cherry-pick fix load inference in quant_post demo (#301) (#303)

demo/quant/quant_post/README.md

@@ -62,7 +62,7 @@ top1_acc/top5_acc= [0.70913923 0.89548034]
 使用以下命令测试离线量化后的模型的精度：
 
 ```
-python eval.py --model_path ./quant_model_train/MobileNet
+python eval.py --model_path ./quant_model_train/MobileNet --model_name __model__ --params_name __params__
 ```
 
 精度输出为

demo/quant/quant_post/image_classification_post_training_quantization_quick_start.ipynb

@@ -282,6 +282,8 @@
    "source": [
     "quant_post_prog, feed_target_names, fetch_targets = fluid.io.load_inference_model(\n",
     "        dirname='./quant_post_model',\n",
+    "        model_filename='__model__',\n",
+    "        params_filename='__params__',\n",
     "        executor=exe)\n",
     "test(quant_post_prog, fetch_targets)"
    ]

docs/en/quick_start/quant_post_tutorial_en.md

@@ -116,6 +116,8 @@ Load the model after post training quantization in ``'./quant_post_model'`` and
 ```python
 quant_post_prog, feed_target_names, fetch_targets = fluid.io.load_inference_model(
         dirname='./quant_post_model',
+        model_filename='__model__',
+        params_filename='__params__',
         executor=exe)
 test(quant_post_prog, fetch_targets)
 ```

docs/zh_cn/quick_start/quant_post_tutorial.md

@@ -123,6 +123,8 @@ slim.quant.quant_post(
 ```python
 quant_post_prog, feed_target_names, fetch_targets = fluid.io.load_inference_model(
         dirname='./quant_post_model',
+        model_filename='__model__',
+        params_filename='__params__',
         executor=exe)
 test(quant_post_prog, fetch_targets)
 ```