Merge pull request mmistakes#77 from pesser/scheduling_variables

Scheduling variables

edflow/tf_util.py

@@ -1,4 +1,5 @@
 import tensorflow as tf
+import numpy as np
 
 
 def make_linear_var(
@@ -50,3 +51,36 @@ def make_exponential_var(step, start, end, start_value, end_value, decay):
     endstep = (np.log(end_value) - np.log(start_value)) / np.log(decay)
     stepper = make_linear_var(step, start, end, startstep, endstep)
     return tf.math.pow(decay, stepper) * start_value
+
+
+def make_var(step, var_type, options):
+    """
+
+    # usage within trainer
+    grad_weight = make_var(step=self.global_step,
+    var_type=self.config["grad_weight"]["var_type"],
+    options=self.config["grad_weight"]["options"])
+
+    # within yaml file
+    grad_weight:
+      var_type: linear
+      options:
+        start:      50000
+        end:        60000
+        start_value:  0.0
+        end_value: 1.0
+        clip_min: 1.0e-6
+        clip_max: 1.0
+
+    Parameters
+    ----------
+    step
+    var_type
+    options
+
+    Returns
+    -------
+
+    """
+    switch = {"linear": make_linear_var, "exponential": make_exponential_var}
+    return switch[var_type](step=step, **options)