diff --git a/algo/algo/index.html b/algo/algo/index.html index 220048d91ebcbb0b58e6332ffa6e7287bbfdacc7..ac7f16636385953f97943e5e62ac11309e1f2a5a 100644 --- a/algo/algo/index.html +++ b/algo/algo/index.html @@ -125,6 +125,9 @@ 算法原理

示例：

点击AIStudio执行以下示例代码。 -

import paddle.fluid as fluid
-from paddle.fluid.param_attr import ParamAttr
-from paddleslim.prune import Pruner
-
-def conv_bn_layer(input,
-                  num_filters,
-                  filter_size,
-                  name,
-                  stride=1,
-                  groups=1,
-                  act=None):
-    conv = fluid.layers.conv2d(
-        input=input,
-        num_filters=num_filters,
-        filter_size=filter_size,
-        stride=stride,
-        padding=(filter_size - 1) // 2,
-        groups=groups,
-        act=None,
-        param_attr=ParamAttr(name=name + "_weights"),
-        bias_attr=False,
-        name=name + "_out")
-    bn_name = name + "_bn"
-    return fluid.layers.batch_norm(
-        input=conv,
-        act=act,
-        name=bn_name + '_output',
-        param_attr=ParamAttr(name=bn_name + '_scale'),
-        bias_attr=ParamAttr(bn_name + '_offset'),
-        moving_mean_name=bn_name + '_mean',
-        moving_variance_name=bn_name + '_variance', )
-
-main_program = fluid.Program()
-startup_program = fluid.Program()
-#   X       X              O       X              O
-# conv1-->conv2-->sum1-->conv3-->conv4-->sum2-->conv5-->conv6
-#     |            ^ |                    ^
-#     |____________| |____________________|
-#
-# X: prune output channels
-# O: prune input channels
-with fluid.program_guard(main_program, startup_program):
-    input = fluid.data(name="image", shape=[None, 3, 16, 16])
-    conv1 = conv_bn_layer(input, 8, 3, "conv1")
-    conv2 = conv_bn_layer(conv1, 8, 3, "conv2")
-    sum1 = conv1 + conv2
-    conv3 = conv_bn_layer(sum1, 8, 3, "conv3")
-    conv4 = conv_bn_layer(conv3, 8, 3, "conv4")
-    sum2 = conv4 + sum1
-    conv5 = conv_bn_layer(sum2, 8, 3, "conv5")
-    conv6 = conv_bn_layer(conv5, 8, 3, "conv6")
-
-place = fluid.CPUPlace()
-exe = fluid.Executor(place)
-scope = fluid.Scope()
-exe.run(startup_program, scope=scope)
-pruner = Pruner()
-main_program, _, _ = pruner.prune(
-    main_program,
-    scope,
-    params=["conv4_weights"],
-    ratios=[0.5],
-    place=place,
-    lazy=False,
-    only_graph=False,
-    param_backup=False,
-    param_shape_backup=False)
-
-for param in main_program.global_block().all_parameters():
-    if "weights" in param.name:
-        print("param name: {}; param shape: {}".format(param.name, param.shape))
+
import paddle.fluid as fluid
+from paddle.fluid.param_attr import ParamAttr
+from paddleslim.prune import Pruner
+
+def conv_bn_layer(input,
+                  num_filters,
+                  filter_size,
+                  name,
+                  stride=1,
+                  groups=1,
+                  act=None):
+    conv = fluid.layers.conv2d(
+        input=input,
+        num_filters=num_filters,
+        filter_size=filter_size,
+        stride=stride,
+        padding=(filter_size - 1) // 2,
+        groups=groups,
+        act=None,
+        param_attr=ParamAttr(name=name + "_weights"),
+        bias_attr=False,
+        name=name + "_out")
+    bn_name = name + "_bn"
+    return fluid.layers.batch_norm(
+        input=conv,
+        act=act,
+        name=bn_name + '_output',
+        param_attr=ParamAttr(name=bn_name + '_scale'),
+        bias_attr=ParamAttr(bn_name + '_offset'),
+        moving_mean_name=bn_name + '_mean',
+        moving_variance_name=bn_name + '_variance', )
+
+main_program = fluid.Program()
+startup_program = fluid.Program()
+#   X       X              O       X              O
+# conv1-->conv2-->sum1-->conv3-->conv4-->sum2-->conv5-->conv6
+#     |            ^ |                    ^
+#     |____________| |____________________|
+#
+# X: prune output channels
+# O: prune input channels
+with fluid.program_guard(main_program, startup_program):
+    input = fluid.data(name="image", shape=[None, 3, 16, 16])
+    conv1 = conv_bn_layer(input, 8, 3, "conv1")
+    conv2 = conv_bn_layer(conv1, 8, 3, "conv2")
+    sum1 = conv1 + conv2
+    conv3 = conv_bn_layer(sum1, 8, 3, "conv3")
+    conv4 = conv_bn_layer(conv3, 8, 3, "conv4")
+    sum2 = conv4 + sum1
+    conv5 = conv_bn_layer(sum2, 8, 3, "conv5")
+    conv6 = conv_bn_layer(conv5, 8, 3, "conv6")
+
+place = fluid.CPUPlace()
+exe = fluid.Executor(place)
+scope = fluid.Scope()
+exe.run(startup_program, scope=scope)
+pruner = Pruner()
+main_program, _, _ = pruner.prune(
+    main_program,
+    scope,
+    params=["conv4_weights"],
+    ratios=[0.5],
+    place=place,
+    lazy=False,
+    only_graph=False,
+    param_backup=False,
+    param_shape_backup=False)
+
+for param in main_program.global_block().all_parameters():
+    if "weights" in param.name:
+        print("param name: {}; param shape: {}".format(param.name, param.shape))
 

