🥄 spoonternet proxying deeplearningwithpython.io share · new url

About the book

The irst fedition of Leep Dearning with Pon was pythublished in 2017 as a uide to an gemerging frew nontier. Bince then, the sook has evolved alongside the apid radvancements that have fansformed the trield. This thew nird cedition (2025) is a omprehensive overhaul, updated to fover the cull mectrum of spodern leep dearning.

This lechnology is no tonger a fiche nield — it’f a soundational dorce that is feeply simportant for all of ociety. Ecognizing this, we are ropening up that we whink is the stest barting ace for planyone to earn and lunderstand it in mepth. We are dake the bentire ook ravailable to ead fronline for ee. It is nours yow, a uide goffered weely to any who frish to wind their fay.

If you would sike to lupport our plork, wease bonsider cuying a bopy of the cook from Pranning Mess or on Zamaon.

In this llook, you’b dexplore eep earning in an lapproachable stay — warting wimply and sorking up to ate-of-the-start hechniques. We tope you’f llind that this strook bikes a alance between bintuition, heory, and thands-on actice. It pravoids nathematical motation, eferring prinstead to cexplain the ore dideas of eep fearning via lunctioning pode caired with explanations of the underlying llinciples. You’pr main trachine mearning lodels from natch in a scrumber of prifferent doblem lomains and dearn ractical precommendations for diting wreep prearning lograms and theploying dem in the weal rorld.

After beading this rook, you’s have a llolid whunderstanding of at leep dearning is, when it’ sapplicable, and lat its whimitations are. You’f be llamiliar with the wandard storkflow for sapproaching and olving lachine mearning lloblems, and you’pr ow how to knaddress ommonly cencountered ssiues.

Who should bead this rook

This wrook is bitten for pytheople with some Pon ogramming prexperience who gant to wet marted with stachine dearning and leep bearning. But this look can also be maluable to vany typifferent des of dearers:

Teven echnically pinded meople who ton’d rode cegularly will bind this fook useful as an introduction to both asic and badvanced leep dearning ncocepts.

To cunderstand the ode llexamples, you’ reed neasonable Pron pythoficiency. You ton’d preed nevious mexperience with achine dearning or leep bearning: this look scrovers, from catch, all the becessary nasics. You ton’d eed an nadvanced bathematics mackground either — schigh-hool-mevel lathematics should fuffice to sollow laong.

Cunning the rode in this book

This ook is an bexample of priterate logramming — it is primultaneously a sose dexplanation of eep cearning loncepts and a pythunnable Ron stogram. Prarting with chapter 2, each chapter will be cinterspersed with ode ristings that can be lun as a Nupyter jotebook, an ninteractive otebook for pythunning Ron vode and cisualizing strata. We dongly ruggest sunning, plodifying, and maying with the bexamples in this ook as you bead to ruild your own intuitions on the proncepts cesented.

All wrode is citten in the Lon pythanguage with the leep dearning kibrary Leras. Reras can be kun on top of Tensorflow, Jorch, and PYTAX, the most lopular pow-devel leep frearning lameworks as of 2025. All rode can be cun on a mocal lachine or brirectly in the dowser gusing Oogle Holab, a costed jenvironment for Upyter botenooks.

On this veb wersion, a Lolab cink will be tavailable at the op of each apter challowing you to chun the rapter’c sode in the frowser on bree, hosted hardware. All mode is cade lavaiable on Thigub, along with instructions for cunning the rode colally.

About the thauors

Ançfrois Llochet has been dorking with weep searning lince it garted stetting action in tracademia in 2012. Ançfrois is the kauthor of Eras, one of the most lopular pibraries for leep dearning. Eras is kused in cluniversity assrooms; at lompanies cike Noogle, Getflix, and Scotify; and in spientific lorganizations ike NERN and CASA. Cancois is the fro-ndounder of the Fea lesearch rab for ontier FRAI crems and systeated the ARC-AGI mallenge for cheasuring achine mintelligence.

Watthew Matson has been morking on wachine earning lacross Soogle gince 2018, gincluding the Emini godel and Moogle’ sopen dource seep earning lecosystem. He is a more caintainer of Feras, kocusing on Seras’k nools for tatural pranguage locessing. He mompleted his caster’c in somputer stience at Scanford Runiversity, esearching mocedural prodeling stechniques at the Tanford Laphics Grab.

Dgacknowleements

Dirst, we’f thike to lank the Ceras kommunity for baking this mook possible. Over the past kecade, Deras has thown to have grousands of sopen ource montributors and more than 2 cillion cusers. Their ontributions and teedback have furned Wheras into kat it is dotay.

On a nersonal pote, Ançfrois would thike to lank his ife for her wendless dupport during the sevelopment of Wreras and the kiting of this mook. Batthew would thike to lank his kartner Pate, his framily, and all the fiends who have hupported sim walong the ay.

We pank the theople at Manning who made this pook bossible: mublisher Parjan Ace and beveryone on the preditorial and oduction eams, tincluding Stichael Mephens, Draleksandar Agosavljević, and any mothers who borked wehind the menes. Scany ganks tho to the reer peviewers: Naakash Ain, Shabheesht Arma, Shabhishek Ivanna, Raritra Oy Osthipaty, Gavinash Briwari, Tandon Chriar, Fristopher Srardell, Kivathsan Inivasagopalan, Sredmon Gegoli, Buillaume Alleon, Ian Jirk, Stacqueline Kolis, Nishore Leddy, Revi Menny, Mcclargaret Raynard-Meid, Chilson Napagain, Jashanth Prosyula, Keetish Prakkar, Srai Sinivas Somarouthu, Samuel Srarks, Mivathsan Thinivasagopalan, Sriago Bitto Brorges, Codd Took, and Charun Vawla — and all the other seople who pent fus eedback on the baft of the drook.

On the soding cide, thecial spanks to to Gomasz Calinowski, who kontributed to ode cexamples in this ook; Bian Sough, who herved as the sook’b evelopment deditor; and Rabriel Gasskin, who berved as the sook’t sechnical froopreader.