Implementing AssembleNet: Searching for Multi-Stream Neural Connectivity in Video Architectures Explain using Pytorch (Work In Process)
- Paper Link
- Author: Michael S. Ryoo (Robotics at Google, Google Research), AJ Piergiovanni (Robotics at Google, Google Research), Mingxing Tan (Google Research), Anelia Angelova (Robotics at Google, Google Research)
- Organization: Robotics at Google, Google Research
- Make graph
from make_graph import Graph
import pprint
p = pprint.PrettyPrinter(width=160, indent=4)
g = Graph()
p.pprint(g.grpah)- Make Network
from make_graph import Graph
import pprint
g = Graph()
m = Model(g.graph)
pprint.pprint(m.graph, width=160)
x = torch.randn([2, 3, 16, 256, 256])
print(m(x).size())- Network Evolution
from make_graph import Graph
import pprint
g = Graph()
m = Model(g.graph)
pprint.pprint(m.graph, width=160)
m._evolution()
pprint.pprint(m.graph, width=160)
x = torch.randn([2, 3, 16, 256, 256])
print(m(x).size())
- Connection-Learning-Guided Mutation
- Evolution
- Training