Soft Parameter Sharing

Reported on 6 benchmarks across 2 tasks · 1 paper · 2 SOTA

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Methodology6 results

Neural Architecture SearchonCIFAR-10 Image Classification
Params· 2019-02-26
33.5
SOTA
Learning Implicitly Recurrent CNNs Through Parameter Sharing arXiv:1902.09701
AutoMLonCIFAR-10 Image Classification
Params· 2019-02-26
33.5
SOTA
Learning Implicitly Recurrent CNNs Through Parameter Sharing arXiv:1902.09701
Neural Architecture SearchonCIFAR-10 Image Classification
Percentage error· 2019-02-26
2.53
best: 1.6 (NAT-M4)
Learning Implicitly Recurrent CNNs Through Parameter Sharing arXiv:1902.09701
Neural Architecture SearchonCIFAR-10
Search Time (GPU days)· 2019-02-26
0.7
best: 224 (AlphaX-1 (cutout NASNet))
Learning Implicitly Recurrent CNNs Through Parameter Sharing arXiv:1902.09701
AutoMLonCIFAR-10 Image Classification
Percentage error· 2019-02-26
2.53
best: 1.6 (NAT-M4)
Learning Implicitly Recurrent CNNs Through Parameter Sharing arXiv:1902.09701
AutoMLonCIFAR-10
Search Time (GPU days)· 2019-02-26
0.7
best: 224 (AlphaX-1 (cutout NASNet))
Learning Implicitly Recurrent CNNs Through Parameter Sharing arXiv:1902.09701