fix

python/paddle/distributed/fleet/meta_optimizers/sharding/offload_helper.py

@@ -41,7 +41,7 @@ class OffloadHelper(object):
             idx,
             type='cast',
             inputs={'X': src_var},
-            outputs={'Y': dst_var},
+            outputs={'Out': dst_var},
             attrs={
                 'in_dtype': src_var.dtype,
                 'out_dtype': dst_var.dtype,
@@ -166,7 +166,7 @@ class OffloadHelper(object):
 
                 assert param in param_to_fp16
                 fp16_param_name = param_to_fp16[param]
-                fp16_param_var = block.var[fp16_param_name]
+                fp16_param_var = block.var(fp16_param_name)
                 fp16_param_var.persistable = True
                 self._insert_cast_op(block, idx + 1, param,
                                      param_to_fp16[param])
@@ -177,7 +177,7 @@ class OffloadHelper(object):
 
             # step3.4: remove cast op
             if op.type == 'cast':
-                input_name = op.desc.input_arg_names[0]
+                input_name = op.desc.input_arg_names()[0]
                 if input_name in param_to_idx:
                     block._remove_op(idx, sync=False)
                     continue