 

 
sensitivity#

 

-
paddleslim.prune.sensitivity(program, place, param_names, eval_func, sensitivities_file=None, pruned_ratios=None) [源代码]

+
paddleslim.prune.sensitivity(program, place, param_names, eval_func, sensitivities_file=None, pruned_ratios=None) 源代码

 

 
计算网络中每个卷积层的敏感度。每个卷积层的敏感度信息统计方法为：依次剪掉当前卷积层不同比例的输出通道数，在测试集上计算剪裁后的精度损失。得到敏感度信息后，可以通过观察或其它方式确定每层卷积的剪裁率。

 

@@ -339,15 +339,15 @@
 
program(paddle.fluid.Program) - 待评估的目标网络。更多关于Program的介绍请参考：Program概念介绍。

 
 
  • 
-
    place(paddle.fluid.Place) - 待分析的参数所在的设备位置，可以是CUDAPlace或CPUPlace。
    
+
    place(paddle.fluid.Place) - 待分析的参数所在的设备位置，可以是CUDAPlace或CPUPlace。Place概念介绍
    
 
    • 
 
  • 
 
    param_names(list) - 待分析的卷积层的参数的名称列表。可以通过以下方式查看模型中所有参数的名称:
    
 
    • 
 
-
    for block in program.blocks:
-    for param in block.all_parameters():
-        print("param: {}; shape: {}".format(param.name, param.shape))
+
    for block in program.blocks:
+    for param in block.all_parameters():
+        print("param: {}; shape: {}".format(param.name, param.shape))
 
    
 
 
      
@@ -365,116 +365,116 @@
 
        
 
      • sensitivities(dict) - 存放敏感度信息的dict，其格式为：
        • 
 
      
-
      {"weight_0":
-   {0.1: 0.22,
-    0.2: 0.33
-   },
- "weight_1":
-   {0.1: 0.21,
-    0.2: 0.4
-   }
-}
+
      {"weight_0":
+   {0.1: 0.22,
+    0.2: 0.33
+   },
+ "weight_1":
+   {0.1: 0.21,
+    0.2: 0.4
+   }
+}
 
      
 
