Source: scikit-optimize Version: 0.10.2-4 Severity: important User: [email protected] Usertags: scikit-learn-1.7
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA512 Dear maintainer, your package has an autopkgtest regression with scikit-learn 1.7.2. Relevant excerpt from https://ci.debian.net/packages/s/scikit-optimize/unstable/amd64/65190549/ follows: 483s =================================== FAILURES =================================== 483s _______________ test_minimizer_api[minimizer7-call_single-True] ________________ 483s 483s verbose = True, call = <function call_single at 0x7f8fed034220> 483s minimizer = functools.partial(<function gbrt_minimize at 0x7f8fecff1580>, acq_func='LCB') 483s 483s @pytest.mark.slow_test 483s @pytest.mark.parametrize("verbose", [True, False]) 483s @pytest.mark.parametrize("call", [call_single, [call_single, check_result_callable]]) 483s @pytest.mark.parametrize("minimizer", MINIMIZERS) 483s def test_minimizer_api(verbose, call, minimizer): 483s n_calls = 7 483s n_initial_points = 3 483s n_models = n_calls - n_initial_points + 1 483s 483s > result = minimizer( 483s branin, 483s [(-5.0, 10.0), (0.0, 15.0)], 483s n_initial_points=n_initial_points, 483s n_calls=n_calls, 483s random_state=1, 483s verbose=verbose, 483s callback=call, 483s ) 483s 483s tests/test_common.py:112: 483s _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 483s /usr/lib/python3/dist-packages/skopt/optimizer/gbrt.py:197: in gbrt_minimize 483s return base_minimize( 483s /usr/lib/python3/dist-packages/skopt/optimizer/base.py:276: in base_minimize 483s optimizer = Optimizer( 483s _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 483s 483s self = <skopt.optimizer.optimizer.Optimizer object at 0x7f8feb1c2d50> 483s dimensions = [(-5.0, 10.0), (0.0, 15.0)] 483s base_estimator = GradientBoostingQuantileRegressor(base_estimator=GradientBoostingRegressor(loss='quantile', 483s ... quantiles=[0.16, 0.5, 0.84], 483s random_state=RandomState(MT19937) at 0x7F8FF0D98440) 483s n_random_starts = None, n_initial_points = 3, initial_point_generator = 'random' 483s n_jobs = 1, acq_func = 'LCB', acq_optimizer = 'sampling', random_state = 1 483s model_queue_size = None, space_constraint = None 483s acq_func_kwargs = {'kappa': 1.96, 'xi': 0.01} 483s acq_optimizer_kwargs = {'n_jobs': 1, 'n_points': 10000, 'n_restarts_optimizer': 5} 483s avoid_duplicates = True 483s 483s def __init__( 483s self, 483s dimensions, 483s base_estimator="gp", 483s n_random_starts=None, 483s n_initial_points=10, 483s initial_point_generator="random", 483s n_jobs=1, 483s acq_func="gp_hedge", 483s acq_optimizer="auto", 483s random_state=None, 483s model_queue_size=None, 483s space_constraint=None, 483s acq_func_kwargs=None, 483s acq_optimizer_kwargs=None, 483s avoid_duplicates=True, 483s ): 483s args = locals().copy() 483s del args['self'] 483s self.specs = {"args": args, "function": "Optimizer"} 483s self.rng = check_random_state(random_state) 483s 483s # Configure acquisition function 483s 483s # Store and creat acquisition function set 483s self.acq_func = acq_func 483s self.acq_func_kwargs = acq_func_kwargs 483s self.avoid_duplicates = avoid_duplicates 483s 483s allowed_acq_funcs = [ 483s "gp_hedge", 483s "EI", 483s "LCB", 483s "MES", 483s "PVRS", 483s "PI", 483s "EIps", 483s "PIps", 483s ] 483s if self.acq_func not in allowed_acq_funcs: 483s raise ValueError( 483s "expected acq_func to be in %s, got %s" 483s % (",".join(allowed_acq_funcs), self.acq_func) 483s ) 483s 483s # treat hedging method separately 483s if self.acq_func == "gp_hedge": 483s self.cand_acq_funcs_ = ["EI", "LCB", "PI"] 483s self.gains_ = np.zeros(3) 483s else: 483s self.cand_acq_funcs_ = [self.acq_func] 483s 483s if acq_func_kwargs is None: 483s acq_func_kwargs = dict() 483s self.eta = acq_func_kwargs.get("eta", 1.0) 483s 483s # Configure counters of points 483s 483s # Check `n_random_starts` deprecation first 483s if n_random_starts is not None: 483s warnings.warn( 483s ("n_random_starts will be removed in favour of " "n_initial_points."), 483s DeprecationWarning, 483s ) 483s n_initial_points = n_random_starts 483s 483s if n_initial_points < 0: 483s raise ValueError( 483s "Expected `n_initial_points` >= 0, got %d" % n_initial_points 483s ) 483s self._n_initial_points = n_initial_points 483s self.n_initial_points_ = n_initial_points 483s 483s # Configure estimator 483s 483s # build base_estimator if doesn't exist 483s if isinstance(base_estimator, str): 483s base_estimator = cook_estimator( 483s base_estimator, 483s space=dimensions, 483s random_state=self.rng.randint(0, np.iinfo(np.int32).max), 483s n_jobs=n_jobs, 483s ) 483s 483s # check if regressor 483s # check if regressor 483s if not is_regressor(base_estimator) and base_estimator is not None: 483s > raise ValueError("%s has to be a regressor." % base_estimator) 483s E ValueError: GradientBoostingQuantileRegressor(base_estimator=GradientBoostingRegressor(loss='quantile', 483s E n_estimators=30), 483s E quantiles=[0.16, 0.5, 0.84], 483s E random_state=RandomState(MT19937) at 0x7F8FF0D98440) has to be a regressor. 483s 483s /usr/lib/python3/dist-packages/skopt/optimizer/optimizer.py:257: ValueError Cheers Timo -----BEGIN PGP SIGNATURE----- iQIzBAEBCgAdFiEEmwPruYMA35fCsSO/zIxr3RQD9MoFAmjs7J4ACgkQzIxr3RQD 9Moohg/9EnKczWoWNHeaW3C+HhoYmp7BZold/yi7CswmMxcwpxGKeq7Kc90VPht+ 1QS9nbRPZqkTUawQU7I/uRMymhVouTaUcYcRl19N4Qe5hjdow4RmgJOWLjjFYbKt Rn7g4/MhaLEHjHFeVW84VIJUZ3s/WBS2XJBmxfBnbUj0A04lAnDowvRRFWTlVWS3 PduLUkTDRDhdjo7XSvatYYZWbRk1Oa6TRkTT7tEextgN2ROdBqtf6INrflk0GzKD 5cLZSnn/MozGvFOdpF3MVM65m0fnEai8to8VGlTg/Tbe2gxxpZAEkVOHOdMM0GGW dShoXTeMY/kOVTHeYBIT7/rp3D1s3xzeInDkPLaE7hAZc3rZ5329hV1yteevSPBn iNiGdBaG44XjecYW6FzYQEcNeEvdRy6DezCY14J7DEvfZuVi92kARqNRWS2vOr0b R0fGJFiGJjf031717ZwEATCdZuBsOmTeXkvy/gY+jAhde0Qelyw3RkiQq0Ptj/eT nwxO4ue5jjksXrWtXOKe/NxvHz0jn02x+hgizak86SrAFA+8ExolGxP+UUjmz3t1 /pH5Jxe1gzS4srrfrtqrxiRRRcGgKE1yjwSGZ1tzAWnO540qH24lRPpuznZYO7JQ BP8R/UcKHqwcCDQdBrfqxZXcqLu0fAttcm2rHT5c8noQGK5rTqM= =FNWx -----END PGP SIGNATURE-----
