nmll

Results 10 comments of nmll

multiple GPU

Will there be a drop when training ReID triplet loss with multi-GPU compared to the single GPU training?

CUDA kernel failed : invalid device function

you can use gdb

In multi_cfg.py, why num_class=1 for extra_head ?

why the featmap_stride=0.4？

Are you not going to make a n-pairs loss miner?

when use the N pair loss，it is increasing.may be which reason?

Are you not going to make a n-pairs loss miner?

Thanks，my learning rate is decaying.I do not use margin and miner,I use cos distance between the objects and then use the N pair loss,but the loss is increasing, I do...

Are you not going to make a n-pairs loss miner?

@KevinMusgrave I pass into N-pairs loss the (anchors, labels,(anchor_index,pos_index,neg_index)),is the function required.

Are you not going to make a n-pairs loss miner?

Thanks a lot!

What is emb_scale used for?

I also have the question about that why to use this operation about self.emb_scale? id_head = self.emb_scale * F.normalize(id_head)

training time

@zzm-hl Have you solved this problem? I train the first stage on 8*3090 bs=2/gpu by 4days with too much time. My part training logs are: import DCN failed import DCN...

ImportError: /usr/local/PyCeres.cpython-37m-x86_64-linux-gnu.so: undefined symbol: cudaDeviceSynchronize

same problem