From 4a5be27d4feb7a8e74c53e915154689869144868 Mon Sep 17 00:00:00 2001 From: Linus Gasser Date: Tue, 6 May 2025 13:43:07 +0200 Subject: [PATCH] WIP --- data/C4DT/projects.yaml | 28 ++++++++ data/DCL/projects.yaml | 60 ++++++++++++++++++ .../products/images/inference4all/logo.png | Bin 0 -> 22714 bytes views/products/demo/inference4all.tpl | 22 +++++++ 4 files changed, 110 insertions(+) create mode 100644 resources/products/images/inference4all/logo.png create mode 100644 views/products/demo/inference4all.tpl diff --git a/data/C4DT/projects.yaml b/data/C4DT/projects.yaml index 9f66ef2..6022c2b 100644 --- a/data/C4DT/projects.yaml +++ b/data/C4DT/projects.yaml @@ -124,3 +124,31 @@ projects: title: Report 2024 matrix.epfl.ch date_added: 2025-03-04 date_updated: 2025-03-04 + + showcase_v2: + name: Showcase-NG + categories: + - Other + applications: + - Info + type: Application + description: > + The next generation of our showcase updates the UI to present our latest projects in a more 2025 compatible + layout. + layman_desc: > + The C4DT showcase is the list of all digital-trust related software projects from our affiliated labs. + It is the first contact point for finding new projects, mainly used internally to communicate with our partners. + For every project you find a short description, and links to the papers and software. + Some of the projects have been evaluated by the C4DT Factory, worked on, or presented as hands-on workshops. + tags: + - Database + incubator: + type: incubated_market + work: 2025/Q2 - active usage + url: https://showcase.c4dt.org + # information: + # - type: Article + # url: https://c4dt.epfl.ch/article/report-2024-matrix-epfl-ch/ + # title: Report 2024 matrix.epfl.ch + date_added: 2025-03-04 + date_updated: 2025-03-04 diff --git a/data/DCL/projects.yaml b/data/DCL/projects.yaml index 96b3ac5..95ff54a 100644 --- a/data/DCL/projects.yaml +++ b/data/DCL/projects.yaml @@ -644,3 +644,63 @@ projects: url: https://dl.acm.org/doi/proceedings/10.1145/3575693 date_added: 2023-03-13 date_updated: 2024-03-22 + + inference4all: + name: Inference 4 all + categories: + - Privacy + - Blockchain + applications: + - Infro + type: Application + description: Distributed ML inference across office computers for privacy + tech_desc: > + The system dynamically distributes large ML model inference workloads across + heterogeneous local computing resources with fault tolerance capabilities. + It employs an intermediate representation dialect that enables seamless + integration of new models without requiring manual hardcoding for each + application. + Unlike static distribution systems, Inference4all supports dynamic client + connections/disconnections and can efficiently partition large models + (up to 70B parameters) across limited hardware resources like standard + office computers. + layman_desc: > + Inference4all enables companies to run complex AI models directly on their + existing office computers instead of sending data to external datacenters. + This approach preserves data privacy by keeping sensitive information within + the organization while still allowing access to powerful AI capabilities. + By distributing computational tasks across regular machines like laptops and + desktops, organizations can leverage AI without expensive hardware investments + or privacy concerns. + language: C + tags: + - Decentralized + - "Machine Learning" + - "Byzantine Resilience" + information: + - type: Paper + title: "The Vital Role