Chenxiao Niu

I found that in the initialization function, weights are initialized to zero, which may cause the problem. Does anyone knows why the fcn8s need to initialize all the weights to...

just update Flask version in web/requirements.txt, let's say Flask==2.0.0

> > just update Flask version in web/requirements.txt, let's say Flask==2.0.0 > > Hi .. Thank you it worked .... There was another problem .. there is a version conflict...

> > > just update Flask version in web/requirements.txt, let's say Flask==2.0.0 > > > > > > Hi .. Thank you it worked .... There was another problem .....

Hi, Does the initialization of IDAUp matters while training CenterNet? I re-implemented one model started training from scratch. The loss-hm plateaus to 9.208 and the network stopped converging. To be...

 llama+lora同样在胡言乱语，请问如何评判模型的准确度和效果呢？

> 请问你们是什么场景，为什么不能内存扩大一点呢？ 假设不能保证用户测试主机内存足够加载一定模型，上述错误导致程序异常推出，没有足够多的信息定位。 期望或许可以从两个方面解决： 1. 优化权重加载的逻辑 2. 给出正确的报错信息，引导用户调试

> 也提到develop分支吧 link to [PR-7115](https://github.com/PaddlePaddle/PaddleDetection/pull/7115)

> 是否可以https://github.com/PaddlePaddle/PaddleDetection/blob/release/2.5/docs/tutorials/GETTING_STARTED_cn.md 这个文档中补充下文档，比如使用方法，参数列表的部分，也加下什么这个命令现在支持了哪些硬件 看了一下这个是PaddleDetection的说明文档，其中的示例指令都是使用的gpu设备。mlu的使用方法和其他异构设备（xpu，npu）类似，这次只对device的传入部分逻辑做修改，并未新增单独的使能接口。另外，文档中没有发现有异构设备的说明部分，如果在gpu指令后面新增部分mlu的说明可能会影响整个文档的行文逻辑。

> NVIDIA is publishing TF1 and TF2 containers by the end of the month that comes with 20.04 and TRT8 support. > > Search for "Nvidia NGC tensorflow container", the...