 
      其中，weight_0是卷积层参数的名称，sensitivities['weight_0']的value为剪裁比例，value为精度损失的比例。
      
 
      示例：
      
 
      点击AIStudio运行以下示例代码。
      
-
      import paddle
-import numpy as np
-import paddle.fluid as fluid
-from paddle.fluid.param_attr import ParamAttr
-from paddleslim.prune import sensitivity
-import paddle.dataset.mnist as reader
-
-def conv_bn_layer(input,
-                  num_filters,
-                  filter_size,
-                  name,
-                  stride=1,
-                  groups=1,
-                  act=None):
-    conv = fluid.layers.conv2d(
-        input=input,
-        num_filters=num_filters,
-        filter_size=filter_size,
-        stride=stride,
-        padding=(filter_size - 1) // 2,
-        groups=groups,
-        act=None,
-        param_attr=ParamAttr(name=name + "_weights"),
-        bias_attr=False,
-        name=name + "_out")
-    bn_name = name + "_bn"
-    return fluid.layers.batch_norm(
-        input=conv,
-        act=act,
-        name=bn_name + '_output',
-        param_attr=ParamAttr(name=bn_name + '_scale'),
-        bias_attr=ParamAttr(bn_name + '_offset'),
-        moving_mean_name=bn_name + '_mean',
-        moving_variance_name=bn_name + '_variance', )
-
-main_program = fluid.Program()
-startup_program = fluid.Program()
-#   X       X              O       X              O
-# conv1-->conv2-->sum1-->conv3-->conv4-->sum2-->conv5-->conv6
-#     |            ^ |                    ^
-#     |____________| |____________________|
-#
-# X: prune output channels
-# O: prune input channels
-image_shape = [1,28,28]
-with fluid.program_guard(main_program, startup_program):
-    image = fluid.data(name='image', shape=[None]+image_shape, dtype='float32')
-    label = fluid.data(name='label', shape=[None, 1], dtype='int64')  
-    conv1 = conv_bn_layer(image, 8, 3, "conv1")
-    conv2 = conv_bn_layer(conv1, 8, 3, "conv2")
-    sum1 = conv1 + conv2
-    conv3 = conv_bn_layer(sum1, 8, 3, "conv3")
-    conv4 = conv_bn_layer(conv3, 8, 3, "conv4")
-    sum2 = conv4 + sum1
-    conv5 = conv_bn_layer(sum2, 8, 3, "conv5")
-    conv6 = conv_bn_layer(conv5, 8, 3, "conv6")
-    out = fluid.layers.fc(conv6, size=10, act="softmax")
-#    cost = fluid.layers.cross_entropy(input=out, label=label)
-#    avg_cost = fluid.layers.mean(x=cost)
-    acc_top1 = fluid.layers.accuracy(input=out, label=label, k=1)
-#    acc_top5 = fluid.layers.accuracy(input=out, label=label, k=5)
-
-
-place = fluid.CPUPlace()
-exe = fluid.Executor(place)
-exe.run(startup_program)
-
-val_reader = paddle.batch(reader.test(), batch_size=128)
-val_feeder = feeder = fluid.DataFeeder(
-        [image, label], place, program=main_program)
-
-def eval_func(program):
-
-    acc_top1_ns = []
-    for data in val_reader():
-        acc_top1_n = exe.run(program,
-                             feed=val_feeder.feed(data),
-                             fetch_list=[acc_top1.name])
-        acc_top1_ns.append(np.mean(acc_top1_n))
-    return np.mean(acc_top1_ns)
-param_names = []
-for param in main_program.global_block().all_parameters():
-    if "weights" in param.name:
-        param_names.append(param.name)
-sensitivities = sensitivity(main_program,
-                            place,
-                            param_names,
-                            eval_func,
-                            sensitivities_file="./sensitive.data",
-                            pruned_ratios=[0.1, 0.2, 0.3])
-print(sensitivities)
+
      import paddle
+import numpy as np
+import paddle.fluid as fluid
+from paddle.fluid.param_attr import ParamAttr
+from paddleslim.prune import sensitivity
+import paddle.dataset.mnist as reader
+
+def conv_bn_layer(input,
+                  num_filters,
+                  filter_size,
+                  name,
+                  stride=1,
+                  groups=1,
+                  act=None):
+    conv = fluid.layers.conv2d(
+        input=input,
+        num_filters=num_filters,
+        filter_size=filter_size,
+        stride=stride,
+        padding=(filter_size - 1) // 2,
+        groups=groups,
+        act=None,
+        param_attr=ParamAttr(name=name + "_weights"),
+        bias_attr=False,
+        name=name + "_out")
+    bn_name = name + "_bn"
+    return fluid.layers.batch_norm(
+        input=conv,
+        act=act,
+        name=bn_name + '_output',
+        param_attr=ParamAttr(name=bn_name + '_scale'),
+        bias_attr=ParamAttr(bn_name + '_offset'),
+        moving_mean_name=bn_name + '_mean',
+        moving_variance_name=bn_name + '_variance', )
+
+main_program = fluid.Program()
+startup_program = fluid.Program()
+#   X       X              O       X              O
+# conv1-->conv2-->sum1-->conv3-->conv4-->sum2-->conv5-->conv6
+#     |            ^ |                    ^
+#     |____________| |____________________|
+#
+# X: prune output channels
+# O: prune input channels
+image_shape = [1,28,28]
+with fluid.program_guard(main_program, startup_program):
+    image = fluid.data(name='image', shape=[None]+image_shape, dtype='float32')
+    label = fluid.data(name='label', shape=[None, 1], dtype='int64')  
+    conv1 = conv_bn_layer(image, 8, 3, "conv1")
+    conv2 = conv_bn_layer(conv1, 8, 3, "conv2")
+    sum1 = conv1 + conv2
+    conv3 = conv_bn_layer(sum1, 8, 3, "conv3")
+    conv4 = conv_bn_layer(conv3, 8, 3, "conv4")
+    sum2 = conv4 + sum1
+    conv5 = conv_bn_layer(sum2, 8, 3, "conv5")
+    conv6 = conv_bn_layer(conv5, 8, 3, "conv6")
+    out = fluid.layers.fc(conv6, size=10, act="softmax")
+#    cost = fluid.layers.cross_entropy(input=out, label=label)
+#    avg_cost = fluid.layers.mean(x=cost)
+    acc_top1 = fluid.layers.accuracy(input=out, label=label, k=1)
+#    acc_top5 = fluid.layers.accuracy(input=out, label=label, k=5)
+
+
+place = fluid.CPUPlace()
+exe = fluid.Executor(place)
+exe.run(startup_program)
+
+val_reader = paddle.batch(reader.test(), batch_size=128)
+val_feeder = feeder = fluid.DataFeeder(
+        [image, label], place, program=main_program)
+
+def eval_func(program):
+
+    acc_top1_ns = []
+    for data in val_reader():
+        acc_top1_n = exe.run(program,
+                             feed=val_feeder.feed(data),
+                             fetch_list=[acc_top1.name])
+        acc_top1_ns.append(np.mean(acc_top1_n))
+    return np.mean(acc_top1_ns)
+param_names = []
+for param in main_program.global_block().all_parameters():
+    if "weights" in param.name:
+        param_names.append(param.name)
+sensitivities = sensitivity(main_program,
+                            place,
+                            param_names,
+                            eval_func,
+                            sensitivities_file="./sensitive.data",
+                            pruned_ratios=[0.1, 0.2, 0.3])
+print(sensitivities)
 
