jiaxinc

Is it a bug that calling`fromHost` without `vkFlushMappedMemoryRanges`

2

Hi, I was trying to understand how vulkan works from your src code. When I looked into file: `deviceArray.hpp` in branch: `vuh2`, ``` auto fromHost(It1 begin, It2 end)-> void {...

Hi, Nice job, but I'm confused that some GAN implementation use direct loss of VGG deep feature map but not `lpips`, I try both these kind of perceptual loss and...

  我的SS服务器使用libev 3.0.7, MacOS上的client连接正常，但是在iOS上跑的时候发现请求都失败了，看上去像是数据包错乱了? 不知道可能是什么原因谢谢了

Fix fp16 arithmetic detection for Cortex-X1 device

2

FP16 arithmetic detection for Cortext-X1 is not covered. CPUs of type Cortext-X1: Snapdragon 888, Samsung Exynos 2100

https://github.com/JDAI-CV/DNNLibrary/blob/e17f11e966b2cce7d747799b76bb9843813d4b01/quant.py#L269 Name of input tensor `data` is presumed which is not appropriate for most other model. https://github.com/JDAI-CV/DNNLibrary/blob/e17f11e966b2cce7d747799b76bb9843813d4b01/quant.py#L216 Also there shouldn't be any presume that model needs normalization with mean/dev.

1. According to section 3 of NeRF [paper](https://arxiv.org/abs/2003.08934): > We encourage the representation to be multi-view consistent by restricting the network to predict the volumn density σ as a function...

DMTet rendering: alpha and mask usage

1

Hi, @ashawkey Be grateful for your great job. After some trials of reading the code related to the latest DMTet feature. I am a little confused about the alpha/mask usage...

2.12.6 随机梯度和批量梯度区别 这里标注有问题，应该是`SGD`、`BGD`而不是`SDG`、`BGD` 其次 > 1、 批量梯度下降的求解思路如下 ... 它得到的虽然是一个全局最优解，但每迭代一步... BGD 得到的并不一定是全局最优解，跟参数初始化、学习率都有关系 如有错误请指正

图片打印不出来

3

想要打印二维码来着， - (void)appendQRCodeWithInfo:(NSString *)info 调用了此方法，打印机打到这里就不动了。 然后作者的添加图片代码我取消注释 [printer appendImage:[UIImage imageNamed:@"ico180"] alignment:HLTextAlignmentCenter maxWidth:300]; 貌似还是打印不出来，不知道是打印机的问题还是什么？？ 感谢作者做出这个强大的类库，希望一起进步

Not convergent in custom dataset.

9

Hi, glad to see the impressive project. I overload the `trainer.py` according to `README` and the training works properly. However, the model doesn't seem to converge and loss is always...