Partial Order Pruning: for Best Speed/Accuracy Trade-off in Neural Architecture Search
snapshots:
df1.caffemodel https://drive.google.com/open?id=1yA9DLSy3PEMQD3R92vKr6CEaOJ0NrOVm
df2.caffemodel https://drive.google.com/open?id=1K0QPFD6XtKnMsrrOLSarnk3C1zmhZqpC
df2a.caffemodel https://drive.google.com/open?id=1H5T-nz1D2DCLtma-alkR_CxGAHKewbql
df1seg.caffemodel https://drive.google.com/open?id=1v-UCb1VIHGtIR9eXiPgdUu9wkDpemNTW
df1seg_mergebn.caffemodel https://drive.google.com/open?id=17ZROC9dJAN8dxkpvTzHQRTS9soCwBOXJ
df2seg1.caffemodel https://drive.google.com/open?id=1mCdozRO4BxDV-NS6secKuvaTNWEFcv_u
df2seg1_mergebn.caffemodel https://drive.google.com/open?id=1RfdYtc7YzM5zYoANsRiB-XJMkvYfy_mv
df2seg2.caffemodel https://drive.google.com/open?id=1D7bgq7h9OUQVY4LYA-x0B2FY8pnVgz3o
df2seg2_mergebn.caffemodel https://drive.google.com/open?id=1FtqRSEN90ynTgMeGH3ee5DC8ubvSHZ3z
df-lite_seg_mergebn.caffemodel https://drive.google.com/open?id=1se9wAkZFyNGYInjrhtXTMIZy39ucMaDu