      
 
 
      merge_sensitive#
      
 
      
-
      paddleslim.prune.merge_sensitive(sensitivities)[源代码]
      
+
      paddleslim.prune.merge_sensitive(sensitivities)源代码
      
 
      
 
      合并多个敏感度信息。
      
 
      
@@ -487,22 +487,22 @@
 
        
 
      • sensitivities(dict) - 合并后的敏感度信息。其格式为：
        • 
 
      
-
      {"weight_0":
-   {0.1: 0.22,
-    0.2: 0.33
-   },
- "weight_1":
-   {0.1: 0.21,
-    0.2: 0.4
-   }
-}
+
      {"weight_0":
+   {0.1: 0.22,
+    0.2: 0.33
+   },
+ "weight_1":
+   {0.1: 0.21,
+    0.2: 0.4
+   }
+}
 
      
 
 
      其中，weight_0是卷积层参数的名称，sensitivities['weight_0']的value为剪裁比例，value为精度损失的比例。
      
 
      示例：
      
 
      load_sensitivities#
      
 
      
-
      paddleslim.prune.load_sensitivities(sensitivities_file)[源代码]
      
+
      paddleslim.prune.load_sensitivities(sensitivities_file)源代码
      
 
      
 
      从文件中加载敏感度信息。
      
 
      
@@ -518,7 +518,7 @@
 
      示例：
      
 
      get_ratios_by_loss#
      
 
      
-
      paddleslim.prune.get_ratios_by_loss(sensitivities, loss)[源代码]
      
+
      paddleslim.prune.get_ratios_by_loss(sensitivities, loss)源代码
      
 
      
 
      根据敏感度和精度损失阈值计算出一组剪切率。对于参数w, 其剪裁率为使精度损失低于loss的最大剪裁率。
      
 
      
diff --git a/api/quantization_api/index.html b/api/quantization_api/index.html
index 8e7d1707364f12f0eb7c8bde4120fa55eba4c838..87d88f93a84917f84685111c5f6f698cb9aa5b81 100644
--- a/api/quantization_api/index.html
+++ b/api/quantization_api/index.html
@@ -172,7 +172,7 @@
     
