From 8c48af11c2da31e8db9e08cec53eb6b00cc9fcec Mon Sep 17 00:00:00 2001
From: Krrish Dholakia <krrishdholakia@gmail.com>
Date: Mon, 2 Oct 2023 14:44:11 -0700
Subject: [PATCH] fixes to get optional params

---
 litellm/__pycache__/main.cpython-311.pyc  | Bin 53167 -> 53083 bytes
 litellm/__pycache__/utils.cpython-311.pyc | Bin 139254 -> 139316 bytes
 litellm/llms/palm.py                      |  76 ++++++++++-----------
 litellm/llms/sagemaker.py                 |  79 +++++++++++-----------
 litellm/main.py                           |   1 -
 litellm/tests/test_completion.py          |   2 +-
 litellm/tests/test_streaming.py           |   2 +-
 litellm/utils.py                          |  14 ++--
 pyproject.toml                            |   2 +-
 9 files changed, 86 insertions(+), 90 deletions(-)

diff --git a/litellm/__pycache__/main.cpython-311.pyc b/litellm/__pycache__/main.cpython-311.pyc
index 6ebbe2ec9899d6b453a336e2d01d6e0387abca44..cdb85c0d8773c59eb261e48de2f69f133d9f77d1 100644
GIT binary patch
delta 11475
zcmds7YjhmNm9Cnn^|D5qmqwDU_tQeMu&k%C(94!>{KOCZ03o}CE!i3ktp<$Uoso${
z43>@TG7AeL$Ru%+gR<aE&ceb8=O}F0@PkAK?^&<l5a7UJcavSlIl(4jligd@(>>EY
z8euK@wS6?z)%Cbl_tsbU-nyk1o;APw9dp_b($bO<Jh&;RPrR6Rw+FvW{hlI|lukdK
z^3c<#Gc9sP<m0NxNtb#p^rI=ahggZ_f(fDL;io*ZQ)%WQwzw*nT4j%`aw}C9vVbJf
zOJ1i!JvpwOS0#}`OooB8)T-3DUOuIYmSj2USXN}}3*(yQs1(xUs&bX8eB#)T4C5}^
zk4VOISY8b8Oj_V`nw`fN&{}U64WC^|@APG6OPPXGw5t+ANJ0{p>{8~Wi@xA1@+C!U
z$ap`J9N)vI@$E=VrX!EJY(kBsX75Kr0WWnYIf#1&bmL#1g+7gfjKp->o4u&Wa?s2Q
z1&g#$M#Llg#0;$wn}Bc9OV4L}S36@$eUfuB%fMZcR3s!5pHu{c+eo(TCOI;E$xhK7
z?McSz|77Q-<&LLI`E2rXvFN2sa&p)zy2v7<G!-6vIa2QA;tMH&9{4GboF^4bq|#%}
zo}iQ_6-#-Oc^53u0Dj7YVdlhhki%^Z3l1?ytwD(7U%=27e##TTz;}t{4rGfQ+)C-d
zLalzGu6Kd1cM_7_l8bevIM<sNJ4_+%$#wf;Njfx=to^Z4mm{F3T|_VDI&mhQ%JT(d
zJ8IIGT#}m<kDDh;F2K4y&)1F-#Yn7Bi6b<i$$Y@0mXdJ>Cmn}Ts!j=*kV;7DxLqpd
zD^pre%A{h3B&nE`MS(-g6`=}lu|_2gNG+8R$&m^sE9mA1mjxCl3wX#1N~H+~RN1IB
zF58&fU(SFWr9Gexem@${fMs2x%wRYL@;Q;bl~hUSegsRs4i+*u(m%+l#H^$q2RG?f
z5&;gXWw&nVY;_ufXZCWjTy23+kJegp1~}u*YUO}#<<(MHC=D+LAk=&%5Zq+xxG5e6
zNnZ~LZpjJj=pwaIAY}Vj6(O-PI_Y90G|6R=HTK~a`seEGTynEcKpM3A@nj{<$ai;H
z<g#y+J%ArThs`_D1Ng6luDM2yLzdJ3ngLmhRi*%0Nn{yA(6E)%MUeC{$<Zxup4vKs
zFa=$#aDE6Q79H7ijcf!p>i1WHW&j~0YV?2x<r}+?G<ePr%w+jQKJE3oaXrmwE-@EO
zWYV@mU(KwPb>nFmklbQ9y_D|?l7?|hJY;p6t3cykuer(-kOs{~8po}w)|wMmJ~WHg
zYSL@1w!*Ac2Cb#ZZT2a|E@l)63derVah#HqN=dV()evck$*+j!<qD}nDooh&mS{_|
zk{NlalsRDz(kgbTLIh|Q(tGPYw7)nbqm?&8K=RgT^B`#>EqYrH4cd5vLb-G-!~Z&I
z=NrQX>b5<iZWkIhQjt0!7>tv|B3FV2T&@#WvDzX~p(_tYHRB?kneEVH*?{y;eXU$6
zu2Tm9o=ep{_n->J4eB=t37iGVD{fLtph~I)2y|TlKtHd6S}vRuNk$+Q!axP2eR35u
zf7>NRaQOxyX$BgE^e70ZUMGl3bw+_oal4uaUaHhQ_prB>ID&x`T}MrT@)B{k+IW6K
z02wT~VHp&l1U>)=fpn|8DaeK3@SiXUp<10Gs9GB!YE8tQIHKB2wSGWzFTlCd%p`bk
zYxPT?KpH_LSBbqUv3ZN5>W0OUKpM#s6UQ1=9CUrTJ@_YsENa#10U*xB?U<X@TG%nM
z{I4_cub0kkhT0gcfW3;E6!|7`zdFEtrK5viok~Tz37B1P@Y7?r(UPxrBbm%L=wx<>
zL1xR<;Q&8&z|TenKj8SuiOjWbx|&yiEOqvBKPH=^Djj6jqjgDSxOIdyYcW47D>;QU
z$r&9O#$_8@@iF{6bVJmuG7!A8m&cLe9sG4V9BnorMUSH`O0DL>CfH_Z_MyNzWQQ0}
zFt$1hzRBD576(o*n&t&CYL0_YqVo;rd8#2>$Af0oG4OZjnBS&nUar>FZa37{NYx`I
zvO}(xYT$i`Tyta7lBzvY4cR$vf|+O&iTKT)7?8#8OR02nm+@#fmgaxMunihi2!QH%
zL{S}ds>&5K=!)HXMhu#CkD)e3SHw<9UMdc&!+nLSxYJ-=!LTg-x|)<L(mPe9pcvOQ
ze+}&YyCSyZvpCzK*+j#cUelunccMo-Ze<1I!8yym<np-Dvlh9VpQ(DIZmPd2?nqUo
z9H|ifd0rMjQ|0WA1A<ic5C-q+uYHhO4F2d`$G52(XGb&0dv8SEFOj_v+i?cA>2_^Y
zRjf(uz?74H2GGsn?F0!iG){!jTJeE|2z5(yI)DJtS_#5sG5wkI%bqo8vhPAFxP|aj
