search

fizyr / keras-retinanet

Keras implementation of RetinaNet object detection.
Apache License 2.0
4.38k stars 1.96k forks source link

Failed to get convolution algorithm #1452

Closed BS-98 closed 4 years ago

BS-98 commented 4 years ago

Hi,

when I start training I get error like this:

tensorflow.python.framework.errors_impl.UnknownError: **Failed to get convolution algorithm. This is probably because cuDNN failed to initialize**, so try looking to see if a warning log message was printed above. [[node retinanet/conv1/Conv2D (defined at keras_retinanet/bin/train.py:551) ]] [Op:__inference_train_function_13978]

but before I get: E tensorflow/stream_executor/cuda/cuda_dnn.cc:328] Could not create cudnn handle: CUDNN_STATUS_ALLOC_FAILED

So, is it a problem with my GPU? I have only 6GB. It could be memory growth issue? Or it is different problem?

I use tensorflow-gpu==2.3.0 and keras==2.4.3

irtiza00 commented 4 years ago

Have a look at this. https://medium.com/@JeansPantRushi/fix-for-tensorflow-v2-failed-to-get-convolution-algorithm-b367a088b56e

stale[bot] commented 4 years ago

This issue has been automatically marked as stale due to the lack of recent activity. It will be closed if no further activity occurs. Thank you for your contributions.