internal merge of PR #1336

PiperOrigin-RevId: 227913649

Author: ywkim

tensor2tensor/data_generators/problem.py

@@ -368,6 +368,7 @@ def eval_metrics(self):
     ]
 
   def eval_metric_fns(self, model_hparams):
+    del model_hparams
     metric_names = self.eval_metrics()
     if not all([m in metrics.METRICS_FNS for m in metric_names]):
       error_str = ("Unrecognized metric. Problem %s specified metrics "

tensor2tensor/models/transformer.py

@@ -1121,22 +1121,6 @@ def body(self, features):
 
     return encoder_output
 
-@registry.register_model
-class TransformerRegressor(TransformerEncoder):
-  """Transformer inheriting from Encoder, for the regression problem.
-  Final res is a tensor that has a shape of (?, 1, 1, 1)
-  """
-
-  def top(self, body_output, features):
-    """Computes single scalar value from body_output
-    """
-    with tf.variable_scope("reg_top_ffn"):
-      # scalar = common_layers.dense(body_output,hparams)
-      x = body_output
-      x = tf.reduce_mean(x, axis=[1, 2], keepdims=True)
-      res = tf.layers.dense(x, 1, name="model_top")
-      return res
-
 
 def features_to_nonpadding(features, inputs_or_targets="inputs"):
   key = inputs_or_targets + "_segmentation"

tensor2tensor/utils/metrics.py

@@ -638,7 +638,7 @@ def create_eager_metrics_for_problem(problem, model_hparams):
   metric_fns = problem.eval_metric_fns(model_hparams)
   tm = problem.get_hparams(model_hparams).modality["targets"]
   return create_eager_metrics_internal(
-        metric_fns, weights_fn=tm.targets_weights_fn)
+      metric_fns, weights_fn=tm.targets_weights_fn)
 
 
 def create_eager_metrics(metric_names, weights_fn=common_layers.weights_all):
@@ -664,7 +664,7 @@ def create_eager_metrics_internal(metric_fns,
   """Create metrics accumulators and averager for Eager mode.
 
   Args:
-    metric_names: dict<metric name, metric function>
+    metric_fns: dict<metric name, metric function>
     weights_fn: function that takes labels and returns a weights mask. Defaults
       to weights of all 1, i.e. common_layers.weights_all. Use
       common_layers.weights_nonzero if labels have 0-padding.