zp830lWUq=`#Uo1b%OJ^nB9h!n?tzHgQI&lB?t-Hec&O^boL6Frbzfoj1(W;dFuGBj
zJEz}vk8zeoFv|nVEQd2mKR*E&z5bQzytvnYfMY`M^|z`m40>t6u&z<t!weodXsAxK
z>^F<Y)B)l}XB3=4Y&QeP@lOAtIxXmRDCT@Md(ew{D+ly~OZJa|;07bhdjd8|lr_eg
z{{}f*CVq3CobA-wOUq_uO$q0<Bnao#BtB_UT%469*XiPY5F;UjD(m#lI{VJw8!S}2
zIv;S<(hYST(Yi5{P?ZEnU}8DcDKGcCidZPKNv#Ko4bic5Pc<|M9O@ceCD#wGmKz4U
z<wk?gutx2|G&|kBEIU_j(rxOsx&{~PvTbs+w0vS6xG8R5lhiypZ1hRAz=sHdkK%H2
zAGyCD%Pj_8V)zt1QvJkj5S1>mukgF21|@G^A+4Ypa0X{#{+@a(z1!>N`L~*jAk=%c
z+1qL%4@kujDT@p0Mh>Ah2^Ou$grr*rHqf>fuUk7Vfl6S&JRDgaA1!Y1(2)iU&SGPE
zEmB3ihatAmU$-o3NEDQa3pK<?7^(n2h5&<jUKKf9KxhYUgeEpa0Lepp8uNUKTd-A0
z82ey|q<>toWZ8q@SH(sv?EW7amS>AP2Ni}iugOyg$Azn3jXD>g?#Q97gWKfx!R>Mf
z+$k`;^m<)ub;#17rz_R)HVd#KNYVgy6aZGXDAokUOdC+u&XH7V15kDFY(+eXnk?~-
z%MHM%b704;@R1_$@zRwoS^rPr(`ST_)&H~sKAKhq6E{j^ZsPQDn_Viv=8D?{$v19L
z@pp+YsRZY3(z+7r{n{J7$6=a7R`t5#u~_{WOj<Bv(k8@z_&?9DU~=R_61Y?FQy#fc
zd{xbZsy`+bKaHL-X_BIU=dmx1V7MoN3cOuM;UTGzp7*%2=Qd`UV9a92{iak0z1%t#
z$_!jk6+|izW?B5ZK@i93##VRVJR*$DMU(84-%1c*tWLmKp;$xTY^|^!S0_PJZcPb3
z99f1<;#5>{^NDwIhcRv$ZZz1*Lm}L}JgiP@1DB22e;#D>ZDn?#UHv*;s6|i`2VEq$
zyh_uDC$yUeuvj@3@`$u*HtVR|wY$ip#yfsePM>Yt&u*31lCt>9hzw0q_w9(Bl-Eez
zo#@Uyd9AcYTAMK3K1wfv9X;Bf_I0e}&8n3orzCU)RuH~#Wyg)LYb)tMN7WjmIy3pJ
zU{Ovh><&9-UlQ_IL>v?3w5s0dZ#um0ghA+hGNx@&4nkvWh_-clm(6d7tOGc8(UYA`
z%ja|J%(T^M6^JygD<ya)!mRPAazNcJpvU);GjQm7M=hDRnw>?BNp>YB4i}2>e;9N6
zEwg#jXJUA;pIv1cJR4b)f8&nvfPCBRYZoxXRU`%oub0*nON4bB<WENWr;Gm*Ws^|-
zvOR>s1yV2f&ao{=Cs&oTU6advUA;R@9v`=ozkwU)C#ZGxErU<`G`hO$tG}JST;3#Y
zN-P#~1O8>Hx^v*BMZLg+YZiZ|j>D(uZ&xqEjWnsdr08k=1#o=iq~e*-=I-Anhc@1R
z4ddU?Rom)A&uz0|e3<@d`>*lC^uZlXi%$TX0)b>EV@T-V-|z1|)Fbo_2ZsE^eZnJj
zddI${53?rsKvH2BR<nFJOSDqnTR6$e`gjH1cSn(V2Itd)o<{oG9h>k)n!d9=FWP~%
zt*~-hGq!**G#C(thiPEvcS=VX0e0jS7@lVq^!fJ<k$(SvWsFfMt1t<pv~Jg7+X!oQ
zjQ(U-Egq#G@7m_oIH@qlPLpohy*)*%6Hd{KyC1|S>5@I0R1(MbIDy3Ddm6J9nv5`$
z2%m%SEX$3CKHc*e&aQ`lM-iT21?&QH#wrd4dJg`3Ippbg@4AMM(x-Y>-8RDNH69}U
zq3lS;P(bV-Vge5I3cZ8F&G|<hgF~b*DsU)CW5Vwl76stCIiG$3|6wyLL0WP5qc|({
z)4OlN_!0V1@0$g}EilN8wO{NT92T~+NwACe8T%n9Ido`HI85K!OKf}<BznufF?=%g
zt9_SE_8Dsc1{3LX_xuIR^!)z2@UgQ8mMg+JNm~w-I{t$Z`yETdEcrdX_dvDn8p~d%
z-#^f4`!mb_g?@Hm3r?l&{U17W^AY-D1br@Il$!ZS<7neZXNU}J#PnAO3-QM9UOniv
zhHWU#6-o(aTGxN>L1~%OnGF*|kig%i)W*xHjnk=(SJJ&twoa#4!ru>rFCV^K)BVfh
zw`(>`*KGJa6=is)JxxDc3<>;Qax`CdG*3I4uh0#$&*N}UdzMWcgarOBIqEJu>ZToa
zp{>W$Y{6gDI@|U(pf{|m7qj<U#T{Puo?cPDxddG*#_Z!#N%EF*^QD@)Ek)+Hi>#1;
zyS$85yyNI9X74M@yK>n3{fu_r`-6)1H1_>bLwf;xf8y&dXYc<qwRzb4GhgmdJ^E}p
zZVR%n`s6^BIgH8zW#(|1mF25A`KzvuJl6aVb!|TOzP7NvkiEaCRmlITl~ez+z;d({
zeObci`ekXx(FXIE4V_J_<<zaU9gRHGS!rqOtU*(q)x6nMXH7<17cvK@x=_bT-lDs;
zYbnocSlV8~Gn+kaMLe^mxy{WpTOIA`JhMH&mcMsoc5%+8_O!Gu=OuU5c3OF6Z$on1
zKA(ANA9l9wbHUqXW#9V%9ol0Jy?^pijDH&HKlP>=kI-dj{t;W~7iZSs&qM8Jy%?M5
zF1iB0OP`?cLA7}9B!2BDk6RYtVrqS6Q<en-(z;o~{#|FrHasMVeMFd{gU{^6_R#b*
zm+hIX%}i3C|GvRKp;zn^(lGt|?@!>$(2fhgSP*;=`U&^4<N!-pG;XWF3m;<NEGQ)$
zXUVr90Zqd78dH9Xa{D$bH3?jwJjzO+XNkm;F_xTxM6Kt#a*~z)n3dX?R?0+5U#?ZS
zk{P8xej(e<2}TR~fTQ%i7uK(Qj}c(;S!EQ%d^m~sp>`&zYjAHL;mYy@*7`%1yi51L
z*nqFmXJ1^5uhZ9F%qsYV726@Ha41@mkIc#NK_8^oU#x;2#TRSw9ii<PCv5nw(Dj!v
z#wS95eA$a@7>6_Vp+KL%r?;OoU6Cxj%cyc9AG2J{bZP4=*$X~^LXhPN+WX29{3$*4
z%I{pP$4ts_pV)JMpRjibj18>QsaGrU7onG5-HrhP=|9_n|2wq*XHQv^K4tyg!b(hs
ze(t3Oum3Z)g|5H;p2_x~P$5Y4U*G7#?}nQG31<08sO6ViEx}z5oE%0hy^^vpZ04Uf
zl<o}Md3FJExWY+1n~YrEa0<_+qRi|kOU~tm)A;8?l<FGm8f^`y^G^p_;21L>m%|zS
zGZUq`$F`1c2s`<w2W9x4be~HPd)a3&oQ1r3Pu_QKDD30a*(fW2!g?+k&f%ZAD8n6I
z#IuVLOfZjUOOeePF5_7~b%lYec9Y5U2L#DgGcqkW?g=m8O^)Q-@zOBTl5FJ3x&nW`
NoGZT2skaJE{|o3*2@?PS

