zhiics commented on a change in pull request #5077: [Refactor][Relay] Refactor Relay Python to use new FFI URL: https://github.com/apache/incubator-tvm/pull/5077#discussion_r393287385
########## File path: python/tvm/relay/__init__.py ########## @@ -19,51 +19,56 @@ import os from sys import setrecursionlimit -from . import call_graph from . import base from . import ty from . import expr from . import type_functor from . import expr_functor from . import adt -from . import analysis +from . import prelude +from . import loops +from . import scope_builder +from . import parser + from . import transform +from . import analysis +from .analysis import call_graph, feature, alpha_equal from .build_module import build, create_executor, optimize from .transform import build_config -from . import prelude -from . import parser from . import debug from . import param_dict -from . import feature from .backend import vm # Root operators from .op import Op +from .op import nn +from .op import image +from .op import vision +from .op import annotation from .op.reduce import * from .op.tensor import * from .op.transform import * from .op.algorithm import * -from . import nn -from . import annotation -from . import vision -from . import contrib -from . import image +from .op.nn import * Review comment: Yeah, the reason I import them was because we widely used relay.conv2d in the code. We had a nn.py before which basically does the same thing as `import *` here. Same to other cases. Any other ways to achieve this? ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
