10 examples of 'tensorflow load checkpoint' in Python

333
def load_checkpoint(sess, saver):
334
    #ckpt = tf.train.get_checkpoint_state('save')
335
    #if ckpt and ckpt.model_checkpoint_path:
336
    #saver.restore(sess, tf.train.latest_checkpoint('save'))
337
    ckpt = 'pretrain_g'+str(config['PRE_GEN_EPOCH'])+'_d'+str(config['PRE_DIS_EPOCH'])+'.ckpt'
338
    saver.restore(sess, './save/' + ckpt)
339
    print 'checkpoint {} loaded'.format(ckpt)
340
    return

147
def load(saver, sess, ckpt_path):
148
    '''Load trained weights.
149
150
    Args:
151
      saver: TensorFlow saver object.
152
      sess: TensorFlow session.
153
      ckpt_path: path to checkpoint file with parameters.
154
    '''
155
    ckpt = tf.train.get_checkpoint_state(ckpt_path)
156
    if ckpt and ckpt.model_checkpoint_path:
157
        # ckpt_name = os.path.basename(ckpt.model_checkpoint_path)
158
        # saver.restore(sess, os.path.join(ckpt_path, ckpt_name))
159
        saver.restore(sess, ckpt.model_checkpoint_path)
160
        # print("Restored model parameters from {}".format(ckpt_name))
161
        print("Restored model parameters")
162
        return True
163
    else:
164
        return False

30
def load_ckpt(saver, sess, ckpt_dir="train"):
31
  """Load checkpoint from the ckpt_dir (if unspecified, this is train dir) and restore it to saver and sess, waiting 10 secs in the case of failure. Also returns checkpoint name."""
32
  while True:
33
    try:
34
      latest_filename = "checkpoint_best" if ckpt_dir=="eval" else None
35
      ckpt_dir = os.path.join(FLAGS.log_root, ckpt_dir)
36
      ckpt_state = tf.train.get_checkpoint_state(ckpt_dir, latest_filename=latest_filename)
37
      tf.logging.info('Loading checkpoint %s', ckpt_state.model_checkpoint_path)
38
      saver.restore(sess, ckpt_state.model_checkpoint_path)
39
      return ckpt_state.model_checkpoint_path
40
    except:
41
      tf.logging.info("Failed to load checkpoint from %s. Sleeping for %i secs...", ckpt_dir, 10)
42
      time.sleep(10)

48
def load_latest_checkpoint(self):
49
    self.saver.restore(self.sess, tf.train.latest_checkpoint('checkpoints'))

155
def load(self, checkpoint_dir):
156
    """
157
        To load the checkpoint use to test or pretrain
158
    """
159
    print("\nReading Checkpoints.....\n\n")
160
    model_dir = "%s_%s_%s" % ("espcn", self.image_size,self.scale)# give the model name by label_size
161
    checkpoint_dir = os.path.join(checkpoint_dir, model_dir)
162
    ckpt = tf.train.get_checkpoint_state(checkpoint_dir)
163
    
164
    # Check the checkpoint is exist 
165
    if ckpt and ckpt.model_checkpoint_path:
166
        ckpt_path = str(ckpt.model_checkpoint_path) # convert the unicode to string
167
        self.saver.restore(self.sess, os.path.join(os.getcwd(), ckpt_path))
168
        print("\n Checkpoint Loading Success! %s\n\n"% ckpt_path)
169
    else:
170
        print("\n! Checkpoint Loading Failed \n\n")

211
def restore_from_checkpoint(self):
212
  """Returns scaffold function to restore parameters from checkpoint."""
213
  def scaffold_fn():
214
    """Loads pretrained model through scaffold function."""
215
    tf.train.init_from_checkpoint(self._checkpoint,
216
                                  {'/': self._checkpoint_prefix,})
217
    return tf.train.Scaffold()
218
  return scaffold_fn if self._checkpoint else None

9
def get_checkpoint(logdir):
10
    ''' Get the first checkpoint '''
11
    ckpt = tf.train.get_checkpoint_state(logdir)
12
    if ckpt:
13
        return ckpt.model_checkpoint_path
14
    else:
15
        print('No checkpoint found')
16
        return None

138
def load_chkpt(saver, sess, chkptdir):
139
	ckpt = tf.train.get_checkpoint_state(chkptdir)
140
	if ckpt and ckpt.model_checkpoint_path:
141
		ckpt_fn = ckpt.model_checkpoint_path.replace('//', '/') 
142
		print('[DEBUG] Loading checkpoint from %s' % ckpt_fn)
143
		saver.restore(sess, ckpt_fn)
144
	else:
145
		raise NameError('[ERROR] No checkpoint found at: %s' % chkptdir)

279
def restore_variables(checkpoint):
280
    if not checkpoint:
281
        return tf.no_op("restore_op")
282
283
    # Load checkpoints
284
    tf.logging.info("Loading %s" % checkpoint)
285
    var_list = tf.train.list_variables(checkpoint)
286
    reader = tf.train.load_checkpoint(checkpoint)
287
    values = {}
288
289
    for (name, shape) in var_list:
290
        tensor = reader.get_tensor(name)
291
        name = name.split(":")[0]
292
        values[name] = tensor
293
294
    var_list = tf.trainable_variables()
295
    ops = []
296
297
    for var in var_list:
298
        name = var.name.split(":")[0]
299
300
        if name in values:
301
            tf.logging.info("Restore %s" % var.name)
302
            ops.append(tf.assign(var, values[name]))
303
304
    return tf.group(*ops, name="restore_op")

221
def restore_checkpoint_if_exists(saver, sess, logdir):
222
    """Looks for a checkpoint and restores the session from it if found.
223
    Args:
224
      saver: A tf.train.Saver for restoring the session.
225
      sess: A TensorFlow session.
226
      logdir: The directory to look for checkpoints in.
227
    Returns:
228
      True if a checkpoint was found and restored, False otherwise.
229
    """
230
    checkpoint = tf.train.get_checkpoint_state(logdir)
231
    if checkpoint:
232
        checkpoint_name = os.path.basename(checkpoint.model_checkpoint_path)
233
        full_checkpoint_path = os.path.join(logdir, checkpoint_name)
234
        saver.restore(sess, full_checkpoint_path)
235
        return True
236
    return False