delta 11155
zcmdTqX>=4vdR24kzUI=5ghm$<0v%ieNr*$oh|3sppGFp{K}ev{&}zVtZV#7j4q*gE
z5@HKmyX<P;3zHaU_U&d(cnL|?dHeS5TG%9St!3kojh#)L*u%yfyn$z@s(QL-x@Tk%
z>pweBR9DycRn>RZ_kHy>KYvaC#_Rg@@1&=v0P=Ir&hx(bX8N;s_|Mp9FVXRO{GV?X
zm-2cFlzvDjWWEY6>qa2}SIIx=2Lb*xHy8h--2$^|&x(x5G^!&g<e&5tb0mN{4q%}G
z14=?BJj-Fhzs|K;QF0uXO`4v9QZ?(^B|uu-3<m|^+FT0`I}FmY8F5HXX?A8Dz$F2)
z;sALRU_jY~-1B5@LT&@f=>{6ArgQ^j`6d(#$kR%mZlOSMA10zWz2@YaT@cw$Bng=#
zdVJGv#aWj!u+5pXL?vgtQ1~i9j>~$jq=g*L!Y9jO1wa4-6wE?SFc&vDOPr}u2)f(@
zgsgADm*G0Vr{VTLwVJpZLGB&^ToD_!C1)XLBbg?>Wdf(Ngb4U-4Cmzr#A1QSZVxjA
z&LCun0J%gbZ&5HQ8O~syS_<2V_C-MOY%zFF!4k|@<B26g373MV2qlCl6DkmG$SpdB
z64Amt*i18P(Pezz<xVdg&J+qMS7Ise#5Y`dl5UH%nv&RY<_Y=1;!z@&tBPodHi|~l
z@iz)=tAUU%lnU-($*6%~kblw-5^j)d53}AO$Yt?UWC#GLRIL&}B(j()*!(WC<4^J&
zC9$|u)wrTfurj<g1-8r#)v9IqUY-r6;cG!_BSndB*qbBw{pu;@n#C=IEem1G8Hk;2
zm1hf9!G@*{>w^`eM3GmOQw;$>L&gE1*Dm*aCE=4-beUS2il*qQ7J)D&Ohwa&%|a>D
zOW8a$Lnx&NCX}KXQFB6-l0-QhKTF<(7;u>Y5Yb4m3U^l7^hG01!a{C?P&UDSG8ffa
zl?~Z^24>zx1=hd;Fpy1jSR?H?kV^XQNZ*KN3g7@BdR<Ahmap0-Q>M$W*7+>8sEN*J
zi#FBFJb5N(C*w8p8|YO4{B&gybSR^O#ZNd*r1pFnU8D9nYFflcsojR=4(sBTp8UJ1
z-6mLw)~u*Ds&<#JtpxB5GT8u>0<K;xCss*S>VymNxoTGc%~Q!uF{B<Cj8mylehea$
zSpIzZL3j{6r{4k&!Y8vW#aM7}Rhn#M0ikR%)C^gCuTTSvGT7^u`^4Uo(R^xG14dLA
zv9EtaZk5stW$cLEZIItV@1zV5nN-VW$;(7-DWGv7k8MjT?i3?sHZAY+NRvj65m-b*
zn+?|E&bFyU+7^7b*&YxqiC0&RD)u1QVwvnMl($C>!-jZkR=KcZ8CT`PI)E^3f~|-T
zLkotDvj3X2i8ZR+m&B{afhn>~OoED6#TZnhY6--jl`^zY@wIqsV|or+th@uULZ}dm
z@#;pBoGpnCqb3^p1rmI6?OCXqUn)zE1cDNLet{i7D9O%lW=KFNMlDhF0BS{x)PbBJ
zv@(QZv1};Y_d4>h17Ri3*AsER8JYuBDsM=_#TjzcrxP2lRf%Y-jNz|PjpOUV<bZ#Q
zDu}kJU8BZvE>%Ot>HI2rZNhH3+-Dh8G2bq~6Zw!7L2&SEV@9SEjV~MZki7XS%bh7m
zES{9%6qAjLkSDmD#69PlEZ;+oXq0e4M01*mMeR40knNw&Z;leyTP63YN^xGfMan4x
z;tc*788m5bXE1Jm7iN<Qw`gVLm^iz$d`QEs5;2JfH4%e(Dv~Q4mYiA)=`-a%@o}-a
z)+)7n{i#w|4TzO|mpo<CoL7E5oIfRV9?10llZNxzGI2!D7vs!-jtnK{AIt628gA9H
zSgnB)SxD>P?*%ggVin&bFFRR@$P3Z=y|wZxLKR`KU6W={bR67?^FV4LYh*c@GktTZ
zobasaESWPClUOU61WDY)i2|#NI>n71O-nrPgJ@l>i|tTms3L3tbSNQKwUFY|Cl?<D
z3ztq}p>n<f=*X4@>D#2*VLlxX(1vrw1L*h!%gj|7%X*DDsEuWV2C5hfUB5RQ;2S}+
zEAc+;#&LJL15T$IY#_K%9T3RXA~su6Vmmig6|X0G<v^Q;^>V28wZt$Q0BEzC7qL19
z4A2%0v__~tqC;E7YN3Waw}~~6N0vbF(e`0oEFmn*l_g=zD+@lOy=4uh`@YFcdS9h=
z2|4Sh;?8<K6*&EA&iW2D+0^ghzpio_qAqSGzh9b3+ewD9nO~2Kdp2@(^DOkT)fsW$
zpGxj~GWVOEYHpSLO>S?zc};YT9`|yCy4WQKMRPx2J-x%=&T0#c-dggu#cA{FyPCva
zdnbD%N%Pr>u)Psse;;{C-NfSGpgMI6WXtd`Pn;=?C{;RuALgb`L%Y-(_^J%Uin=w(
zO4yynACh;Dvq_uqs03@4K+01w^B2nf;O<c}epP`@kaZP-3CayL%y{K66G-)KAp-1)
z2yi{>A>}Gj-YSmlVsa_b<b}wj$qb8vy^k3ONf^whTWRVcl)xT^z!cw?gup!#fde^c
zFH5PlY4l>5WR^znVOF3{qnF4S8sE~VQJ*?A`0#IMXUiGnZ2q{sNW5gUa?wa-c0wo9
zuat@UW9}r0(*u@4?vdp(C&jm&$i*K~+(}~fYRdO%_)^Sr4*$Xvb1^3&7gDin!34#w
z8geUAU(9BzZ(vc*#ks1&7)gMTUuJk)vhSyw#IZ%@gQU!v8dz%-Hl*VgB?Ghwnf7|b
zTA#H9kg%*c1ybGP@ZCCxZ4nyS+t#;KoY%KZoZq)xtfRpjzd6^jAvW#g)nXOvRflw?
zysPHswO=!<L8!$y>mAN|p&@ucn=+8{fmA&>NU8^=)6jG1P!ALrXc&xPp0f+{Mpn^!
