d4b<\/h3>\n<p>\u5728ESP32-CAM\u8bbe\u5907\u4e2d\u90e8\u7f72MobileNetV2-SSD\u6a21\u578b\uff0c\u7ed3\u5408\u91cf\u5316\u4f18\u5316\u540e\u53ef\u5b9e\u73b0\u4eba\u8138\u8bc6\u522b\u4e0e\u5f02\u5e38\u884c\u4e3a\u68c0\u6d4b\u3002\u5728\u63a8\u7406\u90e8\u5206\uff0c\u4f7f\u7528C++\u4e0eTensorFlow Lite Micro\u5b9e\u73b0\uff1a<\/p>\n<pre><code class=\"language-cpp\">const tflite::Model* model = tflite::GetModel(model_tflite);\r\ntflite::MicroInterpreter* interpreter = ...\r\ninterpreter-&gt;Invoke();\r\n<\/code><\/pre>\n<p>\u6b64\u7c7b\u4efb\u52a1\u63a8\u8350\u4f7f\u7528<strong>\u7f8e\u56fd\u4e91\u670d\u52a1\u5668<\/strong>\u5b8c\u6210\u6a21\u578b\u8bad\u7ec3\u4e0e\u6d4b\u8bd5\u6784\u5efa\u8fc7\u7a0b\uff0c\u63d0\u4f9b\u7a33\u5b9a\u7684\u7b97\u529b\u4fdd\u969c\u3002<\/p>\n<h3>\u5de5\u4e1a\u9884\u6d4b\u6027\u7ef4\u62a4<\/h3>\n<p>\u9488\u5bf9\u5982STM32F4 MCU\u7684\u4f4e\u529f\u8017\u5de5\u4e1a\u63a7\u5236\u5668\uff0c\u53ef\u90e8\u7f72LSTM\u6a21\u578b\u7528\u4e8e\u8bbe\u5907\u5f02\u5e38\u68c0\u6d4b\u3002\u5728Python\u4e2d\u6784\u5efa\u4e0e\u4f18\u5316\u6d41\u7a0b\u5982\u4e0b\uff1a<\/p>\n<pre><code class=\"language-python\">model = tf.keras.Sequential([...])\r\nconverter = tf.lite.TFLiteConverter.from_keras_model(model)\r\ntflite_model = converter.convert()\r\n<\/code><\/pre>\n<p>\u501f\u52a9\u8fdc\u7a0b<strong>\u7f8e\u56fd\u670d\u52a1\u5668<\/strong>\u53ef\u52a0\u901f\u6a21\u578b\u8bad\u7ec3\u3001\u8fdc\u7a0b\u8c03\u8bd5\u4e0e\u56fa\u4ef6\u6253\u5305\u6d41\u7a0b\u3002<\/p>\n<h3>\u667a\u80fd\u5bb6\u5c45\u8bed\u97f3\u8bc6\u522b<\/h3>\n<p>\u5728ESP8266\u5e73\u53f0\u4e0a\u8fd0\u884c\u8f7b\u91cf\u5377\u79ef\u795e\u7ecf\u7f51\u7edc\uff08CNN\uff09\u6a21\u578b\uff0c\u7528\u4e8e\u5173\u952e\u8bcd\u8bc6\u522b\u3001\u8bed\u97f3\u63a7\u5236\u706f\u5149\u7b49\u529f\u80fd\u3002\u8fd9\u7c7b\u573a\u666f\u5bf9\u6a21\u578b\u5b9e\u65f6\u6027\u4e0e\u529f\u8017\u63d0\u51fa\u6781\u9ad8\u8981\u6c42\uff0c\u4f18\u5316\u540e\u7684\u6a21\u578b\u9700\u63a7\u5236\u5728\u51e0\u5341KB\u4ee5\u5185\u3002<\/p>\n<h2>\u7f8e\u56fd\u670d\u52a1\u5668\u52a9\u529b\u8fb9\u7f18AI\u90e8\u7f72<\/h2>\n<p>\u867d\u7136\u8fb9\u7f18\u8bbe\u5907\u8d1f\u8d23\u63a8\u7406\uff0c\u4f46\u6a21\u578b\u8bad\u7ec3\u3001