多进程蒸馏¶
+大规模可扩展知识蒸馏框架 Pantheon¶
Teacher¶
pantheon.Teacher() source
@@ -164,8 +164,10 @@Return: An object of class Teacher
pantheon.Teacher.start() source
-: Start teacher service, sychronize with student and launch the thread -to monitor commands from student.
+-
+
- : Start teacher service, sychronize with student and launch the thread +
- to monitor commands from student. +
Args: None
Return: None
pantheon.Teacher.send(data) source
@@ -190,49 +192,68 @@ to monitor commands from student.Return: None
pantheon.Teacher.start_knowledge_service(feed_list, schema, program, reader_config, exe, buf_size=10, times=1) source
-: Start the knowledge service to generate and transfer knowledge data. In GPU mode, the devices to execute knowledge prediction will be determined by the -environment variable FLAGS_selected_gpus, or by CUDA_VISIBLE_DEVICES if it is not set, and by CPU_NUM (default 1) in CPU mode. Only supported in static graph.
-Args:
+-
+
- : Start the knowledge service to generate and transfer knowledge data. In GPU mode, the devices to execute knowledge prediction will be determined by the +
+
+environment variable FLAGS_selected_gpus, or by CUDA_VISIBLE_DEVICES if it is not set, and by CPU_NUM (default 1) in CPU mode. Only supported in static graph.Args:
+
+
feed_list (list): - A list of feed Variables or their names for the -input teacher Program.
+.. code-block:: guess ++
input teacher Program.
+
-schema (dict): - A dict to specify keys and fetched Variablesto generate knowledge.
+schema (dict): - A dict to specify keys and fetched Variables +.. code-block:: guess
++
to generate knowledge.
+program (fluid.Program): - Inference Program of the teacher model.
reader_config (dict): - The config for data reader. Support all the three types of generators used by fluid.io.PyReader and fluid.io.DataLoader, and their configs contain the key-value pair of the generator type and a generator object, plus other necessary argument pairs. See the following:
--
+
- sample generator:
- sample list generator: -
- batch generator: -
exe (fluid.Executor): The executor to run the input program.
buf_size (int): The size of buffers for data reader and knowledge -writer on each device.
+.. code-block:: guess ++
writer on each device.
+times (int): The maximum repeated serving times, default 1. Whenever -the public method get_knowledge_generator() in Student +.. code-block:: guess
++
the public method get_knowledge_generator() in Student object called once, the serving times will be added one, -until reaching the maximum and ending the service.
+until reaching the maximum and ending the service. Only +valid in online mode, and will be ignored in offline mode. +- !!! note “Note” +
- This example should be run with the example of class Student. +
- : The class defined for the student model. Receive knowledge data from +
+
+teacher model and carry out knowledge merging.Args:
+
+- merge_strategy (dict|None): - A dict whose keys are the common schemas shared by different teachers, and each corresponding value specifies the merging strategy for different schemas respectively, supporting sum and mean now.
- in_path (str|None): The input file path. Default None. -
- in_address (str|None): The input IP address, in the format “<IP_address>:<IP_port>“ (e.g. “127.0.0.1:8080“). Default None. +
- in_address (str|None): The input IP address, in the format “<IP_address>:<IP_port>” (e.g. “127.0.0.1:8080”). Default None.
- teacher_id (int): - The id of teacher that receives data from.
- The received data object.
- Knowledge description, which is a dict.
- : Get the real-time size of knowledge queue. If this size is denoted as +
- qsize, it means that there are qsize batch knowledge data already pushed into knowledge queue and waiting for the knowledge -generator to pop out. It‘s dynamic and limited up to 100, the capacity -of the knowledge queue. +generator to pop out. It’s dynamic and limited up to 100, the capacity +of the knowledge queue. +
- The real-time size of knowledge queue.
- batch_size (int): - The batch size of returned knowledge data. @@ -364,8 +398,10 @@ of the knowledge queue. # do something else
- !!! note “Note” +
- This example should be run with the example of class Teacher. +
- English Documents -
- PaddleSlim简介 +
- 介绍
- 安装
- 快速开始
- 进阶教程 @@ -91,7 +91,7 @@
- 模型分析
- SA-NAS
- OneShotNAS -
- 多进程蒸馏 +
- 大规模可扩展知识蒸馏框架 Pantheon
- 卷积层通道剪裁
- Pruner
- sensitivity @@ -499,7 +499,7 @@ } } sensitivities_file = "sensitive_api_demo.data" -with open(sensitivities_file, 'w') as f: +with open(sensitivities_file, 'wb') as f: pickle.dump(sen, f) sensitivities = load_sensitivities(sensitivities_file) print(sensitivities) @@ -553,7 +553,7 @@ Next - Previous + Previous
+
diff --git a/api_cn/prune_api.html b/api_cn/prune_api.html index 4c21c9b6f82e7ad997b423aa241d07d91e0f14e2..a1434d066359b59acab35f89bbb37621a8d3d876 100644 --- a/api_cn/prune_api.html +++ b/api_cn/prune_api.html @@ -36,7 +36,7 @@ - + @@ -83,7 +83,7 @@+reader_config={"sample_generator": some_sample_generator, - "batch_size": batch_size, "drop_last": drop_last} -# drop_last set to True by default +
-reader_config={"sample_generator": some_sample_generator, + "batch_size": batch_size, "drop_last": drop_last} + # drop_last set to True by default + +2) **sample list generator:**
-
-
-reader_config={"sample_list_generator": some_sample_list_generator} +
-reader_config={"sample_list_generator": some_sample_list_generator} + +3) **batch generator:**
-
-
-reader_config={"batch_generator": some_batch_genrator} +
-reader_config={"batch_generator": some_batch_genrator} + +The trial to parse config will be in the order of 1) -> 3), and any other unrelated keys in these configs will be ignored.