    • 量化
      • 
     
    • 
       
-         Edit on GitHub
       
     
      • 
@@ -184,29 +184,50 @@
               
                 
      量化配置#
      
 
      通过字典配置量化参数
      
-
      quant_config_default = {
-    'weight_quantize_type': 'abs_max',
-    'activation_quantize_type': 'abs_max',
-    'weight_bits': 8,
-    'activation_bits': 8,
-    # ops of name_scope in not_quant_pattern list, will not be quantized
-    'not_quant_pattern': ['skip_quant'],
-    # ops of type in quantize_op_types, will be quantized
-    'quantize_op_types':
-    ['conv2d', 'depthwise_conv2d', 'mul', 'elementwise_add', 'pool2d'],
-    # data type after quantization, such as 'uint8', 'int8', etc. default is 'int8'
-    'dtype': 'int8',
-    # window size for 'range_abs_max' quantization. defaulf is 10000
-    'window_size': 10000,
-    # The decay coefficient of moving average, default is 0.9
-    'moving_rate': 0.9,
+
      TENSORRT_OP_TYPES = [
+    'mul', 'conv2d', 'pool2d', 'depthwise_conv2d', 'elementwise_add',
+    'leaky_relu'
+]
+TRANSFORM_PASS_OP_TYPES = ['conv2d', 'depthwise_conv2d', 'mul']
+
+QUANT_DEQUANT_PASS_OP_TYPES = [
+        "pool2d", "elementwise_add", "concat", "softmax", "argmax", "transpose",
+        "equal", "gather", "greater_equal", "greater_than", "less_equal",
+        "less_than", "mean", "not_equal", "reshape", "reshape2",
+        "bilinear_interp", "nearest_interp", "trilinear_interp", "slice",
+        "squeeze", "elementwise_sub", "relu", "relu6", "leaky_relu", "tanh", "swish"
+    ]
+
+_quant_config_default = {
+    # weight quantize type, default is 'channel_wise_abs_max'
+    'weight_quantize_type': 'channel_wise_abs_max',
+    # activation quantize type, default is 'moving_average_abs_max'
+    'activation_quantize_type': 'moving_average_abs_max',
+    # weight quantize bit num, default is 8
+    'weight_bits': 8,
+    # activation quantize bit num, default is 8
+    'activation_bits': 8,
+    # ops of name_scope in not_quant_pattern list, will not be quantized
+    'not_quant_pattern': ['skip_quant'],
+    # ops of type in quantize_op_types, will be quantized
+    'quantize_op_types': ['conv2d', 'depthwise_conv2d', 'mul'],
+    # data type after quantization, such as 'uint8', 'int8', etc. default is 'int8'
+    'dtype': 'int8',
+    # window size for 'range_abs_max' quantization. defaulf is 10000
+    'window_size': 10000,
+    # The decay coefficient of moving average, default is 0.9
+    'moving_rate': 0.9,
+    # if True, 'quantize_op_types' will be TENSORRT_OP_TYPES
+    'for_tensorrt': False,
+    # if True, 'quantoze_op_types' will be TRANSFORM_PASS_OP_TYPES + QUANT_DEQUANT_PASS_OP_TYPES
+    'is_full_quantize': False
 }
 
      
 
 
      参数：
      
 
        
-
      • weight_quantize_type(str) - 参数量化方式。可选'abs_max',  'channel_wise_abs_max', 'range_abs_max', 'moving_average_abs_max'。 默认'abs_max'。
        • 
-
      • activation_quantize_type(str) - 激活量化方式，可选'abs_max', 'range_abs_max', 'moving_average_abs_max'，默认'abs_max'。
        • 
+
      • weight_quantize_type(str) - 参数量化方式。可选'abs_max',  'channel_wise_abs_max', 'range_abs_max', 'moving_average_abs_max'。如果使用TensorRT加载量化后的模型来预测，请使用'channel_wise_abs_max'。 默认'channel_wise_abs_max'。
        • 
+
      • activation_quantize_type(str) - 激活量化方式，可选'abs_max', 'range_abs_max', 'moving_average_abs_max'。如果使用TensorRT加载量化后的模型来预测，请使用'range_abs_max', 'moving_average_abs_max'。，默认'moving_average_abs_max'。
        • 
 
      • weight_bits(int) - 参数量化bit数，默认8, 推荐设为8。
        • 
 
      • activation_bits(int) -  激活量化bit数，默认8， 推荐设为8。
        • 
 
      • not_quant_pattern(str | list[str]) - 所有name_scope包含'not_quant_pattern'字符串的op，都不量化, 设置方式请参考fluid.name_scope。
        • 
@@ -214,6 +235,14 @@
 
      • dtype(int8) - 量化后的参数类型，默认 int8, 目前仅支持int8。
        • 
 
      • window_size(int) -  'range_abs_max'量化方式的window size，默认10000。
        • 
 
      • moving_rate(int) - 'moving_average_abs_max'量化方式的衰减系数，默认 0.9。
        • 
+
      • for_tensorrt(bool) - 量化后的模型是否使用TensorRT进行预测。如果是的话，量化op类型为：TENSORRT_OP_TYPES。默认值为False.
        • 
+
      • is_full_quantize(bool) - 是否量化所有可支持op类型。默认值为False.
        • 
+
      
+
      
+
      注意事项
      
+
      
+
        
+
      • 目前Paddle-Lite有int8 kernel来加速的op只有 ['conv2d', 'depthwise_conv2d', 'mul'], 其他op的int8 kernel将陆续支持。
        • 
 
      
 
      quant_aware#
      
 
      
@@ -237,13 +266,13 @@
 
    
 
