search

kochkinaelena / branchLSTM

branchLSTM model from Turing at SemEval-2017 Task 8: Sequential Approach to Rumour Stance Classification with Branch-LSTM
MIT License
22 stars 3 forks source link

Set hyperopt seed #6

Open LouiseABowler opened 6 years ago

LouiseABowler commented 6 years ago

Hyperopt has the option to initialise its internal RNG via the rstate argument, which isn't set at present (rng_seed in the list of hyperparameter options is used by Lasagne). If we explicitly set rstate, can we reliably obtain the same set of hyperparameters from the optimisation process?

LouiseABowler commented 6 years ago

793612c2dc76a1da2a3869d26a6a1fc800399c35 adds the ability to specify the seed for hyperopt using the --hseed option when running outer.py.