zE`GVz4&`39L1-L#S}w7h#D&6wn2P&?hQbV0#r-gDT4Xm!2~D$o2{NHS5K2i=Dy~`@
zx#ZPMpzjhL5Ek~X#&0w@Y_VlddaQ+rHmxRx)J~9W#tJEQXtw7PUs4gDH}DH_(fld6
zz0pyqHXM4@j=QltD>S!t{)(&=f*o^(9a&3?GPTZ$p+$`jx@m@bQ`}TnyGJAmSnOTj
z>>JRiXPvxh0^3|4g^^tT-lJ_Zdl7nm?}omOVsqanv4s@psrA@(%sj1g6I);cyNVc+
zVqXDTopP&F)EK&nSv4zWm1S=A*@TwpyRlVcP9>04x#l8Me46qolfj=h<T;8}ho73u
z5!qY_+z7Y)o?44nYb{2fS7R+(qZYvpqN6N)smmSPWQ?y)Mvkc+O*BFF@IR1Om~`y&
ze1+G4yiXh~b{&;ndBEWNF^OHrBeAQOzy8Fe6$MX>l8=E`l@EeSHuM$Yh*kk76N8S&
z5WbBN{z8QC9sG?aCtNTGUmQ(bQL3K6vli?4S@_E03gZimpj(UG@Fje2u`MvU(%r-|
zdpu>*llc;*K>ntH-(PH5`VAtOlagSFQ}d~ul2TTfxDX-O#5Tn-or*1+yGSHDEwm+y
zMAA3vGsvJ0p{NocYU-w+MQ5TJ@xX}obi%TA01b-Eg=MYa8MnAXST3xfrcsK{MpF{h
z6zv~QCpFIHgM|c<Rgxbo9M!OJgkBZ^{rpX$1}h;yaZsW`zO|(?p!F#_UnNn%If=J{
zti<9+LlNFb(76OY2Fcbuk&lU9l<<SxG%Grv*xNLVEBPPGYfakQ&?4zF>}{>Doy<p6
zYLw^Xb83OU14J4ZBGO=&1^i{Jv*QA}-Msh2lqEPgf?2s`kvw;@j^%<(@KWfbU)*s-
zJ-=efYOnZIVm12p<og*Urn^d5MgCU;yS<W{@hUWk-EB6<MYUFrL6Ts|DdiuMleLPT
zjC`6_YL3kx4I9xPlTW3W@CR)R;h*3&jdmS+1)ujgjO{}EBy2u4CI!QPw$$N%Rmh6}
zUo*#`P__=RYQry=PAwUh-HD{1_z#~IuZ3nT`#2@%u(_`vdG|IH9!}bGu#@W^Xb9QX
z-G}flPTw%E>~K+6f1iJ^m+$HKO{*yM_4_)IcUR~By?lqilk4moXeb<qg`t)WCI}DV
z6&v3nKGe0TzTgzumZK+NJiDLk>F)7$?(N{b1O9&BfR{Us_igH$`4UChMVbmTM?VyC
z%V?vFvf11q9rH2(Hf%1r@N6C4xw##F7l$@CyQ8=it-{Eb88`VkLNd?2gsod%D?3W(
z9;3}}(u^m0eO>*i$JZ?_f0oYEacA+*wmfH&Q1QmCv*1~La_a_%!bF8WmO}j5){UtO
zjC&bx*mejG;&-;SyVSdJuhV_06vz(s+v{S$Y>-kU;hYVf*?ta^51e^BoR-7c{eCKM
zA~IRVc<zq>g#kRHqb)bl7L*Nqqa)vz5QsnNI1OE)L(f)0cm{u`^GD`!qo3>X@z}q!
zkZA4vPGn-D6!C+dLvS!O*mYZnyLWv7pU1npcfe!#huyW3bOtZ%F3b9ulKtPbxksD-
z!Si-ko9@%@=eT!wtLY2c{Stq7_ga{a%{_OUY6}7Q{0R6B4?$Yask*_sBdxzMB#))!
z4S7gI{%#raZyWN*4Ebz$%FxEKlquxzmZ9Lbp<v8V5PHzF8s;%z%UJc2v6L3+@0OwU
zwxM;*(27^=DTYs9>Dyy*eVCR#mNS2(pETs}R$ATdw7Rjhx_?Qtgt+H(jQHUR8@_hj
zYz(IXcNLy_(j7`ZVK4>Wo@Qz4N&#;-E-j$XpXRkV==0W`X>0SryLphlyqll0&aHp9
zZ0<Uj{ymqG^xt!rQ^0Mrr<6Y5uk_^6=Lbp6Z1&y4=5+f0%j)JL`ur$s*);n6q`t{c
zpW&>0e-#K<!lnRyt4i5dq7Tn6-&df&S70RldnIi7U#GUXDf(|_G&$+>zOlKOK7Tt+
zTK>U2w)(?dgD?d=%xAPdoRTe6>mOEoQYq%mriPX}*73|SG<oX4P0xIWcGFXr-PD$>
z58P}^Zdt@IRwQ|5vd-%IrW)2+Q_)nyI_uJ#Y^<}sq&bsyHqM#Fo|~q67-u)1$!Mx&
zBOQiTBkOdfq%?I^>Th;giG<6^vz&}q>br=QC_iK9^qJETelKJ@`y)L(ia!|oduYVp
zI=>v=4}Egp0igk#a3j2fTkstMF1&aK{!J)&(U1ho@SDT!xkgA#ZW(Rr$eQD(fqstn
zBJStdd3h&H4hffUn{z16c(T{Gzt78c@?LHn;GN$dft4Zs=ns>+X*$8}qYXQiM@Vly
z#|Dqn!4tHhl^*Uz(hz^Z9250kl6N{uM|B+Y3TNo(Roc+fA9tQM7f3_lxaJMlVulst
z8z=(3DRS3n^A{8~;u1BOY(L_g-dMHh7ZjCNEu~E~<S_U00fCMud-`^I5%YL=DfYk8
z<_@-8n-A~fmDdUyKBogTE92g$jhzaS-s`!KXk#W#MV93KqArioaTs5@R!J5ayEY4M
z4e8z-F~N62SH2G+JQe!(Upin7<$2uP@AvvTI(wKHDpI&RbX7*)hqM>N0si<0t|Uq$
zk?tv+{?-imAGrRl|FY5<<EaB)zT<$G+u2Xz2hm*p_37~cLg@NNNQ~n8Uu}Y4ge*5+
zF~VQrKmXg$U{dJH-`&xfKA}5^_=UGS;GNKiZxhNs3jO;(tTzO<XTg*(pzY66Gs1fI
zY66*-u$gs}K$bO}%(^MS>IkQ@ZW_pOU7C8)9ZqMj86eF%<QZHX&SbAyASr7|e?kmr
zv)3GuZW~%Z_;lF9UhN>;dC7J$Gwh(RD4Yu%?o0bG_J^Ge>;k!kBgTt?a2|Wj2idmp
w6xJ;Ogg`gzmI0F`T+X_Tbcum`W}Qy=86eF)J<uhcu!oZv%JZRupDfh<KUC%kE&u=k

