machine-learningoptimizationhyperopt

randint(low=4, high=12) returns a number lower than 4

I am using using Hyperopt's randint in one of my hyperparameter tuning experiments, where my model is getting hyperparamter values from fmin that are lower than than the lower limit specified for randint,

I am passing the following parameters to the hp.randint function (label, low=4, high=12) and the value I got was 0

my env is setup as follows:

hyperopt==0.2.7
numpy==1.19.4
transformers==2.11.0
torch==1.10.2

Any ideas/pointers on how this can be resolved?

Thanks in advance!

Solution

  • Correct signature for this function is hp.randint(label[, low], upper). Try it without kwargs, only with usual args.