auto_set_dir

examples/Atari2600/DQN.py

@@ -137,9 +137,7 @@ class Model(ModelDesc):
         return self.predict_value.eval(feed_dict={'state:0': [state]})[0]
 
 def get_config():
-    basename = os.path.basename(__file__)
-    logger.set_logger_dir(
-        os.path.join('train_log', basename[:basename.rfind('.')]))
+    logger.auto_set_dir()
 
     M = Model()
     dataset_train = ExpReplay(

examples/DisturbLabel/README.md

@@ -14,4 +14,5 @@ Experiements are repeated 15 times for p=0, 10 times for p=0.02 & 0.05, and 5 ti
 of p. All experiements run for 100 epochs, with lr decay, which are enough for them to converge.
 
 I suppose the disturb method works as a random noise to prevent SGD from getting stuck.
-Despite the positive results here, I still doubt whether the method works for ImageNet.
+It doesn't work for harder problems such as SVHN (details to follow). And I don't believe
+it will work for ImageNet.

examples/DisturbLabel/disturb.py

+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+# File: disturb.py
+# Author: Yuxin Wu <ppwwyyxxc@gmail.com>
+
+from tensorpack import ProxyDataFlow, get_rng
+
+class DisturbLabel(ProxyDataFlow):
+    def __init__(self, ds, prob):
+        super(DisturbLabel, self).__init__(ds)
+        self.prob = prob
+        self.rng = get_rng(self)
+
+    def get_data(self):
+        for dp in self.ds.get_data():
+            img, l = dp
+            if self.rng.rand() < self.prob:
+                l = self.rng.choice(10)
+            yield [img, l]
+

examples/DisturbLabel/mnist-disturb.py

@@ -9,15 +9,22 @@ import os, sys
 import argparse
 
 from tensorpack import *
+from disturb import DisturbLabel
 
-BATCH_SIZE = 128
-IMAGE_SIZE = 28
+import imp
+mnist_example = imp.load_source('mnist_example',
+        os.path.join(os.path.dirname(__file__), '..', 'mnist-convnet.py'))
+get_config = mnist_example.get_config
+
+def get_data():
+    dataset_train = BatchData(DisturbLabel(dataset.Mnist('train'), args.prob), 128)
+    dataset_test = BatchData(dataset.Mnist('test'), 256, remainder=True)
+    return dataset_train, dataset_test
+mnist_example.get_data = get_data
 
-class Model(ModelDesc):
-    def _get_input_vars(self):
-        return [InputVar(tf.float32, (None, IMAGE_SIZE, IMAGE_SIZE), 'input'),
-                InputVar(tf.int32, (None,), 'label') ]
+IMAGE_SIZE = 28
 
+class Model(mnist_example.Model):
     def _build_graph(self, input_vars, is_training):
         is_training = bool(is_training)
         keep_prob = tf.constant(0.5 if is_training else 1.0)
@@ -54,62 +61,16 @@ class Model(ModelDesc):
 
         self.cost = tf.add_n([wd_cost, cost], name='cost')
 
