From 6449a8682a419f4814a6fad59ee28e01aa72c91b Mon Sep 17 00:00:00 2001
From: Awni Hannun <awni@apple.com>
Date: Tue, 5 Dec 2023 12:08:05 -0800
Subject: [PATCH] Doc theme (#5)

* change docs theme + links + logo

* move mlx intro to landing page
---
 docs/README.md                |   2 +-
 docs/src/_static/mlx_logo.png | Bin 0 -> 7410 bytes
 docs/src/conf.py              |  12 +++++++++++-
 docs/src/index.rst            |  24 ++++++++++++++++++++++++
 docs/src/quick_start.rst      |  22 ----------------------
 5 files changed, 36 insertions(+), 24 deletions(-)
 create mode 100644 docs/src/_static/mlx_logo.png

diff --git a/docs/README.md b/docs/README.md
index d0c35a31c6..f197ecf432 100644
--- a/docs/README.md
+++ b/docs/README.md
@@ -7,7 +7,7 @@ for example with `conda`:
 
 ```
 conda install sphinx
-pip install sphinx-rtd-theme
+pip install sphinx-book-theme
 ```
 
 ### Build
diff --git a/docs/src/_static/mlx_logo.png b/docs/src/_static/mlx_logo.png
new file mode 100644
index 0000000000000000000000000000000000000000..49400bd8d51fa2077363be082ae6c66c8665ac2f
GIT binary patch
literal 7410
zcmeHMc|25Y`#+ZKX>1|e7`rfIM2sQJU}TV;%06Rxtc|fRk--Qtc4<fyhE%p9Tb2@q
zETIyK8dUbmQvJsBJn#EFzvuJ*{`>v;p81?}U+227@3q|bb*^*nds5F>m>lLf#sL7p
zVN*29n%<5coNO%g>#3p=1A4>kt#7Um0QWLE_uUWC@4=pEYjXgIk_CWxA^_~tS@ARg
z2!#Q_vKs&(3IRYc_*$zKlHNhYIhuNzn*)k;nhiL_zzZ<ZDF*rvU^oUGXroiWX$Fyh
z)7A{~f66cdK#DKG{HM%WdV6r7>E{9FkCy2=!(SS&GyYq8^g7dD^a0EpQ$(gW?7?WK
z5CGuhJ2)ADtJj18fN{Xr){)?7jzM?^1*o`VgFJ955dpymC;%COppyYOf;%`Oz&|hq
z5upS5T>?R;57<x$_;(S4pAN*){0!JI2#*76sK8WU5M2&17>vYYy%5$Y<3Hi_I~|A*
zfe?&<Lc_zuRl?O&g7Dr@RV^(oC=3pT!<Fe0${~?~1osH#z!0fFl>AE%3K!yu_YEfa
z1_go-^tyWlg%WfikOM>iKK}5N;Oq65rNEFs!=eWWJ?Md|s=%QC35JXC{Xbv_J%7M{
zkLwR}$OB~vCu>|tkbmfbTe>IV$lopeFXmtV{Q>w(-Y(FWp!+Y#KiL07IsVoEmzlpa
z{|T_f`{L*W`MY1Lf4cQ|-rwX8qC=eVjllUkqI?5zfguO6z|~+#=zr|_C(<y;KL~FV
