Hi friends: Currently, I am trying to use batch norm in relay api , however, there are parameters which are a little different from batch norm of pytorch api. Here is relay api:
tvm.relay.nn.batch_norm(data, gamma, beta, moving_mean, moving_var, axis=1, epsilon=1e-05, center=True, scale=True)
I am wondering how to initialize moving_mean and moving_var if we just view it as a normal batch norm ? Thank you for help!
Actually， I am a little confused about moving_mean and moving_var batch norm