echo comments

Results 10 comments of


                                            echo

微信二维码不能用了，能重新发一下吗

anyone here?I need QR code,could you help me

This convolution is not supported by cudnn, MXNET convolution is applied

because dilation convolution is not support in cudnn5,using cuda8+cudnn7,install mxnet from src will be ok

Running GPU tests got Some or all tests fail

try to cout some info in inf_test(): cost:0 grads:-0.525744 grads:0.0961584 grads:-0.399427 grads:0.0964459 grads:0.0771148 grads:0.0607103 grads:0.0556602 grads:0.0437479 grads:0.0542926 grads:0.0666461 grads:0.093121 grads:0.0668003 grads:0.0612226 grads:0.0953744 grads:0.0579237

PicoDet训练报错：(External) CUDA error(700), an illegal memory access was encountered.

测试2.3.2 cuda11.6存在同样问题，2.2.2+cuda11.2版本正常

明明显存还有很多，但他报错OSError: (External) CUDA error(700), an illegal memory access was encountered.

picodet 一样的问题

为什么我生成的图片里面的中文只有方框，是不是因为encoding的问题？

I got the same error

Why do inference results differ when setting different nproc_per_node

> CUDA_VISIBLE_DEVICES=0,1代表可见性，main.py这个程序有权利使用两张显卡，但你设置--nproc_per_node=1致使程序只用了一张卡(卡0)测试，当你--nproc_per_node=2程序会将它看见的卡0和卡1两张卡都用来测试，你用两个卡测试肯定会比一个卡测试指标高，我用八张卡测试比用四张卡测试指标也高，并且测试喂入batch数也会影响指标(作者原始batch数最佳)。上述仅为我个人观点。同一个模型，同一个样本，推理结果唯一，指标不唯一？

Update base_model.py

> Hi, would you open an issue to discuss this problem? Since we have no plan to train OpenGait in a multi-machine manner, it may be hard to justify your...