z?Cy!vr5pMk@;BB$rGK~M^ncp<ThE_JB=lgm|2F%7q}K0TdT!}*Aff-xQ(cbM9&9H7
z2s4_Z^lc*;K2uC+C;E=|4Q@Tt`uM$wuP~(d0Z6~^uJwzIIb$9vc@8f@S(Gh56AKro
zz*x7Em(loey4J^QrJa{HUj)6ygbzQXlnC$fZ*^8atod>q5o$SeZbE&t^z-UPYdoCM
z*iD|<OM=?SDZ(I?#RDVzvl)W{20jG<l?G#!%?{{dV+OlH4#oTM7<ZxS4+|v#%$N!Q
zq}<I`h;C+(0>#It@&h-Fgc$gqYPc~of?#Auy+$@EH^?+gx*otNYYmW&x-(;>U&r&X
z>2;wvfESST_*AS2jFEw^fdPD$cwm(2zefN6hPUflvfBpF_r2B5v6HhgIdQv7l_&N-
zHB?S}6)P)ue4qENYxzKR-Z}MaU+u!UY5%X4JDv40TN&CvzP=7>;TY#sijvVO^}M3R
zYm)O6^&9VJhEV<Zb3yIuVi$UOd~Y|sJneU<J*;&h+iY}W5L$Y;<JpudlIv4}Yn-V>
z?AD9geLI%ap+kb_VaoU4;9K_N?Ic>KzRg}N)!kp~LB{8@AJ3LJ56*t&Vz07Xl**s9
zz_#cH5;cRi6Rt*oX&LM0gkEE}a*)A&Eb=bdD*d%D7u`(H<^}?`;0?5MAFh12i_xj{
zi+beG_nD_gD@)_9uAw2uMR9F3k2K|cRfxpGZ;C}|b4eYdI2Yrc&6)7on*#QZk|{o+
z8N?V6yW#n;X<UPqM`36`YT8nFNBX(;YFiPpzq%%z^kHa2+BIg~Xm{shsrS+snH+?E
ztC67d3FUjeyzTWFTR*=u&5<=$XC+0MK6F*Fg%X_aheG(o*$q7xK4e{U@$Tnb(xs1c
zD>Zq@#JOrU^VHA6HPUz)wT@WE#bu`4WUG>vek&(+{KP!da*kY=8;dDq43egltqB{V
z67@a$l?Zc(Wy+~<t`8dlB!);Ms$%4V!<3neyPciNMoTJxsuxo|)<3bTyGs*>u-{)6
zqV{jnU9FtG_$zh+YcT8J8k?tMJux66Emz^`(mMwm!a_3meyA?V_9kLN+j#AIuRBG_
zjawf}9?7;sri~lv;1FO@X%8MFNoOmUu~PGjTRf#etdRWii<qHB9F;0#x!!Z+6fs(c
z%;wS1rAMK*@!&LtKte)~DVT~2-@1gh3^}gI6r6`c1=OC*wJNz!MF3nBHFtqzQF6t(
zkb8=}0*0+gN6*UN;4mD7k&}~yqYkZET;_2ukG3;&q`6J3(Nk~8$ta$R#|;}9y9gOt
z^WsB08<eEGJ?}{mb6`<oh)^zk0OatcSDjVq2FS=F7xSLGs%N>6GC5wB+lGI4!C(F+
zV<Rbs?H1Y2P7T~Jko{o9#hjCksW4qiRh?jZkHvbAau+@Lk_QXsqJ*1897!U!b9EM6
z8`_I#3F`cz(aE_SW8Y?c1mSeEI5*2;Ta8RanGFlcNqcml=3CP-3ket@-PQbLQyP!L
zi*p*Q5U4by!m~$ZGPfuf3o$=!nO-dp$#5B;&l||*DnUlvbiMeaDLVqUad|a;Apd<N
zOM)b_h7ii;dh6saJ=-&A`~x{wioG`dFKZNp)H#nl%K_VTmgw&|o&GpuOPM)F!?u)Y
zQI=oV#JztI1j%69W*@7DrGhMvyW`#_ET>^2_&;b$^GLhq?(0jGWKw4O$v<6;z+YoF
zpBJpWJU;wj)$$OT2IkgV4IBU5757MG%v8tt=Ka|y;{^xOld54<$yYm>yabRPrvRmX
z4HY5XW+i-O+)&baCcvE1di6mG8p7bua~_eN_|)gVq1Y>@I#)n@uwbb|d^D}ZBX(~>
