-
resnet+rotp我在本地的掌纹数据库中测试,把您的resnet代码#直接实例化ResNet18,不替换任何层,就是把main_worker函数里dim = 512,model = replace_pooling(model, k=args.k, dim=dim, f_method=args.f_method)注释掉之后结果为* Acc@1 96.412 Acc@5 99.265
Epo…
dyqii updated
3 weeks ago
-
Hi,
Just out of curiosity, what is the fundamental difference between the ResEnc M, ResEnc L and ResEnc XL in architecture wise? Does it change the depth of the network?
-
The resnet-50 weights that you have shared is any random resnet-50 weights or you first trained it?
-
-
@lhnguyen102 Hi Ha,
I am not sure if it is supposed to be like this, but I believe there is a memory leakage on the ResNet architecture in CUDA.
While running the resnet18_cifar10 example in th…
-
As #197 and #206 have pointed out, there are some inconsistencies with the implementation of Inception-ResNet-v2 and [the paper](https://arxiv.org/pdf/1602.07261). @agsourav highlights that the stem s…
-
-
현재 최대 acc는 valid / test 기준 64-66%를 웃돈다.
현재 신경망은 단순 conv layer만을 이용하는 비교적 단순한 신경망이기에 깊이를 늘려보고자 한다. 다만, 6층부터는 그라디언트 소실 문제가 발생하기 때문에 Resnet을 이용해 문제를 해결하고자 한다.
-
Hello, what is the `strd` and `multiplier` do here? what are the use for?
```
class BasicBlock(nn.Module):
multiplier=1
def __init__(self, input_num_planes, num_planes, strd=1):
…
-
Hi all,
I was testing out this project and noticed that when I run `cargo run --example inference --features tch-gpu ./samples/dog.jpg` it causes a crash with the error: `unsafe precondition(s) viola…