Error in running python jobs (GPU based)

Question

I am using the hyperparameter optimization package spearmint.

In the code, I am using an keras (backend=tensorflow) based jobs. The code seems to work well when CPU is used.

However, when tensorflow uses GPU, the code terminates with the following error: Problem running the job:

(, AttributeError("'google.protobuf.pyext._message.RepeatedScalarConta' object has no attribute '_values'",), )
Traceback (most recent call last):
  File "/home/sujk/myLib/spearmint/spearmint/spearmint/runner.py", line 41, in job_runner
    elif job.language == PYTHON: run_python_job(job)
  File "/home/sujk/myLib/spearmint/spearmint/spearmint/runner.py", line 116, in run_python_job
    dbl_vals = param.dbl_val._values
AttributeError: 'google.protobuf.pyext._message.RepeatedScalarConta' object has no attribute '_values'
None

Following is the code for the job for reproducibility of the error:

import numpy as np

import keras
from keras.datasets import mnist
from keras.models import Sequential
from keras.layers import Dense, Dropout, Flatten
from keras.layers import Conv2D, MaxPooling2D
from keras import backend as K

def main(job_id, params):
    return np.random.uniform(0, 1)

if __name__ == "__main__":
    res = main(None, None)
    print(res)

Following is the code for config.pb file:

language: PYTHON
name: "test"

variable {
    name: "k_size"
    type: INT
    size: 1
    min:  2
    max:  4
}

The above was compiled using the following command:

spearmint config_mnist.pb --driver=local --method=GPEIOptChooser --method-args=noiseless=0 --max-concurrent=1 --grid-size=1

Error in running python jobs (GPU based)

0 Answers