zTVcnSDPR;?aT&7!O^yhC{YV3B(5=X8B}m|RSh`#|IWdXqc+|{$<cu2d$S6dX0eyd}
z@T1XP8pgQoW9dcRRf7*W(Q*ks3vL7Q;gfHI?a^?(iJz4gxem_sH2$$QY6e{#IYb#R
zeL+qnNB`KK9G(T^Dr_m*E2q%0kL{8*&Ro1bzy5rj`gQGETyM`>#w+?0(Ak*2Yg2No
z_O@G{;GKmUs##I3)hQ7zGB#$lV=oY~-pjX{tspg868)v<itaA$S9G<pw{`N=-cz|!
z4E>=Ajhahf-N5@nVfHSOi;Xj3uW$BnYbyoZeZ<n>v4rRn+`P+}w2u5fm$=lrQfEf5
zLinZd3-+$D+im>O3uF_oALE|cF1%m6Yd&?bFtc!U@bxx}v0R}ttA6L5qt?`2El+mY
zX*%t&m_ZY4({(krb(y$b@ae)^2`w@}B$DIy*Pl$ITf9!N#$;~*{9-ARrSy`$)bg=p
zu;xTt11&$_$*ai!fxpwEHLd2^xIMxv-^!cx+6vFCh2ZaR&?*}=JGWn+TWWqCX!b<n
zV$9G$)wANSfvMzzV5aum=;i84d~?GP;<y-LMBn#jp8ddN_N@71>DjTl?k^1eL~o$7
zo_Oy#w)`6QjB#nxKxK6Mf;%fpd<!am<-}9z8*x%{yr<Cw=CY|{ezoN9ft<-_nNx;W
z+rPOWe(i6LB?-C|Hq(c!@k{H<EeDrKnSvCdwiP{Yv7WRxLDv}7%uCogR{ig>2`LhB
znx7lTES9c?44KkzA~Yhn60WGo3Tt6ozP*heV_+4{GI<-&JT0s9b;`2=-T3~B_J_4g
zyJZFVeG;3bMCu6dJnTX;k*@SY>(^I)YGp>Fic(51{dje2TpI;f6OZUt)#mpbkW5b9
z9k><UZ87Vcow6jG+qNZ`nCHS9K6uji+t0~XV@GX08GXQ5q%QQxnoIQOCeht@0%5bU
z-~8XK+`jLE*jdQo^ClU5ly@H?rm1-kVTO^Ln6Jjj*AeeC)bum8zdXRy^I2$0ibRt!
zw7v&U&3&{?#Bvoz_X8@+r}M{Gv*MtR$g|#*&{4&j+H6)}Y4I7VXcvDzaxrf@g#O~a
z>YHP(VLGh4=bs<?Z6>^u9<0=)QHAnAVg4|eOP?}wxgl_y)h3ulBJSJ`&8)e0vSHzr
z8A}*ozwk)Js-*2r>DIUqnVG2fPDyVa^+YrT6GeYYLeR-6n(x!qM(Gjd+3!5biezW4
zqOjvtPreGDjU5Yo_&HytbAMxtR-WlK`tFg+gSkt{da>wF_abvEaz+OAh44!^3%sBP
z(|cbYLRah^`|Es?1SnCG&%KHR@;^pLueOJ`(&wXI1<6;DGB-|XnNfN8DLcdSEZ2>O
z+ir)B5^-6m$1_EilSuCsP_O&(Ba_7K2t_g~@*(9ZQ|bwy8yAmRzeh3hcjcwH<sw$w
zHb?m^kz~1iLbr4HOrD{30EYSeM!4((2;$MjDiIU>%BMU7j}E6tRw=CJg+0Qc3V!o*
zZ?$wA?hTot&qNUl#KTL$4=5~U8FD`@t28#>G62YWy_F$P6U)wsZ`@sr_kJHfR~XE<
zONsqD?b|eAKRAgCX&MSMf2A%dzA0|drr2}Kvg`1jUOm(=(5SvFY$Z?Lq_O+@)BTWF
zqrZ-99&QB6N3+*|e48y*X{Fsj{(AS;&cQj)Bm~VI*>Cod$01yN*>+=uM=8-z)PD6X
z-KRH^{n(+triHnt!c#o0Ivf2`J6bMj60XumHef8eV>`47r_zMf7*WE0S&I18x(D-@
zkrzF<`<5jGJkN|wGG%GquAaSfPEXXHwuX82hWc~4H-2wRJ;djO5stW`q^&TS!N)!V
