未验证 提交 023742e3 编写于 作者: Q qingqing01 提交者: GitHub

Fix one gpu training in deeplabv3+ (#1966)

上级 1af11949
......@@ -184,7 +184,7 @@ if args.parallel:
build_strategy=build_strategy,
exec_strategy=exec_strategy)
else:
binary = fluid.compiler.CompiledProgram(main)
binary = fluid.compiler.CompiledProgram(tp)
if args.use_py_reader:
assert(batch_size % fluid.core.get_cuda_device_count() == 0)
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册