allow scalar logging between steps

tensorpack/callbacks/monitor.py

@@ -130,9 +130,9 @@ class TFSummaryWriter(TrainingMonitor):
         self._writer = tf.summary.FileWriter(logger.LOG_DIR, graph=tf.get_default_graph())
 
     def put_summary(self, summary):
-        self._writer.add_summary(summary, self.trainer.global_step)
+        self._writer.add_summary(summary, self.global_step)
 
-    def _trigger(self):
+    def _trigger(self):     # flush every epoch
         self._writer.flush()
 
     def _after_train(self):
@@ -165,19 +165,25 @@ class JSONWriter(TrainingMonitor):
         self._stat_now = {}
 
         self._last_gs = -1
+        self._total = self.trainer.config.steps_per_epoch
 
-    def put_scalar(self, name, val):
-        gs = self.trainer.global_step
-        if gs != self._last_gs:
+    def _trigger_step(self):
+        # will do this in trigger_epoch
+        if self.local_step != self._total - 1:
             self._push()
-            self._last_gs = gs
-            self._stat_now['epoch_num'] = self.trainer.epoch_num
-            self._stat_now['global_step'] = gs
+
+    def _trigger_epoch(self):
+        self._push()
+
+    def put_scalar(self, name, val):
         self._stat_now[name] = float(val)   # TODO will fail for non-numeric
 
     def _push(self):
         """ Note that this method is idempotent"""
         if len(self._stat_now):
+            self._stat_now['epoch_num'] = self.epoch_num
+            self._stat_now['global_step'] = self.global_step
+
             self._stats.append(self._stat_now)
             self._stat_now = {}
             self._write_stat()
@@ -189,23 +195,42 @@ class JSONWriter(TrainingMonitor):
                 json.dump(self._stats, f)
             shutil.move(tmp_filename, self._fname)
         except IOError:  # disk error sometimes..
-            logger.exception("Exception in StatHolder.finalize()!")
-
-    def _trigger(self):
-        self._push()
+            logger.exception("Exception in JSONWriter._write_stat()!")
 
 
-# TODO print interval
 class ScalarPrinter(TrainingMonitor):
     """
-    Print all scalar data in terminal.
+    Print scalar data into terminal.
     """
-    def __init__(self):
+    def __init__(self, enable_step=False, enable_epoch=True):
+        """
+        Args:
+            enable_step, enable_epoch (bool): whether to print the
+                monitor data (if any) between steps or between epochs.
+        """
         self._whitelist = None
         self._blacklist = set([])
 
+        self._enable_step = enable_step
+        self._enable_epoch = enable_epoch
+
     def _setup_graph(self):
         self._dic = {}
+        self._total = self.trainer.config.steps_per_epoch
+
+    def _trigger_step(self):
+        if self._enable_step:
+            if self.local_step != self._total - 1:
+                # not the last step
+                self._print_stat()
+            else:
+                if not self._enable_epoch:
+                    self._print_stat()
+                # otherwise, will print them together
+
+    def _trigger_epoch(self):
+        if self._enable_epoch:
+            self._print_stat()
 
     def put_scalar(self, name, val):
         self._dic[name] = float(val)
@@ -215,9 +240,6 @@ class ScalarPrinter(TrainingMonitor):
             if self._whitelist is None or k in self._whitelist:
                 if k not in self._blacklist:
                     logger.info('{}: {:.5g}'.format(k, v))
-
-    def _trigger(self):
-        self._print_stat()
         self._dic = {}
 
 

tensorpack/callbacks/stats.py

@@ -15,7 +15,7 @@ class StatPrinter(Callback):
     def __init__(self, print_tag=None):
         log_deprecated("StatPrinter",
                        "No need to add StatPrinter to callbacks anymore!",
-                       "2017-03-26")
+                       "2017-05-26")
 
 
 # TODO make it into monitor?

tensorpack/callbacks/steps.py

@@ -85,7 +85,8 @@ class ProgressBar(Callback):
     def __init__(self, names=[]):
         """
         Args:
-            names(list): list of string, the names of the tensors to display.
+            names(list): list of string, the names of the tensors to monitor
+                on the progress bar.
         """
         super(ProgressBar, self).__init__()
         self._names = [get_op_tensor_name(n)[1] for n in names]