-
Would it be possible for you to release the trained ResNet-50_Arcloss model in PyTorch (.pt) format? It would be easier to perform operations like quantization.
Thanks
-
![2019-11-18 17-46-58 的屏幕截图](https://user-images.githubusercontent.com/28854034/69042086-93564f00-0a2b-11ea-9167-3b6e1addcf84.png)
I can not understand the easy_margin, why the cosine >0, we cho…
-
您好,感谢您的工作,最近训练模型时用到了arcloss,得到的loss值是nan,请问您有遇到这个问题吗?另外cosloss看代码实现没有看到跟余弦、正弦相关的 计算,这个感觉跟定义不太一样。希望能得到您的解答!
-
INFO:root:Epoch[24] Batch [520-540] Speed: 1727.27 samples/sec acc=0.983984 lossvalue=2.195642
INFO:root:Epoch[24] Batch [540-560] Speed: 1288.51 samples/sec acc=0.983203 loss…
-
为什么把loss设置成arcloss之后,loss为87.33,accuracy为1?
layer {
name: "cosin_add_m"
type: "CosinAddm"
bottom: "concat_fc"
bottom: "label"
top: "fc6_margin"
cosin_add_m_param {
m: 0.5
}
…
-
what's the tpr when far is at 0.0001%?
-
郑博士以及各位大神好,代码出现点问题,我一个EPOCH都没跑玩,准确率为1,损失基本为0:
条件是:--name SE_imbalance_s1_384_p0.5_lr2_mt_d0_b24+v+aug --warm_epoch 5 --droprate 0 --stride 1 --erasing_p 0.5 --autoaug --inputsize 384 --lr 0.02 --us…
-
I was trying to train a mobilenet model using arcloss on casia dataset and I am unable to exceed lfw acc more than ~0.9763.
I wanna know is this the best accuracy I can get using this dataset(casi…
-
Hello! In script https://github.com/Ant-Brain/EfficientWord-Net/blob/main/training.ipynb you are referring to
`baseNetwork = tf.keras.models.load_model("/content/drive/MyDrive/Siamese/modelCheckpoin…
-
### Discussed in https://github.com/Lightning-AI/lightning/discussions/13594
Originally posted by **manideep2510** July 11, 2022
Hi, I'm trying to use FSDP training with Pytorch Lightning for …