Closed yananchen1989 closed 5 years ago
Shouldn't you use the Lock in predict_memcache as well? Not really a Keras issue imo.
@yananchen1989 - Could you answer the above question asked by @Dref360
Closing this due to lack of activity. Feel free to reopen when new information is available. Thanks!
My code has a class which includes two thread: one is responsible for load model from disk and the other one is use the loaded model to predict. here is a skeleton of the class: some details have been erased for simplicity
But when running this code, most of the time it is OK, while sometimes the two thread both gives the error below, just as the program execute the
mode.predict
andload_model
operations.So what's wrong with this and I am confused whether the lock has worked right. Any solutions ? Thanks.