ankur6ue

yep.. The Wc and U matrices in eq8 in the original paper are missing as well.

Update: I tried using the latest beta version of kops (1.23.0-beta.2), that uses an updated version of calico as well as the fix described here: https://github.com/kubernetes/kops/issues/13095 With this, my kubernetes...

Having the same issue and this is very annoying. log_to_driver=False does nothing and the console is full of ray system logs making it impossible to see the logs emitted by...

Tagging @JethroMV because I borrowed my values.yaml from his issue.

I tried a different CNI - calico and after setting the host interface to cali+, kiam works. So for whatever reason, the default CNI on kops (cbr0 on kubenet?) doesn't...

Can you review the following code for loss calculation when batch size > 1? idx = batch['sum_idx'] b = logits.shape[0] loss = 0 for i in range(0,b): shift_logits = logits[i,...

same here.. applied few days ago, no response. Odd..

This happens in the middle of the training. The full python stack trace is here: >>> rpn_loss_box: 0.032325 >>> loss_cls: 0.576470 >>> loss_box: 0.398267 >>> lr: 0.001000 speed: 0.460s /...

I haven't yet, as I got looped into other stuff, but basically what is happening is that neither foreground nor background regions are being found for that image so it...

I'm facing very much the same issue with pytorch and torchrec.. I'm using pytorch version 2.6.0+cu124 on cuda 12.4 OSError: /home/ankur/miniconda3/envs/py3_12/lib/python3.12/site-packages/fbgemm_gpu/fbgemm_gpu_config.so: undefined symbol: _ZN5torch3jit17parseSchemaOrNameERKNSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEEb installed torchrec following instructions here: https://github.com/pytorch/torchrec...