-class DisturbLabel(ProxyDataFlow):
-    def __init__(self, ds, prob):
-        super(DisturbLabel, self).__init__(ds)
-        self.prob = prob
-        self.rng = get_rng(self)
-
-    def get_data(self):
-        for dp in self.ds.get_data():
-            img, l = dp
-            if self.rng.rand() < self.prob:
-                l = self.rng.choice(10)
-            yield [img, l]
-
-def get_config(disturb_prob):
-    basename = os.path.basename(__file__)
-    logger.set_logger_dir(
-        os.path.join('train_log', basename[:basename.rfind('.')]))
-
-    # prepare dataset
-    dataset_train = BatchData(DisturbLabel(dataset.Mnist('train'),
-        disturb_prob), 128)
-    dataset_test = BatchData(dataset.Mnist('test'), 256, remainder=True)
-    step_per_epoch = dataset_train.size()
-
-    lr = tf.train.exponential_decay(
-        learning_rate=1e-3,
-        global_step=get_global_step_var(),
-        decay_steps=dataset_train.size() * 10,
-        decay_rate=0.3, staircase=True, name='learning_rate')
-    tf.scalar_summary('learning_rate', lr)
-
-    return TrainConfig(
-        dataset=dataset_train,
-        optimizer=tf.train.AdamOptimizer(lr),
-        callbacks=Callbacks([
-            StatPrinter(),
-            ModelSaver(),
-            InferenceRunner(dataset_test,
-                [ScalarStats('cost'), ClassificationError() ])
-        ]),
-        session_config=get_default_sess_config(0.5),
-        model=Model(),
-        step_per_epoch=step_per_epoch,
-        max_epoch=100,
-    )
-
 if __name__ == '__main__':
     parser = argparse.ArgumentParser()
     parser.add_argument('--gpu', help='comma separated list of GPU(s) to use.') # nargs='*' in multi mode
     parser.add_argument('--load', help='load model')
-    parser.add_argument('--prob', help='disturb prob', type=float)
+    parser.add_argument('--prob', help='disturb prob', type=float, required=True)
     args = parser.parse_args()
     if args.gpu:
         os.environ['CUDA_VISIBLE_DEVICES'] = args.gpu
 
-    config = get_config(args.prob)
+    config = get_config()
     if args.load:
         config.session_init = SaverRestore(args.load)
     QueueInputTrainer(config).train()

examples/DoReFa-Net/svhn-digit-dorefa.py

@@ -164,9 +164,7 @@ class Model(ModelDesc):
         self.cost = tf.add_n([cost, wd_cost], name='cost')
 
 def get_config():
-    basename = os.path.basename(__file__)
-    logger.set_logger_dir(
-        os.path.join('train_log', basename[:basename.rfind('.')]))
+    logger.auto_set_dir()
 
     # prepare dataset
     d1 = dataset.SVHNDigit('train')

examples/Inception/inception-bn.py

@@ -149,6 +149,7 @@ def get_data(train_or_test):
 
 
 def get_config():
+    logger.auto_set_dir()
     # prepare dataset
     dataset_train = get_data('train')
     step_per_epoch = 5000
@@ -184,15 +185,9 @@ if __name__ == '__main__':
     parser = argparse.ArgumentParser()
     parser.add_argument('--gpu', help='comma separated list of GPU(s) to use.') # nargs='*' in multi mode
     parser.add_argument('--load', help='load model')
-    parser.add_argument('--data', help='ImageNet data root directory',
-                        required=True)
-    global args
+    parser.add_argument('--data', help='ImageNet data root directory', required=True)
     args = parser.parse_args()
 
-    basename = os.path.basename(__file__)
-    logger.set_logger_dir(
-        os.path.join('train_log', basename[:basename.rfind('.')]))
-
     if args.gpu:
         os.environ['CUDA_VISIBLE_DEVICES'] = args.gpu
 

examples/ResNet/cifar10-resnet.py

@@ -141,9 +141,7 @@ def get_data(train_or_test):
     return ds
 
 def get_config():
-    basename = os.path.basename(__file__)
-    logger.set_logger_dir(
-        os.path.join('train_log', basename[:basename.rfind('.')]))
+    logger.auto_set_dir()
 
     # prepare dataset
     dataset_train = get_data('train')

examples/ResNet/svhn-resnet.py

@@ -144,9 +144,7 @@ def get_data(train_or_test):
     return ds
 
 def get_config():
-    basename = os.path.basename(__file__)
-    logger.set_logger_dir(
-        os.path.join('train_log', basename[:basename.rfind('.')]))
+    logger.auto_set_dir()
 
