Vincent XU

what if I use another 3DMM model such as BFM2017 ?

11

Hi, The model comes from https://gravis.dmi.unibas.ch/PMM/ could you plz have a look at this model ?

If people use `numpy.core.memmap.memmap` to save arrays then it will crash. use 'numpy' as a check point will be ok.

What about liver lesion segmentation?

1

Hello Bickel, I'm reading your codes in these days and learned many from these great codes and thoughts. But I noticed that in this 'Issues' discussion board we all talk...

I noticed that there're some segmentation models that used modified resnet as backbone. E.g. [this](https://github.com/wuhuikai/FastFCN/blob/master/encoding/dilated/resnet.py#L139-L148) network. I think it's better to train such a backbone on my own to accomplish...

RT. Need a well trained Dilated ResNet-101 backbone PSPNet checkpoint. But currently we do not have too much GPU resources. I think its mIU will be around 78% in multi-scales...

Please review this pr about a new implementation of tf::Semapohre and its RAII manage method tf::SemapohreGuard. I kept the original implementation of tf::Semaphore for compatibility and some-cases' performance.

https://github.com/snuspl/nimble/tree/main_pytorch_v1.7.1/third_party

Hello developers, I'm inspecting smoothquant and use the script below to check the quantized model parameter sizes: ```python from smoothquant.opt import Int8OPTForCausalLM from transformers.models.opt.modeling_opt import OPTForCausalLM import torch model_name =...