HyperparameterValue
| Gradient descent batch size | 64 |
| Entropy | 0.01 |
| Discount factor | 0.999 |
| Loss type | Huber |
| Learning rate | 0.001 |
| Number of experience episodes between each policy-updating iteration | 20 |
| Number of epochs | 3 |
Your rank
26
2020-04-10 21:53:18.446651+0900 DaelimMuseum[61407:9606241] <MobileTourViewController.m:(298)> [request.response statusCode] == >200
2020-04-10 21:53:21.457556+0900 DaelimMuseum[61407:9606241] *** Terminating app due to uncaught exception 'NSInvalidArgumentException', reason: 'MPMoviePlayerController is no longer available. Use AVPlayerViewController in AVKit.'
*** First throw call stack:
(0x1896d896c 0x1893f1028 0x1895c814c 0x195d112c8 0x10295d00c 0x18d1900e4 0x18d194d18 0x18d195104 0x10298b0bc 0x1029859e0 0x10296d1b0 0x10296cfa4 0x1893eccc8 0x18d7da918 0x18d1e9380 0x18d1e96f4 0x18d1e8704 0x18d3a5abc 0x189653524 0x18964e1c4 0x18964e774 0x18964df40 0x1938de534 0x18d7d9580 0x102979d74 0x1894cce18)
libc++abi.dylib: terminating with uncaught exception of type NSException
#import <AVKit/AVPlayerViewController.h>
#import <CoreMedia/CMTime.h>
[_player play];
[_player.player play];
[_player.player stop];
player stop
[_player.player seekToTime:CMTimeMake(1, 1000)];
[_player.player pause];
_player.contentURL
[(AVURLAsset *)_player.player URL]
return (_player.playbackState == MPMoviePlaybackStatePlaying);
return ([_player.player status] == .PlayerbackStatePlaying)
Training algorithm and hyperparameters Info
PPO
A state-of-the-art policy gradient algorithm which uses two neural networks during training – a policy network and a value network.
Hyperparameters
Stop conditionsInfo
Set the conditions for your training job to stop. To avoid run-away jobs, you can limit the length of a job to within a maximum time period (Maximum time).
The training will stop when the specified criteria is met. When your model has stopped training, you will be able to clone your model to start training again using new parameters.
Maximum time
60
Maximum time must be between 5 and 1440 minutes.
An EC2 instance will be provisioned to train your model with Amazon SageMaker and the AWS RoboMaker. It will take about 6 minutes before your training will start, after which you will incur a cost for the training. The estimated cost per training hour starts at $3.36. Charges depend on the track and required RoboMaker Simulation Units. With Free Tier, you will receive $30 of service credits so that your first model creation, evaluation, and submission to the virtual circuit is free. You can track your monthly costs in the AWS Billing Dashboard. To learn more about our Free Tier, please visit our pricing page .
CancelPreviousCreate model