zQ!=i(KO3hVE!72d4XQj;3b@yIbegny4``}o0%yfPIkV37EVJ<a+~>mkbDzJ^i%4yO
zYh{9C1-;IxQ|KR9sf&I*g3kTN$IP6T7P*QdW)n=bmnw4bLOsYV!shdn-n<{v6n+|N
zFG(S}V#jJM$5AYz_HItRa#k+l&XI)tN$+>YY5Y+d${!!6O6v*72DU8b`!7Uf*jVa>
z9!&-Euh8UP+!M|1Yq%Eqp(wKMMxGp!NlGe>f$ntl(RO;?V^@vDdF~JBAy!avXx1Va
z9NjWue{ONaYs~!QbMe)KdV-5-Pp`K5QsHb96x=<v^UNwqAd=b!gLjoAq=!3<T|Fkp
zJJ?G!Q1Y9N{74Z#*?o_mMCun=Vj>IU^l#e8TXsvm6BCc3mA%>N{WN(##JqLsR<T!V
zQrozZ^wn**_#@-OGqv{XA~UoeGqU)_{q=p1`;MIqwipz-)%E`M`o@BChk0S9GpZF#
zJ@xQ*Y{p*PdlV#4w$^jHh{9PLb?;0(Yxi|l8FGB?llRWa{iY`EJ0FU*79PG!7Qx%2
zd#$yqT~`Y@x4$RGy-gsqDLc`+IN|T?d8b2l4tu8;M~&&yzhu_k)5Rl2`)$xC)7vi0
zPic}*;QesPIFII+AVw}T2!n#*Vv@@1tf|TrZr&Q73Z^4SSZUr#SFRlsW4-T*4jyu>
z(4z<M2@F-TFSGr@D-*V~^^n7Ht+PRggf*K;FM_5`2WK_O!Hem?MaXvFp@hHH@`Phj
zwipG5O{$D;G)&t)GP8BrI7q|x%ZFnn+Lh88rjJ6%j_TORf@aH*8>O0T>y3J9=x8C$
z_1{3e7~>be636{U;2B@3#;~J;X!?pE)J146q$0tWh+V*@7Wd+nD3%_+SG~(=NG$JZ
z-cj<neSShyc7X(DNn%^_$07DE@}qmD%*N6KYMG)&r?zZLNo^Ut)eYVL=k1>Kf7jmR
zin@J<d=b&dX_Jk(20LO%dg)YK(`|eE?7IxC9zt-_a?JZ~xpP+hqw3kOXY%p2qd!E@
zsM+oNk3~mb8;z(d909$yVcuQi5Yl*kDQwC;rE%cY>nrH;PbeN!7dA+0ztc|cFL%%l
zorqn1Yb-OI^pgh9h>{jo&%yB4#qw9vd>zT2L*4TIkL0T?ls=@@vAz>yXCz-{LU*Y4
zT|5D(4lmxUS*v|5y}`G<u`RSaHydwWj$-stF0N;Tod%3}kxY_o<#NnkPUnZ0!}h^D
zU|8`v$<<DUR8&N`#*It3PSg#q2WzL}6Ovb~UK#Lx&&)u#^}Il}vG?3+Sc=iOYTR*B
zait>ea4qJgG?Fh>C>O=ZLe)p>X*FM7pFg`8T0(68dM@R0YOmrEmgn3{x46!-bj2iP
z)v(Wx@^HBCea<x<w$ywQR^z^x88>q4<*i+;?UaUvw6U-83ATxG(rt@*uIgKn3nF=v
z<7Cdb-B!)gByURn%*I?E+xiq5`B~D1UZc;1c4=}&i6R~fzc}%>hsNILSluIXI8A1e
zY{RvV6&Xt&b|gt>0ndG-KyG;%ck8%Q!R);?h8hq)1LRj(3%_H>$+rX=HC~E9QEZRr
z92V8tW}GM!7TIw4Iu}Ht*Kqg*_gz_Q5vLd4#khrMp42unx@TG5^`W=J$;H8bePR3H
zrV16|So@ILY4eJE-KVApa3g(L+}61=CGS(W64gHuQ)1|gw+`RK>zE37jn9r+8_A%R
zDqz21aMT65W~i``LfTa|FG($&aa%0Hf((B1vvVp6#6J!lMjaE};H!0bJlOJOv~l3>
z=d>Miov*BWNwaUBdNDjg#sC@1Q?DeglryoV##imzTZxu9PB?qBhN{8IwL5c&je%jo
