SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2107.00860 · ICLR · 2021

Rapid Neural Architecture Search by Learning to Generate Graphs from Datasets

Sung Hwang, Kaist, South Korea, Hayeon Lee, Eunyoung Hyung

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 19 functions out of this paper's own repositories and ran 2 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.

RepositoryRoleRan
HayeonLee/MetaD2A canonical 1 of 9
D-X-Y/AutoDL-Projects canonical 0 of 5
copy not recorded — 1 of 1
yuhuixu1993/PC-DARTS — 0 of 3
xiangning-chen/DrNAS — 0 of 1
FunctionStatusWhere it lives
channel_shuffle Ran this paper's copy was not recorded; identical code first harvested from ddghost/new_darts
pointer only · get_code("b9da06d4f527dd6c")
collate_fn Ran HayeonLee/MetaD2A/MetaD2A_mobilenetV3/loader.py
code served (permissive licence) · get_code("f55c31d782e48ffb")
Architect Not yet run xiangning-chen/DrNAS/architect.py
pointer only (licence: NONE) · get_code("461bb7a03f47f0ad")
Cell Not yet run yuhuixu1993/PC-DARTS/model_search.py
pointer only (licence: NONE) · get_code("18a2cee21566ce44")
MixedOp Not yet run yuhuixu1993/PC-DARTS/model_search.py
pointer only (licence: NONE) · get_code("f58cf0ad967f722d")
Network Not yet run yuhuixu1993/PC-DARTS/model_search.py
pointer only (licence: NONE) · get_code("4b9ef6a87294337c")
backward_step_unrolled Not yet run D-X-Y/AutoDL-Projects/exps/NAS-Bench-201-algos/DARTS-V2.py
code served (permissive licence) · get_code("490b5c41400816e1")
decode_ofa_mbv3_to_igraph Not yet run HayeonLee/MetaD2A/MetaD2A_mobilenetV3/utils.py
code served (permissive licence) · get_code("7b60cfea1c582309")
get_loader Not yet run HayeonLee/MetaD2A/MetaD2A_mobilenetV3/database/metaloader.py
code served (permissive licence) · get_code("14ef56df31b716ed")
get_meta_test_loader Not yet run HayeonLee/MetaD2A/MetaD2A_mobilenetV3/loader.py
code served (permissive licence) · get_code("2d30a7e1cec63917")
get_meta_train_loader Not yet run HayeonLee/MetaD2A/MetaD2A_mobilenetV3/loader.py
code served (permissive licence) · get_code("75092bae3fb86ad0")
get_transform Not yet run HayeonLee/MetaD2A/MetaD2A_mobilenetV3/process_dataset.py
code served (permissive licence) · get_code("f2c3cad885f3c5a6")
is_valid_ofa_mbv3 Not yet run HayeonLee/MetaD2A/MetaD2A_mobilenetV3/utils.py
code served (permissive licence) · get_code("0de0dc03a084ba91")
load_graph_config Not yet run HayeonLee/MetaD2A/MetaD2A_mobilenetV3/utils.py
code served (permissive licence) · get_code("dac9f0aab7051ab2")
mutate_arch_func Not yet run D-X-Y/AutoDL-Projects/exps/NAS-Bench-201-algos/R_EA.py
code served (permissive licence) · get_code("c9db316e18458ccd")
random_size_func Not yet run D-X-Y/AutoDL-Projects/exps/NATS-algos/random_wo_share.py
code served (permissive licence) · get_code("b45190c694bcb277")
select_action Not yet run D-X-Y/AutoDL-Projects/exps/NAS-Bench-201-algos/reinforce.py
code served (permissive licence) · get_code("0e6f925ca93551fa")
str2bool Not yet run HayeonLee/MetaD2A/MetaD2A_mobilenetV3/parser.py
code served (permissive licence) · get_code("b0717d29ccfec107")
train_and_eval Not yet run D-X-Y/AutoDL-Projects/exps/NAS-Bench-201-algos/R_EA.py
code served (permissive licence) · get_code("1ab9fd355247f264")

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Despite the success of recent Neural Architecture Search (NAS) methods on various tasks which have shown to output networks that largely outperform humandesigned networks, conventional NAS methods have mostly tackled the optimization of searching for the network architecture for a single task (dataset), which does not generalize well across multiple tasks (datasets). Moreover, since such task-specific methods search for a neural architecture from scratch for every given task, they incur a large computational cost, which is problematic when the time and monetary budget are limited. In this paper, we propose an efficient NAS framework that is trained once on a database consisting of datasets and pretrained networks and can rapidly search for a neural architecture for a novel dataset. The proposed MetaD2A (Meta Dataset-to-Architecture) model can stochastically generate graphs (architectures) from a given set (dataset) via a cross-modal latent space learned with amortized meta-learning. Moreover, we also propose a meta-performance predictor to estimate and select the best architecture without direct training on target datasets. The experimental results demonstrate that our model meta-learned on subsets of ImageNet-1K and architectures from NAS-Bench 201 search space successfully generalizes to multiple unseen datasets including CIFAR-10 and CIFAR-100, with an average search time of 33 GPU seconds. Even under MobileNetV3 search space, MetaD2A is 5.5K times faster than NSGANetV2, a transferable NAS method, with comparable performance. We believe that the MetaD2A proposes a new research direction for rapid NAS as well as ways to utilize the knowledge from rich databases of datasets and architectures accumulated over the past years. Code is available at https://github.com/HayeonLee/MetaD2A.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2107.00860")
get_code_for_paper("2107.00860")
have("2107.00860")

Connect an agent — have() is free.