Ziteng Gao

Hi, thanks for your interest in our work. Now I am also awaiting the Ant Group to release models since I run the models in the company and the company...

> Hi, so when will you release the trained models? thanks! > > > Hi, thanks for your interest in our work. Now I am also awaiting the Ant Group...

> I reproduce `adamixer_r50_1x_coco.py` using `mmcv_full==1.3.9` and `mmcv_full==1.3.3` respectively. Both `mmcv_full==1.3.9` and `mmcv_full==1.3.3` yield 42.3 mAP, which is 0.4 points lower than the number (42.7 mAP) reported in the paper....

It's ` th main.lua -depth 50 -batchSize 85 -nGPU 4 -nThreads 8 -shareGradInput true -data [imagenet-folder] -dataset imagenet -LR 0.033 -netType resnetdpp -poolingType DPP_sym_lite ` The variant of DPP seems...

The burden 4.14G Flops to 6.59G Flops is for the ResNet-50 one. And 7.89G Flops to 10.32G Flops for ResNet-101.

您好，这个实验我们还没有严格公平对比过，所以没有确切的结论。不过query denosing理论上也可以用在adamixer上的。