     # prepare dataset
     dataset_train = get_data('train')

examples/char-rnn/char-rnn.py

@@ -106,9 +106,7 @@ class Model(ModelDesc):
             [grad], param.grad_clip)[0][0])]
 
 def get_config():
-    basename = os.path.basename(__file__)
-    logger.set_logger_dir(
-        os.path.join('train_log', basename[:basename.rfind('.')]))
+    logger.auto_set_dir()
 
     ds = CharRNNData(param.corpus, 100000)
     ds = BatchData(ds, param.batch_size)

examples/cifar-convnet.py

@@ -107,6 +107,8 @@ def get_data(train_or_test, cifar_classnum):
     return ds
 
 def get_config(cifar_classnum):
+    logger.auto_set_dir()
+
     # prepare dataset
     dataset_train = get_data('train', cifar_classnum)
     step_per_epoch = dataset_train.size()
@@ -144,10 +146,6 @@ if __name__ == '__main__':
                         type=int, default=10)
     args = parser.parse_args()
 
-    basename = os.path.basename(__file__)
-    logger.set_logger_dir(
-        os.path.join('train_log', basename[:basename.rfind('.')]))
-
     if args.gpu:
         os.environ['CUDA_VISIBLE_DEVICES'] = args.gpu
     else:

examples/mnist-convnet.py

@@ -15,7 +15,6 @@ MNIST ConvNet example.
 about 0.6% validation error after 30 epochs.
 """
 
-BATCH_SIZE = 128
 IMAGE_SIZE = 28
 
 class Model(ModelDesc):
@@ -65,14 +64,15 @@ class Model(ModelDesc):
         summary.add_param_summary([('.*/W', ['histogram'])])   # monitor histogram of all W
         self.cost = tf.add_n([wd_cost, cost], name='cost')
 
+def get_data():
+    train = BatchData(dataset.Mnist('train'), 128)
+    test = BatchData(dataset.Mnist('test'), 256, remainder=True)
+    return train, test
+
 def get_config():
-    basename = os.path.basename(__file__)
-    logger.set_logger_dir(
-        os.path.join('train_log', basename[:basename.rfind('.')]))
+    logger.auto_set_dir()
 
-    # prepare dataset
-    dataset_train = BatchData(dataset.Mnist('train'), 128)
-    dataset_test = BatchData(dataset.Mnist('test'), 256, remainder=True)
+    dataset_train, dataset_test = get_data()
     step_per_epoch = dataset_train.size()
 
     lr = tf.train.exponential_decay(

examples/svhn-digit-convnet.py

@@ -56,11 +56,10 @@ class Model(ModelDesc):
         wd_cost = regularize_cost('fc.*/W', l2_regularizer(0.00001))
         tf.add_to_collection(MOVING_SUMMARY_VARS_KEY, wd_cost)
 
-        add_param_summary([('.*/W', ['histogram', 'sparsity'])])   # monitor W
+        add_param_summary([('.*/W', ['histogram', 'rms'])])   # monitor W
         self.cost = tf.add_n([cost, wd_cost], name='cost')
 
-def get_config():
-    # prepare dataset
+def get_data():
     d1 = dataset.SVHNDigit('train')
     d2 = dataset.SVHNDigit('extra')
     data_train = RandomMixData([d1, d2])
@@ -77,11 +76,17 @@ def get_config():
     data_train = AugmentImageComponent(data_train, augmentors)
     data_train = BatchData(data_train, 128)
     data_train = PrefetchData(data_train, 5, 5)
-    step_per_epoch = data_train.size()
 
     augmentors = [ imgaug.Resize((40, 40)) ]
     data_test = AugmentImageComponent(data_test, augmentors)
     data_test = BatchData(data_test, 128, remainder=True)
+    return data_train, data_test
+
+def get_config():
+    logger.auto_set_dir()
+
+    data_train, data_test = get_data()
+    step_per_epoch = data_train.size()
 
     lr = tf.train.exponential_decay(
         learning_rate=1e-3,
@@ -110,10 +115,6 @@ if __name__ == '__main__':
     parser.add_argument('--load', help='load model')
     args = parser.parse_args()
 
