Imininingwane Overview 
 
 
 
 
 
 
 Ngaba usebenzisa i-Keras Keras API izingxenye Ukuhlobisa Imodeli APIs nezinye izixhobo Izigaba ezilandelayo is the high-level API of the TensorFlow platform. It inikeza ukufinyelela, kakhulu-productive interface for solving machine learning (ML) imibuzo, ne-focus on modern deep learning. I-Keras ibandakanya yonke iminyango ye-machine learning workflow, ukususwa kwedatha ukuya ku-hyperparameter tuning kuya ku-deployment. It yakhelwe nge-focus yokuvumela ukuhlolwa okusheshayo. Ukuhlobisa Ngokusebenzisa i-Keras, unayo ukufinyelela ngokuphelele kwe-scalability ne-cross-platform capabilities ye-TensorFlow. Ungasebenza i-Keras ku-TPU Pod noma ama-GPU amancane, futhi ungasebenzisa ama-Keras amamodeli ekusebenziseni ku-browser noma ku-mobile devices. Ungasebenza kuma-Keras amamodeli nge-web API. I-Keras yenzelwe ukunciphisa isisindo se-cognitive ngokufinyelela izici ezilandelayo: 
 
 
 
 
 
 Ukunikeza interface elula, okuqhubekayo. Ukunciphisa inani lwezinto ezidingekayo ezivamile zokusebenza izimo. Ukunikela izindaba ze-error ezibonakalayo, ezisebenzayo. Ngokuhambisana ne-princip of progressive disclosure of complexity: It is easy to get started, futhi ungakwazi ukufinyelela izibuko zokusebenza ezidlulile ngokufunda ngokushesha. Thola uze ushiye ikhodi elula, elula. Ngaba usebenzisa i-Keras Ukuphendula okuphakeme ukuthi wonke abasebenzisi we-TensorFlow kufanele usebenzisa i-Keras APIs ngokuvamile. Noma unjiniyela, umfundisi, noma umfundi we-ML, kufanele uqala nge-Keras. Kukho izimo ezithile zokusetshenziswa (isib. Ukwakha izixhobo phezu TensorFlow noma ukuthuthukiswa kwekhompyutha yakho high-performance) okuyinto zihlanganisa isilinganiso esincane Kodwa uma isibonelo sakho sokusebenzisa akuyona omunye Ngiyaxolisa i-Keras I-TensorFlow Core I-API Izicelo ze-Core API Keras API izingxenye I-Keras isakhiwo se-data core Waze I-layer iyinhlangano elula ye-input / output, futhi i-model iyinhlanganisela ye-acyclic graph (DAG) ye-layers. Ukuhlobisa imikhiqizo Ukuhlobisa Waze Class kuyinto abstract ephambili ku-Keras. A Imininingwane ezivamile (izinga) kanye nekhompyutha (okufinyelelwa ku- Ukulungiswa tf.keras.layers.Layer Layer tf.keras.layers.Layer.call I-Layers iyatholakala ngokushesha: Uma usihlanganisa isibonelo se-layer njenge-attribut ye-layer eyinye, i-layer eyenza ukucubungula ama-weights eyenziwe yi-layer eyinhloko. Ngaphezu kwalokho, ungasebenzisa izindandatho zokuphathelela umsebenzi zokuphathelela idatha, njenge-normalization kanye ne-text vectorization. Izindandatho zokuphathelela zingahlukaniswa ngqo ku-model, noma ngesikhathi noma ngemuva kokufundisa, okwenza i-model ephakeme. Imodeli I-model iyona-object enikezela ama-layers ngokubambisana futhi ingasetshenziswa ku-data. Uhlobo lwe-model elula kakhulu Model, okuyinto ingxubevange linear of layers. Ukuze izakhiwo ezinzima, ungasebenzisa i-Keras functional API, okuvumela ukwakha i-graphs ye-layers, noma usebenzisa i-subclassing ukudala amamodeli ukusuka kwangaphakathi. Sequential Waze Class izici zihlanganisa ukwakhiwa nokuhlola izindlela: tf.keras.Model 
 
 
 
 tf.keras.Model.fit: Ukuqeqesha imodeli ngoba inani elifanele yebhizinisi. tf.keras.Model.predict: I-Generates I-output predictions ye-sample ye-input. tf.keras.Model.evaluate: Iphumela izinga zokushisa kanye nemikhiqizo ye-model; ifakwe ngokusebenzisa indlela ye-tf.keras.Model.compile. Lezi zindlela zibonisa izici zokusebenza ezilandelayo zokusebenza zokusebenza: 
 
