search

ruocwang / darts-pt

[ICLR2021 Outstanding Paper] Rethinking Architecture Selection in Differentiable NAS
Apache License 2.0
102 stars 14 forks source link

Performance in Subspace #3

Open zerocostpt opened 3 years ago

zerocostpt commented 3 years ago

Hi, thanks for the good work, and for the performance in subspace. Are you run each experiment four times with 4 random seeds and select the best to report? Just as what previous work did? This part not be illustrated on paper, just wanna check if it's right.

ruocwang commented 3 years ago

Hello. That is correct; for S1-S4 subspaces we report the best result following the original RobustDARTS, to encourage fair comparison with its baselines. (For the reference of other fellows that might check this thread later: on the original DARTS Space, we report both average and best results of running the whole search algorithms under different seeds, to establish direct comparisons with all previous methods regardless of their adopted reporting schema.)