diff --git a/litellm/__pycache__/utils.cpython-311.pyc b/litellm/__pycache__/utils.cpython-311.pyc
index a9133ee9da266a63eee9cbbe44c1716b9277f92a..06f749149e3995d0d254cb1fc8f0423af3e19af5 100644
GIT binary patch
delta 385
zcmeyipJU4b4&LRwyj%=Gu-{BNHMo&?D=*{BPfR{6lO5DWrtkU0sK=4Xu#Aa;VKvkA
z51$yzrWe#QiYcV9q(cQkIzgZch-=wXSZWx{7>byuSA1p+=jR5?fQcH85}xTF?HDDQ
zU1~VMW=t3O!YDhvu$EDD`h*FLLeoROFq%WS3%)Scf?1RA*9mNQ`pW3e#Pb|zO3^H!
zGm54&O`rOmaV_Jl>CQhGRXJ7&T@=x|BBC=r?+2r~#&s3bODd)tvJV(tRB^ka;&xHM
zt-<pKyU>K(3+&23bc087hV}&>we8z~Fp4uNsyeB<=`kJUc9a8B$Mm>77+8;4*t%P@
z9=B!%vXhyoZ(P7AvYq)i<3UEoRojpLVGL(v+_qizALBem#sk~;{AZM9WV|r-7SmkD
ytJ^CWnSz-ZZ*D)s%;e3^_-nfaH&ZkVqviGzex?ol{AP?I9~j_7!E|GJro#Zoig?rj

