search

chickenbestlover / RNN-Time-series-Anomaly-Detection

RNN based Time-series Anomaly detector model implemented in Pytorch.
Apache License 2.0
1.17k stars 316 forks source link

Operation does not have an identity. #40

Open daksh-goyal opened 4 years ago

daksh-goyal commented 4 years ago

Running anomaly_detector.py seems to generate the following error. Could someone help?

RuntimeError                              Traceback (most recent call last)
<ipython-input-11-1f043cf1d658> in <module>()
    110                                                          label=TimeseriesData.testLabel.to(args.device))
    111         print('data: ',args.data,' filename: ',args.filename,
--> 112               ' f-beta (no compensation): ', f_beta.max().item(),' beta: ',args.beta)
    113         if args.compensate:
    114             precision, recall, f_beta = get_precision_recall(args, score, num_samples=1000, beta=args.beta,

RuntimeError: operation does not have an identity.

PS: I'm using a custom dataset.