    
 
    注意事项
    
+
    
 
      
 
    • 此接口会改变program结构，并且可能增加一些persistable的变量，所以加载模型参数时请注意和相应的program对应。
      • 
 
    • 此接口底层经历了fluid.Program-> fluid.framework.IrGraph->fluid.Program的转变，在fluid.framework.IrGraph中没有Parameter的概念，Variable只有persistable和not persistable的区别，所以在保存和加载参数时，请使用fluid.io.save_persistables和fluid.io.load_persistables接口。
      • 
 
    • 由于此接口会根据program的结构和量化配置来对program添加op，所以Paddle中一些通过fuse op来加速训练的策略不能使用。已知以下策略在使用量化时必须设为False： fuse_all_reduce_ops, sync_batch_norm。
      • 
 
    • 如果传入的program中存在和任何op都没有连接的Variable，则会在量化的过程中被优化掉。
      • 
 
    
-
    
 
    convert#
    
 
    
 
    paddleslim.quant.convert(program, place, config, scope=None, save_int8=False)[源代码]
    
@@ -266,10 +295,10 @@
 
 
    
 
    注意事项
    
-
    因为该接口会对op和Variable做相应的删除和修改，所以此接口只能在训练完成之后调用。如果想转化训练的中间模型，可加载相应的参数之后再使用此接口。
    
 
    
+
    因为该接口会对op和Variable做相应的删除和修改，所以此接口只能在训练完成之后调用。如果想转化训练的中间模型，可加载相应的参数之后再使用此接口。
    
 
    代码示例
    
-
    #encoding=utf8
+
    #encoding=utf8
 import paddle.fluid as fluid
 import paddleslim.quant as quant
 
@@ -311,7 +340,7 @@
 
    更详细的用法请参考 量化训练demo。
    
 
    quant_post#
    
 
    
-
    paddleslim.quant.quant_post(executor, model_dir, quantize_model_path,sample_generator, model_filename=None, params_filename=None, batch_size=16,batch_nums=None, scope=None, algo='KL', quantizable_op_type=["conv2d", "depthwise_conv2d", "mul"])[源代码]
    
+
    paddleslim.quant.quant_post(executor, model_dir, quantize_model_path,sample_generator, model_filename=None, params_filename=None, batch_size=16,batch_nums=None, scope=None, algo='KL', quantizable_op_type=["conv2d", "depthwise_conv2d", "mul"], is_full_quantize=False, is_use_cache_file=False, cache_dir="./temp_post_training")[源代码]
    
 
    
 
    对保存在${model_dir}下的模型进行量化，使用sample_generator的数据进行参数校正。
    
 
    
@@ -329,18 +358,24 @@
 
  • scope(fluid.Scope, optional) - 用来获取和写入Variable, 如果设置为None,则使用fluid.global_scope(). 默认值是None.
    • 
 
  • algo(str) - 量化时使用的算法名称，可为'KL'或者'direct'。该参数仅针对激活值的量化，因为参数值的量化使用的方式为'channel_wise_abs_max'. 当algo 设置为'direct'时，使用校正数据的激活值的绝对值的最大值当作Scale值，当设置为'KL'时，则使用KL散度的方法来计算Scale值。默认值为'KL'。
    • 
 
  • quantizable_op_type(list[str]) -  需要量化的op类型列表。默认值为["conv2d", "depthwise_conv2d", "mul"]。
    • 
+
  • is_full_quantize(bool) - 是否量化所有可支持的op类型。如果设置为False, 则按照 'quantizable_op_type' 的设置进行量化。
    • 
+
  • is_use_cache_file(bool) - 是否使用硬盘对中间结果进行存储。如果为False, 则将中间结果存储在内存中。
    • 
+
  • cache_dir(str) - 如果 'is_use_cache_file'为True, 则将中间结果存储在此参数设置的路径下。
    • 
 
 
    返回
    
 
    无。
    
 
    
 
    注意事项
    
-
    因为该接口会收集校正数据的所有的激活值，所以使用的校正图片不能太多。'KL'散度的计算也比较耗时。
    
 
    
+
      
+
    • 因为该接口会收集校正数据的所有的激活值，当校正图片比较多时，请设置'is_use_cache_file'为True, 将中间结果存储在硬盘中。另外，'KL'散度的计算比较耗时。
      • 
+
    • 目前Paddle-Lite有int8 kernel来加速的op只有 ['conv2d', 'depthwise_conv2d', 'mul'], 其他op的int8 kernel将陆续支持。
      • 
+
    
 
    代码示例
    
 
    
 
    注： 此示例不能直接运行，因为需要加载${model_dir}下的模型，所以不能直接运行。
    
 
    