\u8f6c\u6362\u3001\u6d4b\u8bd5\u548c\u7248\u672c\u7ba1\u7406\u7b49\u6838\u5fc3\u5de5\u4f5c\u4f9d\u8d56\u5f3a\u5927\u7684\u4e91\u7aef\u652f\u6301\u3002\u9009\u62e9\u9ad8\u6027\u80fd\u3001\u4f4e\u5ef6\u8fdf\u7684<strong>\u7f8e\u56fdvps<\/strong>\u4f5c\u4e3a\u5f00\u53d1\u5e73\u53f0\uff0c\u4e0d\u4ec5\u52a0\u901fAI\u6a21\u578b\u7684\u8fed\u4ee3\u6548\u7387\uff0c\u8fd8\u4fbf\u4e8e\u56e2\u961f\u8fdb\u884cCI\/CD\u6d41\u7a0b\u6574\u5408\uff0c\u7279\u522b\u9002\u5408\u9700\u8981\u5f02\u5730\u8fdc\u7a0b\u534f\u4f5c\u7684\u9879\u76ee\u3002<\/p>\n<hr \/>\n<h2>\u603b\u7ed3<\/h2>\n<p>\u8f7b\u91cf\u7ea7AI\u6a21\u578b\u6b63\u6210\u4e3a\u8fb9\u7f18\u8ba1\u7b97\u573a\u666f\u4e0b\u7684\u4e3b\u6d41\u89e3\u51b3\u65b9\u6848\u3002\u901a\u8fc7\u91cf\u5316\u3001\u526a\u679d\u3001\u77e5\u8bc6\u84b8\u998f\u7b49\u6280\u672f\uff0c\u5f00\u53d1\u8005\u53ef\u4ee5\u6709\u6548\u5e94\u5bf9\u7b97\u529b\u4e0e\u80fd\u8017\u7684\u53cc\u91cd\u6311\u6218\u3002\u800c\u501f\u52a9<a href=\"https:\/\/fwq.ai\/\"><strong>\u7f8e\u56fd\u4e91\u670d\u52a1\u5668<\/strong><\/a>\u8fdb\u884c\u6a21\u578b\u5f00\u53d1\u3001\u4f18\u5316\u4e0e\u90e8\u7f72\uff0c\u4e0d\u4ec5\u63d0\u5347\u6574\u4f53\u6548\u7387\uff0c\u4e5f\u80fd\u4e3a\u7cfb\u7edf\u7a33\u5b9a\u6027\u63d0\u4f9b\u4fdd\u969c\uff0c\u4e3a\u60a8\u7684AI\u9879\u76ee\u4fdd\u9a7e\u62a4\u822a\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"<p>&nbsp; \u524d\u8a00 \u968f\u7740AI\u6280\u672f\u5728\u8fb9\u7f18\u8ba1\u7b97\u9886\u57df\u7684\u5e7f\u6cdb\u5e94\u7528\uff0c\u8f7b\u91cf\u7ea7AI\u6a21\u578b\u7684\u4f18\u5316\u4e0e\u90e8\u7f72\u6b63\u9010\u6e10\u6210\u4e3a\u70ed\u95e8\u8bfe\u9898\u3002\u7279\u522b\u662f\u5728\u8ba1\u7b97\u8d44\u6e90\u53d7\u9650\u7684\u73af\u5883\u4e2d\uff0c\u5982\u4f55\u5728\u4fdd\u8bc1\u6027\u80fd\u7684\u540c\u65f6\u5b9e\u73b0\u9ad8\u6548\u63a8\u7406\uff0c\u662f\u6280\u672f\u4eba\u5458\u548c\u4f01\u4e1a\u7528\u6237\u5173\u6ce8\u7684\u7126\u70b9\u3002\u672c\u6587\u5c06\u6df1\u5165\u5256\u6790\u8f7b\u91cf\u7ea7AI\u6a21\u578b\u7684\u4f18\u5316\u65b9\u6cd5\u4e0e\u5b9e\u6218\u90e8\u7f72\u6848\u4f8b\uff0c\u5e76\u63a8\u8350\u9002\u5408\u8fd0\u884c\u8fd9\u4e9b\u4efb\u52a1\u7684\u9ad8\u6027\u80fd\u7f8e\u56fd\u670d\u52a1\u5668\u89e3\u51b3\u65b9\u6848\u3002 \u4e3a\u4ec0\u4e48\u9700\u8981\u8f7b\u91cf\u7ea7AI\u6a21\u578b \u8fb9\u7f18\u8ba1\u7b97\u7684\u73b0\u5b9e\u9700\u6c42 \u5728\u7269\u8054\u7f51\u8bbe\u5907\u3001\u5de5\u4e1a\u73b0\u573a\u63a7\u5236\u5668\u4ee5\u53ca\u5b89\u9632\u7ec8\u7aef\u4e2d\uff0cAI\u6a21\u578b\u7684\u90e8\u7f72\u9762\u4e34\u591a\u4e2a\u6311\u6218\u3002\u8fd9\u4e9b\u8bbe\u5907\u666e\u904d\u5b58\u5728\u5982\u4e0b\u9650\u5236\uff1a \u8ba1\u7b97\u80fd\u529b\u6709\u9650\uff1a\u5982ARM Cortex-M MCU\uff0c\u4ec5\u5177\u5907\u57fa\u7840\u5904\u7406\u6027\u80fd\uff1b \u5185\u5b58\u8d44\u6e90\u53d7\u9650\uff1aRAM\u5e38\u5728\u51e0\u767eKB\u5230\u6570MB\u4e4b\u95f4\uff1b \u529f\u8017\u4e25\u683c\u63a7\u5236\uff1a\u9700\u6ee1\u8db3\u7535\u6c60\u7eed\u822a\u9700\u6c42\uff1b \u5b9e\u65f6\u54cd\u5e94\u8981\u6c42\u9ad8\uff1a\u9002\u7528\u4e8e\u89c6\u9891\u76d1\u63a7\u3001\u673a\u5668\u4eba\u63a7\u5236\u7b49\u5b9e\u65f6\u5e94\u7528\u573a\u666f\u3002 \u56e0\u6b64\uff0c\u4f20\u7edf\u7684AI\u6a21\u578b\u5f80\u5f80\u4e0d\u9002\u5408\u76f4\u63a5\u8fc1\u79fb\u5230\u6b64\u7c7b\u8bbe\u5907\u4e0a\uff0c\u9700\u8981\u4e13\u95e8\u7684\u4f18\u5316\u4e0e\u88c1\u526a\u3002 \u90e8\u7f72AI\u6a21\u578b\u7684\u6838\u5fc3\u74f6\u9888 \u5728\u53d7\u9650\u8bbe\u5907\u4e0a\u90e8\u7f72AI\u65f6\uff0c\u5f00\u53d1\u8005\u901a\u5e38\u9762\u4e34\u5982\u4e0b\u6311\u6218\uff1a \u8fd0\u7b97\u74f6\u9888\uff1a\u7f3a\u5c11\u6d6e\u70b9\u5355\u5143\u6216AI\u52a0\u901f\u5668\uff1b \u5185\u5b58\u4e0d\u8db3\uff1a\u6a21\u578b\u53c2\u6570\u96be\u4ee5\u5b8c\u5168\u8f7d\u5165\uff1b \u7535\u6e90\u7ea6\u675f\uff1a\u5904\u7406\u5668\u9700\u7ef4\u6301\u4f4e\u529f\u8017\u8fd0\u884c\uff1b \u7cfb\u7edf\u7a33\u5b9a\u6027\uff1a\u9700\u8981\u4fdd\u8bc1\u957f\u671f\u65e0\u5f02\u5e38\u8fd0\u884c\u3002 \u6b64\u7c7b\u95ee\u9898\u4fc3\u4f7f\u6280\u672f\u4eba\u5458\u8f6c\u5411\u8f7b\u91cf\u7ea7AI\u6a21\u578b\u8bbe\u8ba1\uff0c\u5e76\u7ed3\u5408\u9002\u914d\u6027\u5f3a\u7684\u670d\u52a1\u5668\u8d44\u6e90\u8fdb\u884c\u63a8\u7406\u524d\u8bad\u7ec3\u548c\u6a21\u578b\u8f6c\u6362\uff0c\u63a8\u8350\u4f7f\u7528\u7f8e\u56fdvps\u8fdb\u884c\u6a21\u578b\u5f00\u53d1\u548c\u90e8\u7f72\u6d4b\u8bd5\u3002 \u8f7b\u91cf\u7ea7\u6a21\u578b\u4f18\u5316\u5173\u952e\u6280\u672f \u4e3a\u4e86\u9002\u5e94\u8fb9\u7f18\u73af\u5883\uff0c\u4ee5\u4e0b\u4e09\u79cd\u4f18\u5316\u6280\u672f\u5c24\u4e3a\u5173\u952e\uff1a 1. \u91cf\u5316\u6280\u672f\uff08Quantization\uff09 \u5c06\u6d6e\u70b9\u6a21\u578b\u8f6c\u4e3a\u5b9a\u70b9\u683c\u5f0f\uff08\u5982INT8\u6216FP16\uff09\uff0c\u53ef\u663e\u8457\u964d\u4f4e\u6a21\u578b\u4f53\u79ef\u548c\u8fd0\u7b97\u8d1f\u8f7d\u3002\u4f8b\u5982\u4f7f\u7528TensorFlow Lite\u5b9e\u73b0\u5982\u4e0b\u4f18\u5316\uff1a import tensorflow as tf model = tf.keras.models.load_model(&#8216;model.h5&#8217;) converter = tf.lite.TFLiteConverter.from_keras_model(model) converter.optimizations = [tf.lite.Optimize.DEFAULT] converter.target_spec.supported_types = [tf.float16] quantized_model = converter.convert() 2. \u526a\u679d\u6280\u672f\uff08Pruning\uff09 \u901a\u8fc7\u79fb\u9664\u795e\u7ecf\u7f51\u7edc\u4e2d\u7684\u5197\u4f59\u6743\u91cd\uff0c\u5b9e\u73b0\u6a21\u578b\u538b\u7f29\u3002\u4f8b\u5982\uff1a import tensorflow_model_optimization as tfmot pruning_params = {&#8216;pruning_schedule&#8217;: [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[13],"tags":[],"class_list":["post-68749","post","type-post","status-publish","format-standard","hentry","category-ai"],"_links":{"self":[{"href":"https:\/\/fwq.ai\/blog\/wp-json\/wp\/v2\/posts\/68749","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/fwq.ai\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/fwq.ai\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/fwq.ai\/blog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/fwq.ai\/blog\/wp-json\/wp\/v2\/comments?post=68749"}],"version-history":[{"count":1,"href":"https:\/\/fwq.ai\/blog\/wp-json\/wp\/v2\/posts\/68749\/revisions"}],"predecessor-version":[{"id":68750,"href":"https:\/\/fwq.ai\/blog\/wp-json\/wp\/v2\/posts\/68749\/revisions\/68750"}],"wp:attachment":[{"href":"https:\/\/fwq.ai\/blog\/wp-json\/wp\/v2\/media?parent=68749"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/fwq.ai\/blog\/wp-json\/wp\/v2\/categories?post=68749"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/fwq.ai\/blog\/wp-json\/wp\/v2\/tags?post=68749"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}