Difference between GDAS variants

D-X-Y / AutoDL-Projects

Automated deep learning algorithms implemented in PyTorch.

MIT License

1.56k stars 281 forks source link

Difference between GDAS variants #119

Open D-X-Y opened 2 years ago

D-X-Y commented 2 years ago

Which Algorithm GDAS: Searching for A Robust Neural Architecture in Four GPU Hours

Describe the Question In search_cells.py, there are two kinds of GDAS implementations:

First, self._ops[index](x) * weights[index]
Second, weights[_ie] * edge(nodes[j]) if _ie == argmaxs else weights[_ie] for _ie, edge in enumerate(self.edges[node_str])

D-X-Y commented 2 years ago

Make a commitment to ablatively study this issue.

D-X-Y commented 2 years ago

TODO:

Run python ./exps/NATS-algos/search-cell.py --dataset cifar10 --data_path $TORCH_HOME/cifar.python --algo gdas_v1 --rand_seed 777 and python ./exps/NATS-algos/search-cell.py --dataset cifar10 --data_path $TORCH_HOME/cifar.python --algo gdas --rand_seed 777 to see the performance difference.

Xiao-congxi commented 1 month ago

May I ask the author if you have conducted experimental comparisons between these two implementations? Will they bring significant differences?