 
 
 I-Callbacks. Ungasebenzisa i-callbacks eyenziwe ku-start early, i-model checkpointing, ne-TensorBoard monitoring. Ungasebenzisa futhi i-callbacks e-custom. Ukuqeqeshwa okuzenzakalelayo. Ungahambisa ngempumelelo ukuqeqeshwa kwakho ku-GPU amaningi, i-TPU, noma izixhobo. Ngokusebenzisa i-steps_per_execution argument ku-tf.keras.Model.compile, ungakwazi ukucubungula ama-batches amaningi ku-tf.function eyodwa, okwandisa kakhulu ukusetshenziswa kwe-device ku-TPUs. Ukuze ukubuyekeza okuhlobene indlela yokusebenzisa , bheka isitimela se-training kanye ne-evaluation. Ukuze ufunde indlela yokusebenza kwe-in-training kanye ne-evaluation loop, bheka Ukusebenza kwe-Customizing what happens in . fit fit() APIs nezinye izixhobo I-Keras inikeza ezinye i-API kanye nezixhobo zokufunda, kuhlanganise: 
 
 
 
 
 Izinzuzo Ukuhlobisa Ukukhangisa Ubuchwepheshe Data Loading Ukuze uhlu oluphelele le-API ezikhoyo, bheka i-API Ukuze uthole okwengeziwe mayelana neminye iiprojekthi ne-Keras, bheka . Imininingwane ye-API I-Ecosystem ye-Keras Izigaba ezilandelayo Ukuze uqala ukusebenzisa i-Keras nge-TensorFlow, bheka ngezinto ezilandelayo: 
 
 
 
 
 
 
 
 
 
 
 
 
 
 Umbala we-Sequential Model I-API ye-Functional Ukwakhiwa & Ukubuyekezwa nge-Methodology Ehlukile Ukwenza izindwangu ezintsha kanye nemodeli ngokusebenzisa subclassing Ukusebenza kwe-Serialization ne-Sparing Ukusebenza nge preprocessing layers Ukucubungula okufanayo ku-fit() Ukubhalisa Ukuqeqesha Ukubhalisa Ukubhalisa Ukubhalisa Ukusebenza nge-RNNs Ukubuyekezwa kwe-masking & padding Yenza i-callbacks yakho I-Transfer Learning & i-Fine-Tuning Multi-GPU kanye nokuhlanganiswa ukuqeqeshwa Ukuze uthole okwengeziwe mayelana ne-Keras, bheka izihloko ezilandelayo ku : Ngena ngemvume 
 
 
 
 
 
 mayelana Keras Ukuhlolwa kwe-Keras kubasebenzi Ukuhlolwa ku-Keras for Researchers Imininingwane ye-API I-Ecosystem ye-Keras 
 
 Okwakhiwa kwebhizinisi le-TensorFlow, le ncwadi ifumaneke lapha ngaphansi kwebhizinisi elisha futhi isetshenziswe ngaphansi kwe-CC BY 4.0. Isampula se-codes ifakwe ngaphansi kwe-Apache 2.0 License. Okwakhiwa kwebhizinisi le-TensorFlow, le ncwadi ifumaneke lapha ngaphansi kwebhizinisi elisha futhi isetshenziswe ngaphansi kwe-CC BY 4.0. Isampula se-codes ifakwe ngaphansi kwe-Apache 2.0 License. Ukuhlobisa Ukuhlobisa

Walkthroughs, tutorials, guides, and tips. This story will teach you how to do something new or how to do something better.

Lo msindo ukhiqizwa ngolimi lokuqala lwendaba!

I-Keras 101: Umhlahlandlela we-beginner-friendly ku-Building Neural Networks

About Author

IMIBONO

HANG TAGS

LESI SIHLOKO SETHULWE NGAPHAKATHI

Related Stories

The Billionaires F*cked Up

Zulu Network: Moving The Bitcoin Economy Forward With a Two-Tiered Bitcoin Layer 2 Architecture

The Billionaires F*cked Up

Zulu Network: Moving The Bitcoin Economy Forward With a Two-Tiered Bitcoin Layer 2 Architecture

Light-Mode

Classic

Newspaper

Dark-Mode

Neon Noir

Minty

HN StartUps