Duo Li
Duo Li
@andyhahaha Thanks a lot for reminding me, and I got the split from SegNet repo. But there exists some conflicts with the [CamVid official site](http://mi.eng.cam.ac.uk/research/projects/VideoRec/CamVid/) as far as i know:...
@andyhahaha Furthermore, as for your original question, I think shrinking `n_class` to 11 naively won't work, because if so, the network will output 11 classes while the ground truth still...
@bilylee Hi, I tried this way, but still got little improvement in the 5-shot scenario (specifically 67.1%). This is a code snippet ```python class Convnet(nn.Module): def __init__(self, x_dim=3, hid_dim=64, z_dim=64):...
@speedinghzl Sorry, I have the same issue, my program is even stuck in 4xTitan Xp with the default settings.
@aasharma90 Thanks for your kind advice! Shrinking the batch size can definitely fit the model into GPU memory with ease, but we have to sacrifice the performance as demonstrated in...
Actually, 4x12g VRAM is not enough. I have run the `run_local.sh` with 4 Tesla M40 GPUs, the memory usage of GPU 0 is over 12800M. Without modifying any default settings...
Hi, @thomasverelst Thanks for your prompt reply and sharing! I have realized your concern about the computational resource, but two weeks is still a fairly long experimental period :). Furthermore,...
Yeah, it seems to work now. I have successfully run this branch with ResNet-32 on CIFAR for fast prototyping (with matched accuracy and reduced FLOPs). As an additional note, the...
It should work after modifying certain nested structures, refer to #22. @bkj
The same.