def genotype(self):
n, start = 3, 2
weightsr2 = [F.softmax(beta, dim=-1) for beta in self.beta_reduce[0:2]]
weightsn2 = [F.softmax(beta, dim=-1) for beta in self.beta_normal[0:2]]
for i in range(self.n_nodes):
end = start + n
tw2 = [
F.softmax(beta, dim=-1) for beta in self.beta_reduce[start:end]
]
tn2 = [
F.softmax(beta, dim=-1) for beta in self.beta_normal[start:end]
]
start = end
n += 1
weightsr2 = weightsr2 + tw2
weightsn2 = weightsn2 + tn2
# normalize edge-leve beta parameters
gene_normal = gt.parse(self.alpha_normal, weightsn2, k=2)
gene_reduce = gt.parse(self.alpha_reduce, weightsr2, k=2)
concat = range(2, 2 + self.n_nodes) # concat all intermediate nodes
return gt.Genotype(normal=gene_normal,
normal_concat=concat,
reduce=gene_reduce,
reduce_concat=concat)
def genotype(self):
gene_normal = gt.parse(self.alpha_normal, k=2)
gene_reduce = gt.parse(self.alpha_reduce, k=2)
concat = range(2, 2+self.n_nodes) # concat all intermediate nodes
return gt.Genotype(normal=gene_normal, normal_concat=concat,
reduce=gene_reduce, reduce_concat=concat)
def genotypetoaugment(self, layers_augment=20):
genotype = {}
genotype["normal"] = gt.parse(self.alpha_normal, k=2)
genotype["reduce"] = gt.parse(self.alpha_reduce, k=2)
for n in range(layers_augment):
if n == layers_augment // 3 or n == 2 * layers_augment // 3:
genotype["cell_%d" % n] = genotype["reduce"]
else:
genotype["cell_%d" % n] = genotype["normal"]
genotype["normal_concat"] = range(
2, 2 + self.n_nodes) # concat all intermediate nodes
genotype["reduce_concat"] = range(
2, 2 + self.n_nodes) # concat all intermediate nodes
return genotype
def genotype(self):
with torch.no_grad():
gene_normal = parse(weights=F.softmax(self.alphas_normal, dim=-1),
num_nodes=self.num_nodes,
k_strongest=2)
gene_reduce = parse(weights=F.softmax(self.alphas_reduce, dim=-1),
num_nodes=self.num_nodes,
k_strongest=2)
concat = range(2, self.num_nodes + 2)
genotype = Genotype(normal=gene_normal,
normal_concat=concat,
reduce=gene_reduce,
reduce_concat=concat)
return genotype
def DAG(self):
gene_DAG1 = gt.parse(self.alpha_DAG[0 * self.n_big_nodes:1 *
self.n_big_nodes],
k=2)
gene_DAG2 = gt.parse(self.alpha_DAG[1 * self.n_big_nodes:2 *
self.n_big_nodes],
k=2)
gene_DAG3 = gt.parse(self.alpha_DAG[2 * self.n_big_nodes:3 *
self.n_big_nodes],
k=2)
concat = range(self.n_big_nodes, self.n_big_nodes + 2)
return gt.Genotype2(DAG1=gene_DAG1,
DAG1_concat=concat,
DAG2=gene_DAG2,
DAG2_concat=concat,
DAG3=gene_DAG3,
DAG3_concat=concat)
def DAG(self):
gene_DAG1 = gt.parse(self.alpha_DAG[0 * self.n_big_nodes:1 *
self.n_big_nodes],
k=2)
gene_DAG2 = gt.parse(self.alpha_DAG[1 * self.n_big_nodes:2 *
self.n_big_nodes],
k=2)
gene_DAG3 = gt.parse(self.alpha_DAG[2 * self.n_big_nodes:3 *
self.n_big_nodes],
k=2)
# concat = range(2, 2+self.n_big_nodes) # concat all intermediate nodes
# concat = range(2+self.n_big_nodes-2, 2+self.n_big_nodes)
concat1 = gt.parse_concat(self.alpha_concat[0])
concat2 = gt.parse_concat(self.alpha_concat[1])
concat3 = gt.parse_concat(self.alpha_concat[2])
return gt.Genotype2(DAG1=gene_DAG1,
DAG1_concat=concat1,
DAG2=gene_DAG2,
DAG2_concat=concat2,
DAG3=gene_DAG3,
DAG3_concat=concat3)
def genotype(self):
gene_normal = gt.parse(self.alpha_normal, k=2)
concat = range(1, 1 + self.n_nodes) # concat all intermediate nodes
return gt.Genotype(normal=gene_normal, normal_concat=concat)
