Dan Hendrycks

@jacobgil I can confirm. Here is the main code modification I made before encountering this issue. ```python em = cv2.ml.EM_create() em.setClustersNumber(N) em.trainEM(descriptors) return np.float32(em.getMeans()), np.float32(em.getCovs()), np.float32(em.getWeights())[0] ```

On colab there must not have been a directory name, as directories on drive might be structured unusually. Best, Dan Hendrycks On Tue, Jun 21, 2022 at 7:07 PM lazir0lufias...

If I recall correctly, the whole network is modified. On Sun, Dec 12, 2021 at 3:33 AM Hongxin Wei ***@***.***> wrote: > Hi, > > In the paper, it writes:...

This is a very low-level issue, and unfortunately "NCCL Error 1: unhandled cuda error" means that even CUDA does not know what it is. I could only suggest updating drivers...

Just load the tar as a checkpoint and it will work.

It's convenient for mapping back and forth between [-1,1], and [0,1]. On Mon, Apr 10, 2023 at 2:17 AM futawaseda ***@***.***> wrote: > > https://github.com/hendrycks/pre-training/blob/10cd02a3c21698dfc790a50fe442fddf417ad17b/robustness/adversarial/train_imagenet.py#L127 > > Hi, I assume...

Tiny ImageNet just has 200 of the ImageNet-1K classes. Some might use the 800 classes as OE data.

Should be in pytorch's vision library. On Mon, Mar 11, 2024 at 6:32 AM Guru Swaroop ***@***.***> wrote: > I have noticed that the pre-trained weights for the models are...