-    basename = os.path.basename(__file__)
-    logger.set_logger_dir(
-        os.path.join('train_log', basename[:basename.rfind('.')]))
-
     if args.gpu:
         os.environ['CUDA_VISIBLE_DEVICES'] = args.gpu
     else:

tensorpack/utils/logger.py

@@ -14,7 +14,7 @@ from .fs import mkdir_p
 
 __all__ = []
 
-class MyFormatter(logging.Formatter):
+class _MyFormatter(logging.Formatter):
     def format(self, record):
         date = colored('[%(asctime)s @%(filename)s:%(lineno)d]', 'green')
         msg = '%(message)s'
@@ -28,17 +28,17 @@ class MyFormatter(logging.Formatter):
             # Python3 compatibilty
             self._style._fmt = fmt
         self._fmt = fmt
-        return super(MyFormatter, self).format(record)
+        return super(_MyFormatter, self).format(record)
 
-def getlogger():
+def _getlogger():
     logger = logging.getLogger('tensorpack')
     logger.propagate = False
     logger.setLevel(logging.INFO)
     handler = logging.StreamHandler(sys.stdout)
-    handler.setFormatter(MyFormatter(datefmt='%m%d %H:%M:%S'))
+    handler.setFormatter(_MyFormatter(datefmt='%m%d %H:%M:%S'))
     logger.addHandler(handler)
     return logger
-logger = getlogger()
+_logger = _getlogger()
 
 
 def get_time_str():
@@ -52,8 +52,8 @@ def _set_file(path):
         info("Log file '{}' backuped to '{}'".format(path, backup_name))
     hdl = logging.FileHandler(
         filename=path, encoding='utf-8', mode='w')
-    hdl.setFormatter(MyFormatter(datefmt='%m%d %H:%M:%S'))
-    logger.addHandler(hdl)
+    hdl.setFormatter(_MyFormatter(datefmt='%m%d %H:%M:%S'))
+    _logger.addHandler(hdl)
 
 def set_logger_dir(dirname, action=None):
     """
@@ -63,10 +63,10 @@ def set_logger_dir(dirname, action=None):
     """
     global LOG_FILE, LOG_DIR
     if os.path.isdir(dirname):
-        logger.warn("""\
+        _logger.warn("""\
 Directory {} exists! Please either backup/delete it, or use a new directory \
 unless you're resuming from a previous task.""".format(dirname))
-        logger.info("Select Action: k (keep) / b (backup) / d (delete) / n (new):")
+        _logger.info("Select Action: k (keep) / b (backup) / d (delete) / n (new):")
         if not action:
             while True:
                 act = input().lower().strip()
@@ -92,10 +92,21 @@ unless you're resuming from a previous task.""".format(dirname))
     LOG_FILE = os.path.join(dirname, 'log.log')
     _set_file(LOG_FILE)
 
+# export logger functions
+for func in ['info', 'warning', 'error', 'critical', 'warn', 'exception', 'debug']:
+    locals()[func] = getattr(_logger, func)
+
 def disable_logger():
+    """ disable all logging ability from this moment"""
     for func in ['info', 'warning', 'error', 'critical', 'warn', 'exception', 'debug']:
         globals()[func] = lambda x: None
 
-# export logger functions
-for func in ['info', 'warning', 'error', 'critical', 'warn', 'exception', 'debug']:
-    locals()[func] = getattr(logger, func)
+def auto_set_dir(action=None):
+    """ set log directory to a subdir inside 'train_log', with the name being
+    the main python file currently running"""
+    mod = sys.modules['__main__']
+    basename = os.path.basename(mod.__file__)
+    set_logger_dir(
+            os.path.join('train_log',
+                basename[:basename.rfind('.')]),
+            action=action)