xray1111
xray1111
@kba May I ask why did this feature be removed? It seems like useful while processing some domain specific documents.
@amitdo Thank you! Maybe I'll dig into int some day...
@nilboy Thanks! I've changed the api and it runs ok. Another question: How to use the model I've trained in the demo.py? I try to change the line below but...
Also meet this problem, when attach a remote process id using gdb: vscode version: 1.51.1 c++tools Extension version: 1.1.3 OS version: macos 10.14.6 Remote OS version: ubuntu 18.04.1 Remote gdb...
@Leavingseason Sorry for late response, that problem may caused by a wrong when calcuating RMSE, I changed the code then it's fine. Thanks!
我也有这个OOM的问题，请问最后解决了吗？我这里用了4块A800(80G)，主机内存512G，把batchsize减半后也是跑20个step左右会把内存耗完然后报错 @huyuelin
请问你的主机环境是多少G的内存呢？ @huyuelin
你的问题是因为GPU显存不足吗？我这里是主机内存占用太多造成的，我主机512G，我的GPU是A800，每块80G显存，我看报错不是CUDA显存不够，而是主机内存被占用太多。 @huyuelin