delta 343
zcmdmTfaBYK4&LRwyj%=GpkXAPD%Z%nm6vhmCnhzP$qwow)AxL0)SF(=#>hYY_b0|&
zp%fNLkV-J90+O}t)8~C=4Cm*BiUFw_juNivAMF?=nO$l)rZ-$?)ZnUNNMW-9OHDp-
zLwtJX7e<fC$}M8keQbf|Ui-qBJ6-rTquAsJb^O~izB0Nq@yudkU?`dnBAzo$-~F9&
zEo0O4%pZ)ZY!^i|u83$%pZ0^%Txf&#6&1US0(K3aH#kHmWMAM=+Wz$iqd1eIkdvyL
zG}93aM>!yMOq$DGm-U#ct-A*6aSc`=JDGX<#s!Qb+Z}&19%N)(xt;khV>lz@*6q>%
z80RrE?%)3FKcg%o<N4_lj7)PGuWaAO$P~=Pcw;*k3zIiH<InA3+)U9dj27EB@iT4U
T=Qm{(`M>}t@}?KcGaUv1wdHB{

diff --git a/litellm/llms/palm.py b/litellm/llms/palm.py
index 0521f0b042..8a7203d891 100644
--- a/litellm/llms/palm.py
+++ b/litellm/llms/palm.py
@@ -56,48 +56,44 @@ def completion(
     ## COMPLETION CALL
     response = palm.chat(messages=prompt)
 
