Failed to call ThenRnnForward

第一次出现这个错误，通过降低n卡版本解决了，过程很曲折。。

没想到今天又出现了这个错误。感觉可能是我的笔记本的gpu显存太低了（gtx 1050,4g），训练参数多的网络就不太行了。。

 

Probably your are running out of memory on the gpu. 

https://stackoverflow.com/questions/53972814/cudnnlstm-failed-to-call-thenrnnforward

设置gpu内存自增长（注意：要写到import tensorflow as tf的后面，不要写到其他位置）

有用，谢谢之后还报了个Failed to call ThenRnnForward with model config:的错，把batch_size调小一点后解决了

https://blog.csdn.net/weixin_44493841/article/details/104032615#comments_12917030

 

===第一次的笔记===

RuntimeError: Physical devices cannot be modified after being initialized
https://cloud.tencent.com/developer/ask/230462

Failed to call ThenRnnForward
https://github.com/tensorflow/tensorflow/issues/40403
https://github.com/tensorflow/tensorflow/issues/35950
https://github.com/tensorflow/tensorflow/issues?q=+Failed+to+call+ThenRnnBackward+
https://www.jianshu.com/p/d8cfd390e30b
https://stackoverflow.com/questions/53972814/cudnnlstm-failed-to-call-thenrnnforward
n卡驱动431.36
http://bbs.pcbeta.com/forum.php?mod=viewthread&tid=1822598&ordertype=2
查看n卡驱动版本
https://mp.weixin.qq.com/s?__biz=MzAxMzUzOTg3OQ==&mid=2455934821&idx=1&sn=616f6795a18d78b429fb9fd76876cc93
dch版本驱动
https://www.expreview.com/70334.html

UnkownError:Fail to find the dnn implementation. 解决方案及附加问题
https://blog.csdn.net/weixin_44545603/article/details/103322446
https://stackoverflow.com/questions/54473254/cudnnlstm-unknownerror-fail-to-find-the-dnn-implementation

VMware虚拟机上不能使用CUDA/CUDNN
https://blog.csdn.net/m0_37605642/article/details/98725096