Fred

----> 9 mean = x.mean(-1, keepdmin=True) keepdmin is a typo, the correct argument is `keepdim`

i don't think it is polite to leave your information on other's repo

> (I should clarify - my understanding is that the resources _will_ be freed by the GC when the variable is reassigned on the next step. We're not leaking significant...

A random thought. As a community, would be great to have a discord channel for discussions and updates.

@yair-schiff I am just curious, did you solve the > What if the sequences have paddings? E.g. Input is [1 2 3 0 0 0] So flipped input would be...

I came up with a solution to the padding issue. Say a tensor [1,2,3,0,0], where 0 is the padding token. We flip it to get [0,0,1,2,3], pass it to the...

I don't get your question... Why EMA affects data? Isn't it only tuning the model params?

Based on my experiments, the utilization is very low. Most of the time, the GPU is not computing anything but waiting. for input of batch size 6, img size 128*128*3,...

I think the reason is that the data loading is slow. Because many image preprocessing operations are applied when load an img from dataset.

This is my docker container ``` nvcr.io/nvidia/clara/bionemo-framework:latest "/workspace/bionemo/…" 4 days ago Up 13 hours bionemo ```