The trial to parse config will be in the order of 1) -> 3), and any other unrelated keys in these configs will be ignored.
Return: None
@@ -271,15 +292,21 @@ until reaching the maximum and ending the service. exe=exe)!!! note “Note“ -This example should be run with the example of class Student.
+-
+
Student¶
pantheon.Student(merge_strategy=None) source
-: The class defined for the student model. Receive knowledge data from -teacher model and carry out knowledge merging.
-Args:
+-
+
Args:
Return: None
pantheon.Student.start() source
-: End teachers‘ registration and synchronize with all of them.
+: End teachers’ registration and synchronize with all of them.
Args: None
Return: None
pantheon.Student.send(self, data, teacher_ids=None) source
@@ -306,7 +333,8 @@ teacher model and carry out knowledge merging.Return: None
pantheon.Student.recv(teacher_id) source
: Receive data from one teacher.
-Args:
++
Args:pantheon.Student.get_knowledge_desc() source
-: Get description for knowledge, including shape, data type and lod level for each schema.
++
: Get description for knowledge, including shape, data type and lod level for each schema.
Args: None
Return:
+pantheon.Student.get_knowledge_qsize() source
-: Get the real-time size of knowledge queue. If this size is denoted as -qsize, it means that there are qsize batch knowledge data +
+
-
+
Args: None
Return:
+pantheon.Student.get_knowledge_generator(batch_size, drop_last=False) source
-: Get the generator for knowledge data, return None if last generator doesn‘t finish yet.
+: Get the generator for knowledge data, return None if last generator doesn’t finish yet.
Args:
!!! note “Note“ -This example should be run with the example of class Teacher.
+-
+
- English Documents -
- PaddleSlim简介 +
- 介绍
- 安装
- 快速开始
- 进阶教程 @@ -91,7 +91,7 @@
- 模型分析
- SA-NAS
- OneShotNAS -
- 多进程蒸馏 +
- 大规模可扩展知识蒸馏框架 Pantheon
- 卷积层通道剪裁
- 量化
- 量化配置 diff --git a/api_cn/search_space.html b/api_cn/search_space.html index 7c6d2db25341a9d49c6ad46b5dcfbde9506aba8c..1ffb75264a77aef5769dbfdff19ca9d38f53e26d 100644 --- a/api_cn/search_space.html +++ b/api_cn/search_space.html @@ -83,7 +83,7 @@
- English Documents -
- PaddleSlim简介 +
- 介绍
- 安装
- 快速开始
- 进阶教程 @@ -91,7 +91,7 @@
- 模型分析
- SA-NAS
- OneShotNAS -
- 多进程蒸馏 +
- 大规模可扩展知识蒸馏框架 Pantheon
- 卷积层通道剪裁
- 量化
- 简单蒸馏 diff --git a/api_cn/single_distiller_api.html b/api_cn/single_distiller_api.html index fca9f79b5a15b44d1d585bba7815cd127127c073..07c9f411a08f8d4a6e4245899097158d7cf94b01 100644 --- a/api_cn/single_distiller_api.html +++ b/api_cn/single_distiller_api.html @@ -83,7 +83,7 @@
- English Documents -
- PaddleSlim简介 +
- 介绍
- 安装
- 快速开始
- 进阶教程 @@ -91,7 +91,7 @@
- 模型分析
- SA-NAS
- OneShotNAS -
- 多进程蒸馏 +
- 大规模可扩展知识蒸馏框架 Pantheon
- 卷积层通道剪裁
- 量化
- 简单蒸馏
-
diff --git a/api_cn/table_latency.html b/api_cn/table_latency.html
index 50863963ac4076b1e4ddde3a464a762ca9e0b838..37427d5d9e620702906d6a21b7e5305a747d0a24 100644
--- a/api_cn/table_latency.html
+++ b/api_cn/table_latency.html
@@ -83,7 +83,7 @@
- English Documents -
- PaddleSlim简介 +
- 介绍
- 安装
- 快速开始
- 进阶教程 @@ -91,7 +91,7 @@
- 模型分析
- SA-NAS
- OneShotNAS -
- 多进程蒸馏 +
- 大规模可扩展知识蒸馏框架 Pantheon
- 卷积层通道剪裁
- 量化
- 简单蒸馏 diff --git a/api_en/index_en.html b/api_en/index_en.html index 478b13a8df26fbd609878c68cbca69d59591b14d..d7c09cefc02d98b1d79808750688bd58fe351603 100644 --- a/api_en/index_en.html +++ b/api_en/index_en.html @@ -35,7 +35,7 @@ - + @@ -95,6 +95,7 @@
- paddleslim.nas.one_shot package
- paddleslim.pantheon package
- search space +
- Table about hardware lantency
- Model Zoo @@ -154,6 +155,7 @@
- paddleslim.nas.one_shot package
- paddleslim.pantheon package
- search space +
- Table about hardware lantency