-
    import paddle.fluid as fluid
+
    import paddle.fluid as fluid
 import paddle.dataset.mnist as reader
 from paddleslim.quant import quant_post
 val_reader = reader.train()
@@ -383,7 +418,7 @@
 
    返回类型
    
 
    fluid.Program
    
 
    代码示例
-
    import paddle.fluid as fluid
+
    import paddle.fluid as fluid
 import paddleslim.quant as quant
 
 train_program = fluid.Program()
diff --git a/api/single_distiller_api/index.html b/api/single_distiller_api/index.html
index 334c738d15c0c23c78fd4050a96f012ca220c810..11bb3de84331c403ca527ddb20718897d178e2d2 100644
--- a/api/single_distiller_api/index.html
+++ b/api/single_distiller_api/index.html
@@ -172,7 +172,7 @@
     
  • 知识蒸馏
    • 
     
  • 
       
-         Edit on GitHub
       
     
    • 
@@ -184,9 +184,9 @@
               
                 
    merge#
    
 
    
-
    paddleslim.dist.merge(teacher_program, student_program, data_name_map, place, scope=fluid.global_scope(), name_prefix='teacher_') [源代码] 
    
+
    paddleslim.dist.merge(teacher_program, student_program, data_name_map, place, scope=fluid.global_scope(), name_prefix='teacher_') [源代码]
    
 
    
-
    merge将两个paddle program（teacher_program, student_program）融合为一个program，并将融合得到的program返回。在融合的program中，可以为其中合适的teacher特征图和student特征图添加蒸馏损失函数，从而达到用teacher模型的暗知识（Dark Knowledge）指导student模型学习的目的。
    
+
    merge将teacher_program融合到student_program中。在融合的program中，可以为其中合适的teacher特征图和student特征图添加蒸馏损失函数，从而达到用teacher模型的暗知识（Dark Knowledge）指导student模型学习的目的。
    
 
    
 
    
 
    参数：
    
@@ -198,13 +198,13 @@
 
  • scope(Scope)-该参数表示程序使用的变量作用域，如果不指定将使用默认的全局作用域。默认值：fluid.global_scope()
    • 
 
  • name_prefix(str)-merge操作将统一为teacher的Variables添加的名称前缀name_prefix。默认值：'teacher_'
    • 
 
-
    返回： 由student_program和teacher_program merge得到的program
    
+
    返回： 无
    
 
    
 
    Note
    
 
    data_name_map 是 teacher_var name到student_var name的映射，如果写反可能无法正确进行merge
    
 
    
 
    使用示例：
    
-
    import paddle.fluid as fluid
+
    import paddle.fluid as fluid
 import paddleslim.dist as dist
 student_program = fluid.Program()
 with fluid.program_guard(student_program):
@@ -220,7 +220,7 @@
 data_name_map = {'y':'x'}
 USE_GPU = False
 place = fluid.CUDAPlace(0) if USE_GPU else fluid.CPUPlace()
-main_program = dist.merge(teacher_program, student_program,
+dist.merge(teacher_program, student_program,
                           data_name_map, place)
 
    
 
@@ -241,7 +241,7 @@
 
 
    返回： 由teacher_var1, teacher_var2, student_var1, student_var2组合得到的fsp_loss
    
 
    使用示例：
    
-
    import paddle.fluid as fluid
+
    import paddle.fluid as fluid
 import paddleslim.dist as dist
 student_program = fluid.Program()
 with fluid.program_guard(student_program):
@@ -257,8 +257,8 @@
 data_name_map = {'y':'x'}
 USE_GPU = False
 place = fluid.CUDAPlace(0) if USE_GPU else fluid.CPUPlace()
-main_program = merge(teacher_program, student_program, data_name_map, place)
-with fluid.program_guard(main_program):
+merge(teacher_program, student_program, data_name_map, place)
+with fluid.program_guard(student_program):
     distillation_loss = dist.fsp_loss('teacher_t1.tmp_1', 'teacher_t2.tmp_1',
                                       's1.tmp_1', 's2.tmp_1', main_program)
 
    
@@ -272,13 +272,13 @@
 
    
 
    参数：
    
 
      
-
    • teacher_var_name(str): teacher_var的名称. 
      • 
+
    • teacher_var_name(str): teacher_var的名称.
      • 
 
    • student_var_name(str): student_var的名称.
      • 
 
    • program(Program): 用于蒸馏训练的fluid program。默认值：fluid.default_main_program()
      • 
 
    
 