z9n30O(q>zvL{5#3+1)Oy-_O~2hPs+tQkWG*V&+%^GkqK29d=ThOOZALFA7GY>ZvWf
zi&^^ff}~qAy=r89!A>R!bb)ONY%}qOgpZI)B4m$!#)yYC+6GWA719KX$TIlQ9BuY*
z^d$@}iM=`znb|Slb+m;3jb?b6>S`jO#lG%HsxsqPuo3Jw%U*V{ue_k6Mt+>MPh9Fl
zs|Gi(tZlP7f1}DquA_*A(0<7c62Z=ey<O2J{871sb1{8Yp`NfoTmkuNHWYBf-{}nS
zMoKchyBZf>rgU*#xYvdxSI=K6c<Oj>@s=HuuX@T)`G=CHEOCC(&kk)8hwincE_qK#
zmva-da(`)-&CJ^2z<k@?HWt6Z$+At}$La(_c;faY41(t_w91f4Go1V9PN&CY^(u<b
zX6LI6wPzF7HoeAk#%to*vFW=%-Q?3-?ek~l8DpN1BH7HwgaV*aeN#=*30d63qsN0N
zEIu8*=WCS8m}V9dF>QAR7zPK_cb*;WO0bU~QY{p`Cz+STRQFIuSjE`+n&5?tg(#o?
zI#x7fc+6ZpN>XJtZ_9GfwlZ+1xWgNwp7EVX%aMDLJdW5DNEd_NxMPAkk<=au;^Wny
zhkICiFIS5R(x8J9R-LVIOhqgDN7{yamJqlmgm^lp4vt0>Uo-?+ymPfHQtU^0X*3&H
z9EB0?I5<t@*&&xtDLg$$1E0I%koYjD?ekDrAUB$S$5pC|6|W_A?o_e8T<=(2ic6VK
zVz8H;to0bllBD}tNWNIGnkrP&!Tm^w$7(o5D-E-&A6vH`hfCc(6O?pw;r0d=w*e_B
zLi5O56mc1c7_`=tUCq1o)*b!PcIZlPviBD3*L*>64N1CvJS7Y?Z<Kv*V^8a(Ma6}a
z4@`fQj|fQez4y$kyNsrJY<)Zqdesgp>m1XQ$x)LOe<)7vXI`&5wW;`O43XKpfuoq>
z<feMgGr`5wxZ5wkOZn`RS9v&5I&bn@3^fdToyl3r-1AMjMHUHR0I^>uINvJRZ%hPb
zhGnSb1yym>3U|RS&9$nk*vDT*)1*rtgW1H`Y^nyU_z;B-ZbEIxg^lHKszP!1MT;7%
zB_f4Hch(Ks)TlNkx(0IRBKs|uO+JFZLrPogFFmmg(Fg6;D3Vjc5;8j|+RLqBFMSuv
zD94!fq!Fay-lo<`r|N#ct))Tt@uXpMni1>cE~jKo%)_-RA&t+H+hZ)R%(b4&*}B9&
zaw;VcCg>uWY6#|X%Y%>YS@f3mV=r&7)IzsCv*uhgHvC0>7f^~Cmv4*3#1>kW4^=rT
zX@ByLXIWq?ud@*(+-t*}6IM3*fWT>Qx}k|q=_73umCDDiBR0caZ1O7#3Lyg9E4b_9
zO%F~_Dum=tQYA`6;a0<tBHN$|iXZm?yM%7dOnI%J)mQ&t@}D}i?-lxyhDnv~=|l9j
z#j|W5%`ef9(}ktf{Fe@$|5@GwcsEG@GRuDvn48Lw_{$+p&^?c<XJNJtG)fH&-M0A_
z{x(vMF9*g_wD0%4b(Jvk1Z-arPm|*J0ttEFc)R5))`rsC%>&(8I)R6%&wy9d7~kt%
zkCGL%J#H+(En^J2BFigPYi$!%{`$2-P!_0()VW3r4WtG|ElvK2Tdsw_uZk`guVy<v
zmXY^GcKfuPm}r^91_b+*7IQ`ax)r3)cG^0<Ye8~^Mx9wupH+u5s?s-cRW;9^9;EN$
zo(%*7<GAQszFb%GNAdJ+;(U?#Vu?ih261Bf%|=lMzUoDmxerh1+sS=%J_^bvU8vt$
m7#{5WSH*wF;y>rbzTjR}UK0!C1O1Og05CPOK-C$zUHTuIjo7yU

