wrong lambdas parameter h2o4gpu.ElasticNet

[sologubix]

• OS platform, distribution and version (e.g. Linux Ubuntu 16.04): Linux Ubuntu 18.04
• Installed from (source or binary): binary
• Version: 0.2.0
• Python version (optional): 3.6
• CUDA/cuDNN version: 9.0
• GPU model (optional): 1080TI GTX
• CPU model: Intel(R) Core(TM) i7-7820X CPU @ 3.60GHz
• RAM available: 64GB

Description I was trying to set lambdas parameter for h2o4gpu.ElasticNet and model ignores inputted values .

``model = h2o4gpu.ElasticNetH2O(family='elasticnet', alphas = [0.2, 0.3, 0.8], lambdas=[0.5, 1, 10]) model.fit(train_x, train_y)

model.get_params()

{'alpha_max': 1.0, 'alpha_min': 0.0, 'alphas': array([[0.2], [0.3], [0.8]]), 'double_precision': 1, 'family': 'elasticnet', 'fit_intercept': 1, 'glm_stop_early': 1, 'glm_stop_early_error_fraction': 1.0, 'gpu_id': 0, 'lambda_max': -1.0, 'lambda_min_ratio': 1e-07, 'lambda_stop_early': 1, 'lambdas': array([[2.], [0.], [0.]]), 'max_iter': 5000, 'n_alphas': 3, 'n_folds': 5, 'n_gpus': 1, 'n_lambdas': 3, 'n_threads': 1, 'order': None, 'store_full_path': 0, 'tol': 0.01, 'tol_seek_factor': 0.1, 'verbose': 0}

[mdymczyk]

@sologubix hey! Looks like it's just a bug in our print method, the actual property passed to our C backend is model.lambdas_list, this one seems to be set properly. I'll mark this as low priority and we'll try to fix the printout in the future.

[sologubix]

@mdymczyk Thanks for clarification.