    返回： 由teacher_var, student_var组合得到的l2_loss
    
 
    使用示例：
    
-
    import paddle.fluid as fluid
+
    import paddle.fluid as fluid
 import paddleslim.dist as dist
 student_program = fluid.Program()
 with fluid.program_guard(student_program):
@@ -294,8 +294,8 @@
 data_name_map = {'y':'x'}
 USE_GPU = False
 place = fluid.CUDAPlace(0) if USE_GPU else fluid.CPUPlace()
-main_program = merge(teacher_program, student_program, data_name_map, place)
-with fluid.program_guard(main_program):
+merge(teacher_program, student_program, data_name_map, place)
+with fluid.program_guard(student_program):
     distillation_loss = dist.l2_loss('teacher_t2.tmp_1', 's2.tmp_1',
                                      main_program)
 
    
@@ -309,15 +309,15 @@
 
    
 
    参数：
    
 
      
-
    • teacher_var_name(str): teacher_var的名称. 
      • 
-
    • student_var_name(str): student_var的名称. 
      • 
+
    • teacher_var_name(str): teacher_var的名称.
      • 
+
    • student_var_name(str): student_var的名称.
      • 
 
    • program(Program): 用于蒸馏训练的fluid program。默认值：fluid.default_main_program()
      • 
-
    • teacher_temperature(float): 对teacher_var进行soft操作的温度值，温度值越大得到的特征图越平滑 
      • 
-
    • student_temperature(float): 对student_var进行soft操作的温度值，温度值越大得到的特征图越平滑 
      • 
+
    • teacher_temperature(float): 对teacher_var进行soft操作的温度值，温度值越大得到的特征图越平滑
      • 
+
    • student_temperature(float): 对student_var进行soft操作的温度值，温度值越大得到的特征图越平滑
      • 
 
    
 
    返回： 由teacher_var, student_var组合得到的soft_label_loss
    
 
    使用示例：
    
-
    import paddle.fluid as fluid
+
    import paddle.fluid as fluid
 import paddleslim.dist as dist
 student_program = fluid.Program()
 with fluid.program_guard(student_program):
@@ -333,8 +333,8 @@
 data_name_map = {'y':'x'}
 USE_GPU = False
 place = fluid.CUDAPlace(0) if USE_GPU else fluid.CPUPlace()
-main_program = merge(teacher_program, student_program, data_name_map, place)
-with fluid.program_guard(main_program):
+merge(teacher_program, student_program, data_name_map, place)
+with fluid.program_guard(student_program):
     distillation_loss = dist.soft_label_loss('teacher_t2.tmp_1',
                                              's2.tmp_1', main_program, 1., 1.)
 
    
@@ -348,13 +348,13 @@
 
    
 
    参数：
    
 
 
    返回：自定义的损失函数loss
    
 
    使用示例：
    
-
    import paddle.fluid as fluid
+
    import paddle.fluid as fluid
 import paddleslim.dist as dist
 student_program = fluid.Program()
 with fluid.program_guard(student_program):
@@ -370,13 +370,13 @@
 data_name_map = {'y':'x'}
 USE_GPU = False
 place = fluid.CUDAPlace(0) if USE_GPU else fluid.CPUPlace()
-main_program = merge(teacher_program, student_program, data_name_map, place)
+merge(teacher_program, student_program, data_name_map, place)
 def adaptation_loss(t_var, s_var):
     teacher_channel = t_var.shape[1]
     s_hint = fluid.layers.conv2d(s_var, teacher_channel, 1)
     hint_loss = fluid.layers.reduce_mean(fluid.layers.square(s_hint - t_var))
     return hint_loss
-with fluid.program_guard(main_program):
+with fluid.program_guard(student_program):
     distillation_loss = dist.loss(main_program, adaptation_loss,
             t_var='teacher_t2.tmp_1', s_var='s2.tmp_1')
 
    
diff --git a/index.html b/index.html
index 50fd8ff300c708a82f23e82fcbccc4d40e25c726..f006c752332664dcc875cb52a8d1e649f5c0b0c9 100644
--- a/index.html
+++ b/index.html
@@ -168,7 +168,7 @@
     
  • Home
    • 
     
  • 
       
-         Edit on GitHub
       
     
    • 
@@ -211,15 +211,15 @@
 
      
 
    • 安装develop版本
      • 
 
    
-
    git clone https://github.com/PaddlePaddle/PaddleSlim.git
-cd PaddleSlim
-python setup.py install
+
    git clone https://github.com/PaddlePaddle/PaddleSlim.git
+cd PaddleSlim
+python setup.py install
 
    
 
 
      
 
    • 安装官方发布的最新版本
      • 
 
    
-
    pip install paddleslim -i https://pypi.org/simple
+
    pip install paddleslim -i https://pypi.org/simple