-
-    if "stream" in optional_params and optional_params["stream"] == True:
-        return response.iter_lines()
-    else:
-        ## LOGGING
-        logging_obj.post_call(
-                input=prompt,
-                api_key="",
-                original_response=response,
-                additional_args={"complete_input_dict": {}},
-            )
-        print_verbose(f"raw model_response: {response}")
-        ## RESPONSE OBJECT
-        completion_response = response.last
-
-        if "error" in completion_response:
-            raise PalmError(
-                message=completion_response["error"],
-                status_code=response.status_code,
-            )
-        else:
-            try:
-                model_response["choices"][0]["message"]["content"] = completion_response
-            except:
-                raise PalmError(message=json.dumps(completion_response), status_code=response.status_code)
-
-        ## CALCULATING USAGE - baseten charges on time, not tokens - have some mapping of cost here. 
-        prompt_tokens = len(
-            encoding.encode(prompt)
-        ) 
-        completion_tokens = len(
-            encoding.encode(model_response["choices"][0]["message"]["content"])
+    ## LOGGING
+    logging_obj.post_call(
+            input=prompt,
+            api_key="",
+            original_response=response,
+            additional_args={"complete_input_dict": {}},
         )
+    print_verbose(f"raw model_response: {response}")
+    ## RESPONSE OBJECT
+    completion_response = response.last
 
-        model_response["created"] = time.time()
-        model_response["model"] = "palm/" + model
-        model_response["usage"] = {
-            "prompt_tokens": prompt_tokens,
-            "completion_tokens": completion_tokens,
-            "total_tokens": prompt_tokens + completion_tokens,
-        }
-        return model_response
+    if "error" in completion_response:
+        raise PalmError(
+            message=completion_response["error"],
+            status_code=response.status_code,
+        )
+    else:
+        try:
+            model_response["choices"][0]["message"]["content"] = completion_response
+        except:
+            raise PalmError(message=json.dumps(completion_response), status_code=response.status_code)
+
+    ## CALCULATING USAGE - baseten charges on time, not tokens - have some mapping of cost here. 
+    prompt_tokens = len(
+        encoding.encode(prompt)
+    ) 
+    completion_tokens = len(
+        encoding.encode(model_response["choices"][0]["message"]["content"])
+    )
+
+    model_response["created"] = time.time()
+    model_response["model"] = "palm/" + model
+    model_response["usage"] = {
+        "prompt_tokens": prompt_tokens,
+        "completion_tokens": completion_tokens,
+        "total_tokens": prompt_tokens + completion_tokens,
+    }
+    return model_response
 
 def embedding():
     # logic for parsing in - calling - parsing out model embedding calls
diff --git a/litellm/llms/sagemaker.py b/litellm/llms/sagemaker.py
index 4f1ae1acc5..8bcc8d51d1 100644
--- a/litellm/llms/sagemaker.py
+++ b/litellm/llms/sagemaker.py
@@ -6,6 +6,7 @@ import time
 from typing import Callable
 from litellm.utils import ModelResponse, get_secret
 import sys
+from copy import deepcopy
 
 class SagemakerError(Exception):
     def __init__(self, status_code, message):
@@ -60,9 +61,12 @@ def completion(
                 )
         else:
             prompt += f"{message['content']}"
+    # pop streaming if it's in the optional params as 'stream' raises an error with sagemaker
+    inference_params = deepcopy(optional_params)
+    inference_params.pop("stream", None)
     data = {
         "inputs": prompt,
-        "parameters": optional_params
+        "parameters": inference_params
     }
 
     ## LOGGING
@@ -79,46 +83,43 @@ def completion(
         CustomAttributes="accept_eula=true",
     )
     response = response["Body"].read().decode("utf8")
-    if "stream" in optional_params and optional_params["stream"] == True:
-        return response.iter_lines()
-    else:
-        ## LOGGING
-        logging_obj.post_call(
-                input=prompt,
-                api_key="",
-                original_response=response,
-                additional_args={"complete_input_dict": data},
-            )
-        print_verbose(f"raw model_response: {response}")
-        ## RESPONSE OBJECT
-        completion_response = json.loads(response)
-        if "error" in completion_response:
-            raise SagemakerError(
-                message=completion_response["error"],
-                status_code=response.status_code,
-            )
-        else:
-            try:
-                model_response["choices"][0]["message"]["content"] = completion_response[0]["generation"]
-            except:
-                raise SagemakerError(message=json.dumps(completion_response), status_code=response.status_code)
-
-        ## CALCULATING USAGE - baseten charges on time, not tokens - have some mapping of cost here. 
-        prompt_tokens = len(
-            encoding.encode(prompt)
-        ) 
-        completion_tokens = len(
-            encoding.encode(model_response["choices"][0]["message"]["content"])
+    ## LOGGING
+    logging_obj.post_call(
+            input=prompt,
+            api_key="",
+            original_response=response,
+            additional_args={"complete_input_dict": data},
         )
+    print_verbose(f"raw model_response: {response}")
+    ## RESPONSE OBJECT
+    completion_response = json.loads(response)
+    if "error" in completion_response:
+        raise SagemakerError(
+            message=completion_response["error"],
+            status_code=response.status_code,
+        )
+    else:
+        try:
+            model_response["choices"][0]["message"]["content"] = completion_response[0]["generation"]
+        except:
+            raise SagemakerError(message=json.dumps(completion_response), status_code=response.status_code)
 
