[singa] branch dev-postgresql updated: Add implementation for mlp models in model selection

[zh...@apache.org]

This is an automated email from the ASF dual-hosted git repository.

zhaojing pushed a commit to branch dev-postgresql
in repository https://gitbox.apache.org/repos/asf/singa.git


The following commit(s) were added to refs/heads/dev-postgresql by this push:
     new 3f1d9cc7 Add implementation for mlp models in model selection
     new d47766c4 Merge pull request #1083 from zmeihui/23-9-5-ms
3f1d9cc7 is described below

commit 3f1d9cc7e487400ab22e8c1c04bfb67c4e8bf721
Author: zmeihui <me...@bit.edu.cn>
AuthorDate: Tue Sep 5 21:56:47 2023 +0800

    Add implementation for mlp models in model selection
---
 examples/model_selection_psql/msmlp/model.py | 153 +++++++++++++++++++++++++++
 1 file changed, 153 insertions(+)

diff --git a/examples/model_selection_psql/msmlp/model.py b/examples/model_selection_psql/msmlp/model.py
new file mode 100644
index 00000000..44cf44f9
--- /dev/null
+++ b/examples/model_selection_psql/msmlp/model.py
@@ -0,0 +1,153 @@
+#
+# 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.
+#
+
+from singa import layer
+from singa import model
+from singa import tensor
+from singa import opt
+from singa import device
+from singa.autograd import Operator
+from singa.layer import Layer
+from singa import singa_wrap as singa
+import argparse
+import numpy as np
+
+np_dtype = {"float16": np.float16, "float32": np.float32}
+
+singa_dtype = {"float16": tensor.float16, "float32": tensor.float32}
+
+class MSMLP(model.Model):
+
+    def __init__(self, data_size=10, perceptron_size=100, num_classes=10):
+        super(MSMLP, self).__init__()
+        self.num_classes = num_classes
+        self.dimension = 2
+
+        self.relu = layer.ReLU()
+        self.linear1 = layer.Linear(perceptron_size)
+        self.linear2 = layer.Linear(num_classes)
+        self.softmax_cross_entropy = layer.SoftMaxCrossEntropy()
+        self.sum_error = SumErrorLayer()
+    
+    def forward(self, inputs):
+        y = self.linear1(inputs)
+        y = self.relu(y)
+        y = self.linear2(y)
+        return y
+
+    def train_one_batch(self, x, y, synflow_flag, dist_option, spars):
+        out = self.forward(x)
+        if synflow_flag:
+            loss = self.sum_error(out)
+        else:  # normal training
+            loss = self.softmax_cross_entropy(out, y)
+
+        if dist_option == 'plain':
+            pn_p_g_list = self.optimizer(loss)
+        elif dist_option == 'half':
+            self.optimizer.backward_and_update_half(loss)
+        elif dist_option == 'partialUpdate':
+            self.optimizer.backward_and_partial_update(loss)
+        elif dist_option == 'sparseTopK':
+            self.optimizer.backward_and_sparse_update(loss,
+                                                      topK=True,
+                                                      spars=spars)
+        elif dist_option == 'sparseThreshold':
+            self.optimizer.backward_and_sparse_update(loss,
+                                                      topK=False,
+                                                      spars=spars)
+        return pn_p_g_list, out, loss
+
+    def set_optimizer(self, optimizer):
+        self.optimizer = optimizer
+
+
+def create_model(pretrained=False, **kwargs):
+    """Constructs a CNN model.
+
+    Args:
+        pretrained (bool): If True, returns a pre-trained model.
+    
+    Returns:
+        The created CNN model.
+    """
+    model = MSMLP(**kwargs)
+
+    return model
+
+
+__all__ = ['MLP', 'create_model']
+
+if __name__ == "__main__":
+    np.random.seed(0)
+
+    parser = argparse.ArgumentParser()
+    parser.add_argument('-p',
+                        choices=['float32', 'float16'],
+                        default='float32',
+                        dest='precision')
+    parser.add_argument('-g',
+                        '--disable-graph',
+                        default='True',
+                        action='store_false',
+                        help='disable graph',
+                        dest='graph')
+    parser.add_argument('-m',
+                        '--max-epoch',
+                        default=1001,
+                        type=int,
+                        help='maximum epochs',
+                        dest='max_epoch')
+    args = parser.parse_args()
+
+    # generate the boundary
+    f = lambda x: (5 * x + 1)
+    bd_x = np.linspace(-1.0, 1, 200)
+    bd_y = f(bd_x)
+
+    # generate the training data
+    x = np.random.uniform(-1, 1, 400)
+    y = f(x) + 2 * np.random.randn(len(x))
+
+    # choose one precision
+    precision = singa_dtype[args.precision]
+    np_precision = np_dtype[args.precision]
+
+    # convert training data to 2d space
+    label = np.asarray([5 * a + 1 > b for (a, b) in zip(x, y)]).astype(np.int32)
+    data = np.array([[a, b] for (a, b) in zip(x, y)], dtype=np_precision)
+
+    dev = device.create_cuda_gpu_on(0)
+    sgd = opt.SGD(0.1, 0.9, 1e-5, dtype=singa_dtype[args.precision])
+    tx = tensor.Tensor((400, 2), dev, precision)
+    ty = tensor.Tensor((400,), dev, tensor.int32)
+    model = MLP(data_size=2, perceptron_size=3, num_classes=2)
+
+    # attach model to graph
+    model.set_optimizer(sgd)
+    model.compile([tx], is_train=True, use_graph=args.graph, sequential=True)
+    model.train()
+
+    for i in range(args.max_epoch):
+        tx.copy_from_numpy(data)
+        ty.copy_from_numpy(label)
+        out, loss = model(tx, ty, 'fp32', spars=None)
+
+        if i % 100 == 0:
+            print("training loss = ", tensor.to_numpy(loss)[0])