Is there no code about guilded mutation in esr-ea or I missed it ?

[32l]

Hi, thanks for your awesome work, but i doubt whether the implement of EA's mutation is correct, the paper says that we should acquire the block credits during model evaluation procedure, which can be used to guide the mutation to accelerate searching and find better architecture, I find that vega's implementation is general mutaion, could you help me? thanks!

[zhangjiajin]

Please check file: vega/algorithms/nas/esr_ea/esr_ea_individual.py

    def mutation_node(self, mutation_rate=0.05):
        """Mutate the active node type of individual.
        :param mutation_rate: the prosibility to mutate, defaults to 0.05
        :type mutation_rate: float
        """
        for node_ind in range(self.node_num):
            if self.gene[node_ind][0] and np.random.rand() < mutation_rate:
                type_prob = self.net_info.func_prob
                self.gene[node_ind][1] = bisect_right(type_prob, random())
        self.active_net = self.active_net_list()
        self.parameter = self.network_parameter()
        self.flops = self.network_flops()

It's guided mutation, not normal mutation.
The func_prob is guide information.

[32l]

ok, thanks. I will double check the code.