-        model_response["created"] = time.time()
-        model_response["model"] = model
-        model_response["usage"] = {
-            "prompt_tokens": prompt_tokens,
-            "completion_tokens": completion_tokens,
-            "total_tokens": prompt_tokens + completion_tokens,
-        }
-        return model_response
+    ## CALCULATING USAGE - baseten charges on time, not tokens - have some mapping of cost here. 
+    prompt_tokens = len(
+        encoding.encode(prompt)
+    ) 
+    completion_tokens = len(
+        encoding.encode(model_response["choices"][0]["message"]["content"])
+    )
+
+    model_response["created"] = time.time()
+    model_response["model"] = model
+    model_response["usage"] = {
+        "prompt_tokens": prompt_tokens,
+        "completion_tokens": completion_tokens,
+        "total_tokens": prompt_tokens + completion_tokens,
+    }
+    return model_response
 
 def embedding():
     # logic for parsing in - calling - parsing out model embedding calls
diff --git a/litellm/main.py b/litellm/main.py
index 5b1871b282..f991889df5 100644
--- a/litellm/main.py
+++ b/litellm/main.py
@@ -977,7 +977,6 @@ def completion(
                 encoding=encoding,
                 logging_obj=logging
             )
-
             if "stream" in optional_params and optional_params["stream"]==True: ## [BETA]
                 # sagemaker does not support streaming as of now so we're faking streaming:
                 # https://discuss.huggingface.co/t/streaming-output-text-when-deploying-on-sagemaker/39611
diff --git a/litellm/tests/test_completion.py b/litellm/tests/test_completion.py
index ba9390f16b..c01b605d92 100644
--- a/litellm/tests/test_completion.py
+++ b/litellm/tests/test_completion.py
@@ -926,7 +926,7 @@ def test_completion_with_fallbacks():
     except Exception as e:
         pytest.fail(f"Error occurred: {e}")
 
-test_completion_with_fallbacks()
+# test_completion_with_fallbacks()
 # def test_completion_with_fallbacks_multiple_keys():
 #     print(f"backup key 1: {os.getenv('BACKUP_OPENAI_API_KEY_1')}")
 #     print(f"backup key 2: {os.getenv('BACKUP_OPENAI_API_KEY_2')}")
diff --git a/litellm/tests/test_streaming.py b/litellm/tests/test_streaming.py
index b0f0abba9d..f016ae97db 100644
--- a/litellm/tests/test_streaming.py
+++ b/litellm/tests/test_streaming.py
@@ -709,7 +709,7 @@ def test_completion_sagemaker_stream():
     except Exception as e:
         pytest.fail(f"Error occurred: {e}")
 
-# test_completion_sagemaker_stream()
+test_completion_sagemaker_stream()
 
 # test on openai completion call
 def test_openai_text_completion_call():
diff --git a/litellm/utils.py b/litellm/utils.py
index c030713628..e25dab254b 100644
--- a/litellm/utils.py
+++ b/litellm/utils.py
@@ -977,9 +977,9 @@ def get_optional_params(  # use the openai defaults
             raise ValueError("LiteLLM.Exception: Function calling is not supported by this provider")
 
     def _check_valid_arg(supported_params): 
-        print(f"checking params for {model}")
-        print(f"params passed in {passed_params}")
-        print(f"non-default params passed in {non_default_params}")
+        print_verbose(f"checking params for {model}")
+        print_verbose(f"params passed in {passed_params}")
+        print_verbose(f"non-default params passed in {non_default_params}")
         unsupported_params = [k for k in non_default_params.keys() if k not in supported_params]
         if unsupported_params:
             raise ValueError("LiteLLM.Exception: Unsupported parameters passed: {}".format(', '.join(unsupported_params)))
@@ -1225,7 +1225,6 @@ def get_optional_params(  # use the openai defaults
     for k in passed_params.keys(): 
         if k not in default_params.keys(): 
             optional_params[k] = passed_params[k]
-    print(f"final params going to model: {optional_params}")
     return optional_params
 
 def get_llm_provider(model: str, custom_llm_provider: Optional[str] = None):
@@ -3441,14 +3440,15 @@ def completion_with_split_tests(models={}, messages=[], use_client=False, overri
 
 def completion_with_fallbacks(**kwargs):
     print(f"kwargs inside completion_with_fallbacks: {kwargs}")
-    nested_kwargs = kwargs.pop("kwargs")
+    nested_kwargs = kwargs.pop("kwargs", {})
     response = None
     rate_limited_models = set()
     model_expiration_times = {}
     start_time = time.time()
     original_model = kwargs["model"]
-    fallbacks = [kwargs["model"]] + nested_kwargs["fallbacks"]
-    del nested_kwargs["fallbacks"]  # remove fallbacks so it's not recursive
+    fallbacks = [kwargs["model"]] + nested_kwargs.get("fallbacks", [])
+    if "fallbacks" in nested_kwargs:
+        del nested_kwargs["fallbacks"]  # remove fallbacks so it's not recursive
 
     while response == None and time.time() - start_time < 45:
         for model in fallbacks:
diff --git a/pyproject.toml b/pyproject.toml
index 44398321ad..3747784598 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "litellm"
-version = "0.1.811"
+version = "0.1.812"
 description = "Library to easily interface with LLM API providers"
 authors = ["BerriAI"]
 license = "MIT License"