singa-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From jiny...@apache.org
Subject svn commit: r1691203 - /incubator/singa/site/trunk/content/markdown/docs/examples.md
Date Wed, 15 Jul 2015 13:05:57 GMT
Author: jinyang
Date: Wed Jul 15 13:05:56 2015
New Revision: 1691203

URL: http://svn.apache.org/r1691203
Log:
the examples

Added:
    incubator/singa/site/trunk/content/markdown/docs/examples.md

Added: incubator/singa/site/trunk/content/markdown/docs/examples.md
URL: http://svn.apache.org/viewvc/incubator/singa/site/trunk/content/markdown/docs/examples.md?rev=1691203&view=auto
==============================================================================
--- incubator/singa/site/trunk/content/markdown/docs/examples.md (added)
+++ incubator/singa/site/trunk/content/markdown/docs/examples.md Wed Jul 15 13:05:56 2015
@@ -0,0 +1,170 @@
+Title:
+Notice:    Licensed to the Apache Software Foundation (ASF) under one
+           or more contributor license agreements.  See the NOTICE file
+           distributed with this work for additional information
+           regarding copyright ownership.  The ASF licenses this file
+           to you under the Apache License, Version 2.0 (the
+           "License"); you may not use this file except in compliance
+           with the License.  You may obtain a copy of the License at
+           .
+             http://www.apache.org/licenses/LICENSE-2.0
+           .
+           Unless required by applicable law or agreed to in writing,
+           software distributed under the License is distributed on an
+           "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+           KIND, either express or implied.  See the License for the
+           specific language governing permissions and limitations
+           under the License.
+
+Here are the examples of SINGA, including MLP, CNN, RBM and RNN models. This tutorial will
show you some basic information about how to configure SINGA.
+To run a SINGA job, you need to configure two files, model.conf to specify  the deep learning
model and cluster.conf to define the distributed training architecture.
+
+model.conf
+====
+model.conf is the file that configures the deep learning model you want to train. 
+It should contain the neurualnet structure, training algorithm(backforward or contrastive
divergence etc.), 
+SGD update algorithm(e.g. Adagrad), number of training/test steps and training/test frequency,

+and display features and etc. 
+SINGA will read model.conf as a Google protobuf class 
+[ModelProto](https://github.com/apache/incubator-singa/blob/master/src/proto/model.proto).

+Here is a simple example simplified from our [MLP example](https://github.com/apache/incubator-singa/blob/master/examples/mnist/model.conf):
+
+    name: "simple-mlp"
+    train_steps: 1000
+    test_steps:10
+    test_frequency:60
+    display_frequency:30
+    alg: kBackPropagation
+    updater{
+      base_lr: 0.001
+      lr_change: kStep
+      type: kSGD
+      step_conf{
+        change_freq: 60
+        gamma: 0.997
+      }
+    }
+
+    neuralnet {
+    layer {
+      name: "data"
+      type: kShardData
+      sharddata_conf {
+        path: "examples/mnist/mnist_train_shard"
+        batchsize: 1000
+      }
+      exclude: kTest
+    }
+
+    layer {
+      name: "data"
+      type: kShardData
+      sharddata_conf {
+        path: "examples/mnist/mnist_test_shard"
+        batchsize: 1000
+      }
+      exclude: kTrain
+    }
+
+    layer{
+      name:"mnist"
+      type: kMnist
+      srclayers: "data"
+      mnist_conf {
+        norm_a: 127.5
+        norm_b: 1
+      }
+    }
+
+    layer{
+      name: "label"
+      type: kLabel
+      srclayers: "data"
+    }
+
+    layer{
+      name: "fc"
+      type: kInnerProduct
+      srclayers:"mnist"
+      innerproduct_conf{
+        num_output: 2500
+      }
+      param{
+        name: "weight"
+        init_method: kUniform
+        low:-0.05
+        high:0.05
+      }
+      param{
+        name: "bias"
+        init_method: kUniform
+        low: -0.05
+        high:0.05
+      }
+    }
+
+    layer{
+      name: "tanh"
+      type: kTanh
+      srclayers:"fc1"
+    }
+
+    layer{
+      name: "pre-softmax"
+      type: kInnerProduct
+      srclayers:"tanh1"
+      innerproduct_conf{
+        num_output: 2000
+      }
+      param{
+        name: "weight"
+        init_method: kUniform
+        low:-0.05
+        high:0.05
+      }
+      param{
+        name: "bias"
+        init_method: kUniform
+        low: -0.05
+        high:0.05
+      }
+    }
+
+    layer{
+      name: "loss"
+      type:kSoftmaxLoss
+      softmaxloss_conf{
+        topk:1
+      }
+      srclayers:"pre-softmax"
+      srclayers:"label"
+    }
+    }
+
+In this example, we define a neuralnet that contains one hidden layer. fc+tanh is the hidden
layer(fc is for the inner product part, and tanh is for the non-linear activation function),
and the final softmax layer is represented as pre-softmax+loss (inner product and softmax).
For each layer, we define its name, input layer(s), basic configurations (e.g. number of nodes,
parameter initialization settings). 
+You can also get more details about[programming model](http://singa.incubator.apache.org/docs/programming-model.html)
from our website.
+
+cluster.conf
+====
+cluster.conf is the file that configures the distributed architecture you want to use. 
+SINGA will read cluster.conf as a Google protobuf class [ClusterProto](https://github.com/apache/incubator-singa/blob/master/src/proto/cluster.proto).

+By configuring cluster.conf, you can let SINGA run in single machine, Sandblaster, Downpour,
Hogwild, AllReduce mode and etc.
+The details about architecture settings are described in [System Architecture](http://singa.incubator.apache.org/docs/architecture.html)
in our website. Below is a basic single machine configuration:
+
+
+    nworker_groups: 1
+    nserver_groups: 1
+    nservers_per_group: 1
+    nworkers_per_group: 1
+    nservers_per_procs: 1
+    nworkers_per_procs: 1
+    workspace: "examples/mnist/"
+
+
+List of examples
+====
+* [MLP using MNIST](http://singa.incubator.apache.org/docs/mlp.html)
+  - A simple backforward model : multilayer perception.
+* [CNN using CIFAR10](http://singa.incubator.apache.org/docs/cnn.html)
+  - A convolutional nereual network example, using more types of layers.
+ 



Mime
View raw message