Merge pull request #261 from hackingmaterials/tpot_downgrade

Tpot downgrade
hackingmaterials · Nov 8, 2019 · 0a5d0dd · 0a5d0dd
2 parents 39c267f + 26bd6c7
commit 0a5d0dd
Show file tree

Hide file tree

Showing 4 changed files with 42 additions and 37 deletions.
diff --git a/automatminer/automl/adaptors.py b/automatminer/automl/adaptors.py
@@ -69,8 +69,8 @@ def __init__(self, **tpot_kwargs):
         tpot_kwargs["cv"] = tpot_kwargs.get("cv", 5)
         tpot_kwargs["n_jobs"] = tpot_kwargs.get("n_jobs", -1)
         tpot_kwargs["verbosity"] = tpot_kwargs.get("verbosity", 3)
-        tpot_kwargs["memory"] = tpot_kwargs.get("memory", "auto")
-        tpot_kwargs["use_dask"] = True
+        # tpot_kwargs["memory"] = tpot_kwargs.get("memory", "auto")
+        # tpot_kwargs["use_dask"] = True
 
         self.mode = None
         self.tpot_kwargs = tpot_kwargs
@@ -109,10 +109,10 @@ def fit(self, df, target, **fit_kwargs):
         # Determine learning type based on whether classification or regression
         self.mode = regression_or_classification(df[target])
 
-        mltype_str = "Classifier" if self.mode == AMM_CLF_NAME else "Regressor"
-        self.tpot_kwargs["template"] = self.tpot_kwargs.get(
-            "template", "Selector-Transformer-{}".format(mltype_str)
-        )
+        # mltype_str = "Classifier" if self.mode == AMM_CLF_NAME else "Regressor"
+        # self.tpot_kwargs["template"] = self.tpot_kwargs.get(
+        #     "template", "Selector-Transformer-{}".format(mltype_str)
+        # )
 
         if self.mode == AMM_CLF_NAME:
             self.tpot_kwargs["config_dict"] = self.tpot_kwargs.get(

diff --git a/automatminer/automl/config/tpot_configs.py b/automatminer/automl/config/tpot_configs.py
@@ -80,14 +80,14 @@
         "bootstrap": [True, False],
     },
     "sklearn.linear_model.RidgeCV": {},
-    "xgboost.XGBRegressor": {
-        "n_estimators": tree_estimators,
-        "max_depth": tree_max_depths,
-        "learning_rate": tree_learning_rates,
-        "subsample": tree_ensemble_subsample,
-        "min_child_weight": range(1, 21, 4),
-        "nthread": [1],
-    },
+    # "xgboost.XGBRegressor": {
+    #     "n_estimators": tree_estimators,
+    #     "max_depth": tree_max_depths,
+    #     "learning_rate": tree_learning_rates,
+    #     "subsample": tree_ensemble_subsample,
+    #     "min_child_weight": range(1, 21, 4),
+    #     "nthread": [1],
+    # },
     # Preprocesssors
     "sklearn.preprocessing.Binarizer": {"threshold": np.arange(0.0, 1.01, 0.05)},
     "sklearn.decomposition.FastICA": {"tol": np.arange(0.0, 1.01, 0.05)},
@@ -214,14 +214,14 @@
         "C": [1e-4, 1e-3, 1e-2, 1e-1, 0.5, 1.0, 5.0, 10.0, 15.0, 20.0, 25.0],
         "dual": [True, False],
     },
-    "xgboost.XGBClassifier": {
-        "n_estimators": tree_estimators,
-        "max_depth": tree_max_depths,
-        "learning_rate": tree_learning_rates,
-        "subsample": tree_ensemble_subsample,
-        "min_child_weight": range(1, 21),
-        "nthread": [1],
-    },
+    # "xgboost.XGBClassifier": {
+    #     "n_estimators": tree_estimators,
+    #     "max_depth": tree_max_depths,
+    #     "learning_rate": tree_learning_rates,
+    #     "subsample": tree_ensemble_subsample,
+    #     "min_child_weight": range(1, 21),
+    #     "nthread": [1],
+    # },
     # Preprocesssors
     "sklearn.preprocessing.Binarizer": {"threshold": np.arange(0.0, 1.01, 0.05)},
     "sklearn.decomposition.FastICA": {"tol": np.arange(0.0, 1.01, 0.05)},

diff --git a/automatminer_dev/submit.py b/automatminer_dev/submit.py
@@ -15,7 +15,7 @@
 
 if __name__ == "__main__":
 
-    N_JOBS = 10
+    N_JOBS = 2
 
     pipe_config = {
         "learner_name": "TPOTAdaptor",
@@ -25,7 +25,8 @@
             "max_time_mins": 1440,
             "max_eval_time_mins": 20,
             "population_size": 200,
-            "memory": "auto",
+            #todo: change this back
+            # "memory": "auto",
             "n_jobs": N_JOBS,
         },
         # "reducer_kwargs": {"reducers": ("corr",)},
@@ -49,7 +50,7 @@
     }
 
     pipe_config_debug = {
-        "autofeaturizer_kwargs": {"preset": "debug", "n_jobs": 10},
+        "autofeaturizer_kwargs": {"preset": "debug", "n_jobs": N_JOBS},
         "reducer_kwargs": {"reducers": ()},
         "learner_name": "rf",
         "learner_kwargs": {"n_estimators": 500},
@@ -77,18 +78,22 @@
         "debug"
     ]
 
-    # from automatminer_dev.config import MP_E_FORM, JDFT2D, GLASS
-    # wf = wf_benchmark("lrc", pipe_config, **GFA, cache=True, tags=tags)
+    from automatminer_dev.config import EXPT_IS_METAL, EXPT_GAP
+    worker = "lrc"
+    # wf = wf_benchmark(worker, pipe_config, **EXPT_IS_METAL, cache=True, tags=tags)
+    # wf = wf_benchmark(worker, pipe_config, **EXPT_GAP, cache=True, tags=tags)
+    # wf = wf_benchmark(worker, pipe_config_debug, **EXPT_IS_METAL, cache=True, tags=tags)
+    wf = wf_benchmark(worker, pipe_config_debug, **EXPT_GAP, cache=True, tags=tags)
 
-    wf = wf_evaluate_build(
-        "cori",
-        "24 hr tpot express 99% reducing with all mean cleaning samples",
-        BENCHMARK_FULL_SET,
-        pipe_config,
-        include_tests=False,
-        cache=True,
-        tags=tags,
-    )
+    # wf = wf_evaluate_build(
+    #     "cori",
+    #     "24 hr tpot express 99% reducing with all mean cleaning samples",
+    #     BENCHMARK_FULL_SET,
+    #     pipe_config,
+    #     include_tests=False,
+    #     cache=True,
+    #     tags=tags,
+    # )
 
     # wf = wf_run_test("local", "initial_test")
 

diff --git a/requirements.txt b/requirements.txt
@@ -1,7 +1,7 @@
 # Direct requirements of this project alone
 matminer==0.6.2
 xgboost==0.80
-tpot==0.11.0
+tpot==0.9
 skrebate==0.6
 pyyaml==5.1.2