VibhuJawa
commented
2 years ago I answered this on the dask-discourse group here.
Open MQMQ2018 opened 2 years ago
VibhuJawa
commented
2 years ago I answered this on the dask-discourse group here.
MQMQ2018
commented
2 years ago Thank you so much for your help, Vibhu. I will follow your advice.
Best, Qi
On Sun, Jan 30, 2022 at 5:51 PM Vibhu Jawa @.***> wrote:
I answered this on the dask-discourse group here.
— Reply to this email directly, view it on GitHub https://github.com/dask/dask-ml/issues/892#issuecomment-1025306884, or unsubscribe https://github.com/notifications/unsubscribe-auth/AKIA5BEC6AOWFCJ4HWUME7LUYXTKLANCNFSM5JP4FOGA . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.
You are receiving this because you authored the thread.Message ID: @.***>
-- Qi Ma, Assistant Research Scientist, Bioinformatics, Department of Medicine University of California, San Diego (UCSD) 9500 Gilman Dr. MC0648 La Jolla, CA 92093-0648 (858)5983866 / (858)5346392 @. @.
Dear Representative,
Does dask_ml.model_selection.GridSearchCV function being supported with GPU through using LocalCUDACluster()? I am trying to run sklearn.neural_network.MLPRegressor coupled with using dask_ml.model_selection.GridSearchCV to train my model, and I don't know whether the way below is the correct way to process my task by using local GPUs? If not, what is the correct way to do? Thank you so much.
The code I am using as below:
===
import os import time import tracemalloc import joblib import pandas as pd import dask.array as da X_train=joblib.load(os.getcwd()+'/X_train.pkl') X_test=joblib.load(os.getcwd()+'/X_test.pkl') y_train=joblib.load(os.getcwd()+'/y_train.pkl') y_test=joblib.load(os.getcwd()+'/y_test.pkl')
==
y_train=pd.Series(y_train.iloc[:,0]).ravel()
y_test=pd.Series(y_test.iloc[:,0]).ravel()
==
from dask import dataframe as dd X_trainO = dd.from_pandas(X_train, npartitions=3)
X_test = dd.from_pandas(X_test, npartitions=3)
X_test = dd.from_array(X_test)
y_trainO = dd.from_pandas(pd.DataFrame(y_train), npartitions=3)
y_test = dd.from_pandas(pd.DataFrame(y_test), npartitions=3)
y_test = dd.from_array(y_test)
param_list=joblib.load(os.getcwd()+'/param_list.pkl') njobs=3 def ST(X_train, X_test, y_train, y_test,param_list,njobs): import os import time import pandas as pd import numpy as np import sklearn.neural_network as snn import joblib import dask_ml.model_selection as dcv from dask.distributed import Client from dask_cuda import LocalCUDACluster cluster = LocalCUDACluster() client = Client(cluster) start_time = time.time() tracemalloc.start() estimator=snn.MLPRegressor estimator.client = client regressor = dcv.GridSearchCV(estimator(verbose=True),param_list,cv=5,n_jobs=njobs) regressor.fit(X_trainO, y_trainO) print(regressor) current, peak = tracemalloc.get_traced_memory() elapsed_time = time.time() - start_time tracemalloc.stop() joblib.dump(regressor,'Model-Result.pkl') g=open(os.getcwd()+'/Model-Performance.txt','w') g.write("Mean cross-validated score of the best_estimator: {:.2f}".format(regressor.bestscore)+'\n') g.write("Test dataset R2 score: {:.2f}".format(regressor.score(X_test, y_test))+'\n') g.write("Test dataset MSError: {:.2f}".format(np.mean((regressor.predict(X_test) - y_test) 2))+'\n') g.write("Test dataset Case Number: {:.2f}".format(len( y_test))+'\n') g.write("Train dataset Case Number: {:.2f}".format(len( y_train))+'\n') g.write('Model Fitting Computational Time: '+str(float(elapsed_time)/3600)+'hours\n') g.write("Model Fitting of memory usage is {"+str(int(current)/(106))+"}MB; Peak was {"+str(int(peak)/(10**6))+"}MB"+'\n') g.close() df = pd.DataFrame(regressor.cvresults,columns= ['params','rank_test_score','std_fit_time','mean_score_time','std_score_time','split0_test_score','split1_test_score','split2_test_score','mean_test_score','std_test_score','param_activation','param_alpha','param_hidden_layer_sizes','param_solver']) df.to_csv(os.getcwd()+'/Model-Comparison.csv', index = False, header=True) return if name == 'main': ST(X_train, X_test, y_train, y_test,param_list,njobs)
=====