fix hard code bias in eval (#2187)

dygraph/ppdet/engine/trainer.py

@@ -115,7 +115,7 @@ class Trainer(object):
             return
         if self.cfg.metric == 'COCO':
             # TODO: bias should be unified
-            bias = 1 if 'bias' in self.cfg else 0
+            bias = self.cfg['bias'] if 'bias' in self.cfg else 0
             self._metrics = [
                 COCOMetric(
                     anno_file=self.dataset.get_anno(), bias=bias)