I-Keras iwumtapo wezincwadi wePython onamandla wokwakha nokuklama amamodeli enethiwekhi ye-neural, ikakhulukazi amamodeli okufunda ajulile. Lo mtapo wezincwadi we-neural-network womthombo ovulekile ubhalwe nge-Python, oklanyelwe ukunika amandla ukuhlola okusheshayo ngamanethiwekhi ajulile e-neural; igxile ekubeni umsebenzisi-friendly, modular, futhi ande. Nokho, ngenkathi usebenza nama-Keras, ungase uhlangane nenkinga evamile - keras.datasets no module. Leli phutha libonisa ukuthi imojuli ye-keras.datasets ayitholakali noma ayifakiwe kahle ohlelweni lwakho. Le mojula ibaluleke kakhulu njengoba iqukethe imisebenzi eminingana yokufinyelela nokulanda amasethi edatha asetshenziswa ekufundeni okujulile nokufunda komshini.
import keras from keras.datasets import mnist (train_images, train_labels), (test_images, test_labels) = mnist.load_data()
Ikhuluma nenkinga 'ye-keras.datasets no module'
Ukubhekana ne-keras.datasets ayikho impande yephutha lemojuli inkinga engaxazululwa ngokufaka kahle noma ukufaka kabusha ama-keras kanye nemitapo yolwazi ye-tensorflow. Uma inkinga iqhubeka, ukulandela izinyathelo ezingezansi kungasiza.
Okokuqala, kuhlale kuwumkhuba omuhle ukubuyekeza umphathi wakho wephakheji ye-Python, i-pip, enguqulweni yakamuva. Bese, udinga ukukhipha ama-keras akhona kanye nokufakwa kwe-tensorflow usebenzisa umyalo wokukhipha i-pip.
Ngemva kokukhipha ngempumelelo, faka kabusha i-keras ne-tensorflow futhi.
Amazwibela wekhodi yePython abonisa lezi zinyathelo:
pip install --upgrade pip pip uninstall keras pip uninstall tensorflow pip install keras pip install tensorflow
Incazelo enemininingwane yekhodi
Ake ngikuhambise ezitebhisini ezingenhla.
Okokuqala, sibuyekeza i-pip. Ukugcina i-pip isesikhathini kukunikeza ukufinyelela kumaphakheji akamuva namapeshi okuphepha.
Okulandelayo, sikhipha ama-keras namalabhulali e-tensorflow ukuze sisuse noma yiziphi izinguqulo zangaphambilini noma ukufakwa okungaphelele okungase kubangele iphutha elithi 'keras.datasets no module'.
Ngemuva kwalokhu, sifaka kabusha i-keras ne-tensorflow. Lokhu kufakwa okusha kufanele kubhekane nephutha lemojuli.
Khumbula, i-Python nemitapo yolwazi yayo iyazwela ngendawo osebenza kuyo, ngakho-ke qiniseka ukuthi ufaka futhi ubhekisela kumaphakheji endaweni efanele uma usebenzisa into efana ne-virtualenv noma indawo ye-conda.
Imisebenzi ehilelekile kule nkinga
Umyalo we-pip iyithuluzi lokufaka nokuphatha amaphakheji wePython. Ngemiyalo ye-pip, sibuyekeza i-pip, siyakhipha futhi sifaka kabusha ama-keras kanye ne-tensorflow.
Imojula yedathasethi ye-Keras isetshenziselwa ukulayisha idatha ku-Python, engase isetshenziselwe ukuqeqesha amanethiwekhi e-neural. Idathasethi ye-mnist esetshenziswe esibonelweni sesethulo iyisizindalwazi samadijithi abhalwe ngesandla. Ilayishwa futhi ikhishwe kusetshenziswa indlela ye-load_data(), eyingxenye yemojula ye-keras.datasets.mnist.
