fix pserver weight decay multi inputs test=develop

Author: typhoonzero

python/paddle/fluid/transpiler/distribute_transpiler.py

@@ -1706,13 +1706,27 @@ def _get_param_block(opt_op):
             outputs=outputs,
             attrs=opt_op.all_attrs())
 
-    def _is_splited_grad_var(self, var, var_dict):
+    def _get_pserver_grad_param_var(self, var, var_dict):
+        """
+        Return pserver side grad/param variable, return None
+        if the variable is not grad/param, e.g.
+
+            a@GRAD -> [email protected]
+            a@GRAD -> a@GRAD (a is not splited)
+            fc_0.w_0 -> fc_0.w_0.block_0
+            fc_0.w_0 -> fc_0.w_0 (weight is not splited)
+            _generated_var_123 -> None
+        """
         grad_block = None
         for _, g in six.iteritems(var_dict):
             if self._orig_varname(g.name) == self._orig_varname(var.name):
+                # skip per trainer vars
                 if g.name.find(".trainer_") == -1:
-                    grad_block = g
-                    break
+                    # only param or grads have splited blocks
+                    if self._orig_varname(g.name) in self.grad_name_to_param_name or\
+                        self._orig_varname(g.name) in self.param_name_to_grad_name:
+                        grad_block = g
+                        break
         return grad_block
 
     def _clone_lr_op(self, program, block, op):
@@ -1745,32 +1759,38 @@ def _append_pserver_non_opt_ops(self, optimize_block, opt_op):
         for key, varlist in six.iteritems(inputs):
             if not isinstance(varlist, list):
                 varlist = [varlist]
-            for var in varlist:
-                # for ops like clipping and weight decay, get the splited var
+            for i in range(len(varlist)):
+                var = varlist[i]
+                # for ops like clipping and weight decay, get the splited var (xxx.block0)
                 # for inputs/outputs
-                grad_block = self._is_splited_grad_var(
+                grad_block = self._get_pserver_grad_param_var(
                     var, program.global_block().vars)
                 if grad_block:
-                    inputs[key] = grad_block
+                    varlist[i] = grad_block
                 elif var.name not in program.global_block().vars:
-                    program.global_block().create_var(
-                        name=var.name,
-                        persistable=var.persistable,
-                        dtype=var.dtype,
-                        shape=var.shape)
+                    tmpvar = program.global_block()._clone_variable(var)
+                    varlist[i] = tmpvar
+                else:
+                    varlist[i] = program.global_block().vars[var.name]
+            inputs[key] = varlist
 
         outputs = self._get_output_map_from_op(
             self.origin_program.global_block().vars, opt_op)
         for key, varlist in six.iteritems(outputs):
             if not isinstance(varlist, list):
                 varlist = [varlist]
-            for var in varlist:
-                grad_block = self._is_splited_grad_var(
+            for i in range(len(varlist)):
+                var = varlist[i]
+                grad_block = self._get_pserver_grad_param_var(
                     var, program.global_block().vars)
                 if grad_block:
-                    outputs[key] = grad_block
+                    varlist[i] = grad_block
                 elif var.name not in program.global_block().vars:
-                    program.global_block()._clone_variable(var)
+                    tmpvar = program.global_block()._clone_variable(var)
+                    varlist[i] = tmpvar
+                else:
+                    varlist[i] = program.global_block().vars[var.name]
+            outputs[key] = varlist
 
         return optimize_block.append_op(
             type=opt_op.type,