查了一下paddle.fluid将来版本会废弃,自从paddlepaddle 2.1之后is_compiled_with_rocm就在paddle.device下面了,我先改成了device还是报错说明引擎是2.0的。并且is_compiled_with_rocm是paddlepaddle为了适配amd和海光的gpu的,跟nvidia的显卡关系不大。把相关的两个if not 注释掉就可以了 paddle-bot-oldclosed this as ...
CosineAnnealing 这个api 已经在我们排期优化中；
from paddle.fluid.dygraph.baseimportto_variable from paddle.fluidimportdygraphasdygraphwithfluid.dygraph.guard():x=np.random.randn(1,32,24,24).astype('float32')x=to_variable(x)x1,x2,x3,x4=fluid.layers.split(x,num_or_sections=4,dim=1)print("X1 shape is ",x1.shape)print("X2 shape ...