literal 0
HcmV?d00001

diff --git a/docs/src/conf.py b/docs/src/conf.py
index 5d3be4e579..132a85fb2b 100644
--- a/docs/src/conf.py
+++ b/docs/src/conf.py
@@ -39,7 +39,17 @@
 
 # -- Options for HTML output -------------------------------------------------
 
-html_theme = "sphinx_rtd_theme"
+html_theme = "sphinx_book_theme"
+
+html_theme_options = {
+    "show_toc_level": 2,
+    "repository_url": "https://github.com/ml-explore/mlx",
+    "use_repository_button": True,
+    "navigation_with_keys": False,
+}
+
+html_logo = "_static/mlx_logo.png"
+
 
 # -- Options for HTMLHelp output ---------------------------------------------
 
diff --git a/docs/src/index.rst b/docs/src/index.rst
index 3c57db7de7..4459703701 100644
--- a/docs/src/index.rst
+++ b/docs/src/index.rst
@@ -1,6 +1,30 @@
 MLX
 ===
 
+MLX is a NumPy-like array framework designed for efficient and flexible
+machine learning on Apple silicon.
+
+The Python API closely follows NumPy with a few exceptions. MLX also has a
+fully featured C++ API which closely follows the Python API.
+
+The main differences between MLX and NumPy are:
+
+ - **Composable function transformations**: MLX has composable function
+   transformations for automatic differentiation, automatic vectorization,
+   and computation graph optimization.
+ - **Lazy computation**: Computations in MLX are lazy. Arrays are only
+   materialized when needed.
+ - **Multi-device**: Operations can run on any of the supported devices (CPU,
+   GPU, ...)
+
+The design of MLX is strongly inspired by frameworks like `PyTorch
+<https://pytorch.org/>`_, `Jax <https://github.com/google/jax>`_, and
+`ArrayFire <https://arrayfire.org/>`_. A noteable difference from these
+frameworks and MLX is the *unified memory model*. Arrays in MLX live in shared
+memory. Operations on MLX arrays can be performed on any of the supported
+device types without performing data copies. Currently supported device types
+are the CPU and GPU.
+
 .. toctree::
    :caption: Install
    :maxdepth: 1
diff --git a/docs/src/quick_start.rst b/docs/src/quick_start.rst
index 3439580ba6..ae79124873 100644
--- a/docs/src/quick_start.rst
+++ b/docs/src/quick_start.rst
@@ -1,28 +1,6 @@
 Quick Start Guide
 =================
 
-MLX is a NumPy-like array framework designed for efficient and flexible
-machine learning on Apple silicon. The Python API closely follows NumPy with
-a few exceptions. MLX also has a fully featured C++ API which closely follows
-the Python API.
-
-The main differences between MLX and NumPy are: 
-
- - **Composable function transformations**: MLX has composable function
-   transformations for automatic differentiation, automatic vectorization,
-   and computation graph optimization.
- - **Lazy computation**: Computations in MLX are lazy. Arrays are only
-   materialized when needed.
- - **Multi-device**: Operations can run on any of the supported devices (CPU,
-   GPU, ...) 
-
-The design of MLX is strongly inspired by frameworks like `PyTorch
-<https://pytorch.org/>`_, `Jax <https://github.com/google/jax>`_, and
-`ArrayFire <https://arrayfire.org/>`_. A noteable difference from these
-frameworks and MLX is the *unified memory model*. Arrays in MLX live in shared
-memory. Operations on MLX arrays can be performed on any of the supported
-device types without performing data copies. Currently supported device types
-are the CPU and GPU.
 
 Basics
 ------