About StarNet-150, StarNet-100, StarNet-050 pretrained model

[puyiwen]

Hi, thank you for your great work! Could you please provide StarNet-150, StarNet-100, StarNet-50 pretrained model? My adress is [e-mail]. Thank you very much again!

[Xu Ma]

@puyiwen Thanks. We mainly focus on s1,s2,s3,and s4 variants, and the smaller models / checkpoints are not relseased. I really need, we can retrain later and update.

[puyiwen]

@puyiwen Thanks. We mainly focus on s1,s2,s3,and s4 variants, and the smaller models / checkpoints are not relseased. I really need, we can retrain later and update.

ok, thank you for your reply. I want to compare the downstream task performace of mobilenetv2-0.75 and starnet-150, so I need pretrained model. If you can update, I would be grateful.

[puyiwen]

@ma-xu Sorry to bother you again. If I have only 4 GPU, how can I train the starnet? Should I need to change some parameters, like lr, batch-size?

[The-HHEE]

Hi @ma-xu, I also have the request about the checkpoint of starnet050, as we want to evaluate it on resource-constraint devices. Thank you!

[Xu Ma]

@KokyoK @puyiwen Hi, thanks for your message.

Let me train these models on this weekend since there are some GPUs available, and probably will update the link to checkpoints by Wednesday. If not, please feel free to directly send me an email. Thank you!