of Gradient Clipping in Byzantine-Resilient Distributed Learning" + url: https://arxiv.org/abs/2405.14432 + - type: Paper + title: "Byzantine-Robust Federated Learning: Impact of Client Subsampling and Local" + url: https://arxiv.org/abs/2402.12780 + notes: + - label: Published at + text: ICML'24 + url: https://dl.acm.org/doi/10.5555/3692070.3692116 + - type: Paper + title: "Chop Chop: Byzantine Atomic Broadcast to the Network Limit" + url: https://infoscience.epfl.ch/entities/publication/d83dd7af-b83d-4f99-b50a-0b6d11e786be + notes: + - label: Published at + text: OSDI 2024 + url: https://www.usenix.org/conference/osdi24/presentation/camaioni + - type: Paper + title: "Robust Distributed Learning: Tight Error Bounds and Breakdown Point under Data Heterogeneity" + url: https://infoscience.epfl.ch/entities/publication/83afd663-8967-40d7-a0d9-5418371431f0 + notes: + - label: Published at + text: NeurIPS 2023 + url: https://papers.nips.cc/paper_files/paper/2023/hash/8f182e220092f7f1fc44f3313023f5a0-Abstract-Conference.html + date_added: 2025-03-21 + date_updated: 2025-03-21 diff --git a/resources/products/images/inference4all/logo.png b/resources/products/images/inference4all/logo.png new file mode 100644 index 0000000000000000000000000000000000000000..41358062633c62f7014ca3cf7eded63150b1eeaf GIT binary patch literal 22714 zcmb5Wc|6qL7e77=W8af~jbsnm*FqE}6eHUpNyfg7eI{b;Nw#b)mSmljeVH;*)*?)V zLC89mFxlq&^8WmOkH`1lugAl~%zfQ+&$;J5&vNfQFNwEqny@hPGlD=MmTRU)Rv-`r z_y_^f!+<}3Q9llVKMa1Rc7Y%eYuD*7xX1U+J)n^<$QT}E?R!5c)Fr?j6dD?;;OXNP z=;q?*uHYNsk-wqC4+4pTt{GjvjVf513N3KFGfLeWDlvpeyrg-;2^lU)3Nn&@^om~Q z)v$0KT8N3>=GJX?9KzP%jI5}*afbMdGu-T@rWsG;IPXFj#B}ox58d3OQ|2WC$yFcr z^_Hf|hr1Jss7MWuG0pv=`u!<)Z+4FVr=JW{$5JSJMdB=80x?ob6=X>PUj+wZ?C|w? zC_W0G>Hg^qFkZR>!h>{=q=<*S_yON)z z--V0fGo@H8_KSfUXpm-+!**-mwg_|+WYZsLCoV2NT+EI0Ik9P_v@a-;uzz!;fVQah zggXRH*^`ci0K~u#9vPtNBJ<&BXWYB27>h7|qWS_Yz$vID3*#m2yu2@;jA4^(^Wosg zNK*w<=oVxT7qUifMY9487b_wAw>lpAmPJZ_*9A1Od_fkXZrS1LTN1yWNU zNh@qQEfnDebeR~pd{*1^x;F^P9g8*1eFyK?o0FNA z#$SG4ROl<{rB?V%$G37gg7zhY!5GbwpRna2A-;I2*PC8Iv#| ziLK*P%TN@r5sG!?*r{U!&8DyXCyxu$b6{NzG5nf8e9_y>rcLmg;!O!+hu z6=P^5dZX!kSev>!220;{olcbVp5)S2GTK8N;`wQQaK3%~u7?0b+pU>f4K~)3jZEv_ zR_p&5?_cJ|Cr0kc4xWy8cR$0Cp;iy>NDNNDpAxL-Q57@D>*Z;PZpQ?oE0|A;bkn_m z1)Zr27?Wk^$S4`>v8~X{QW_iAbZb{|t17KxEJ#u9Z>v0ikljp7%BT)>1lG5tUaRA} z3qwPpFYELaqvFsR4~BVkEu=oNmU5WwgPdDZv#T*TGJ4AUxc!?h_a|7M>+vI%LFao( zD&-0!Q&%g^#=B>s9#8e?;;>kWZJLfGQm(oG$pD{6FY>XZIjH7*%;%X#9oSkC<*u=j z1SWs$4k+de`}_paFIH;swF)7xuM*?=toA6PSDa-1VA7o<8WO&`9b;X1w8D|+P^^>jMTURhXq%}gXfc?;^G+_AF0E|tmceqvaXeLFuzDYhvxP4 z)cTwnqP?{K^#Ca&BZ(gsgtotA@EA+d@;c#8_%uuU^$B{j5&M=aBIaK|TmcLL=IJ)q zl2Z(JTELbvX<(Xg8xHy0r58C*t#bQgb{M{4HWUE;6-Sd_$uVM^-f|-dSto`g)r)B` zKUA{P?XNk~&WRCyg%8}(Skuz>Y)TGw2fvSws&iaAPywUZ{M;XP>IZAaG_I(Xw)7p} zl^0>^P!|{HTsu2_2NGDP35kJ3GyS5`xiaDIO-^ptwJUwVo+UHn#-hu?QSrvss<`w% zG%F_4_w}m>+8K*QItlBEel*z~pcmm?gT)nm{1KG6r>1!KT^zu0Z74>p2 zG+j7Lwmvknaa&+GRp;>ifz=1lUcjPT4T!#o#9hCvsm*s|lPvrQgU z@X6ZBP77$J{SU0cjCaP5q(8{JWHs~KTwGkJoJr&H$^t!#lc+QsytFu#wYANx)G@;6 z1{Jso*bx6uk;%QrpwBa)8oIHY;UB_Sx{43>-!S{vB*o-(hb-ex-RvqT=T^;J3|NCT zzxdnD^RumrW@uIA&j%hJRU$)Y%1xWJHjt`EKoTHT(c`H2i_uCLQwe*SBF^z80-+ zt<}c9k`pYy=-XoVR-?I#6%ZCA9XUP=!v4Cjob0sLKZVzC#%EidZQWCLA9z%Gc%qM79p7;Ow*V$@A6^1|E>OuZr6_Zv9O*< zE}F_#vhmISLcR~sGk>%}>zEd4Eh^%L?P{tP$38`eEl_S9WXF7S>Z_CJ{!HnxrCVfc z)*zq7r+cLAP7cC5pDNjGO`gqXvw4-)%nVmNmK+;-N&KC}DPnG6#~hg$elzXar5~UN zQ}LAIE{Dw3HhWqxTYNY#tl;{;)Q|UF42RzRU?ghX{mNGA`%h*4gMvNGf(1R6ikI;p ziqe)dEO5g-9@Jp#WiO{x8Fq**-0x%KI~p7Fc9R=DM1=RABPz2mBOij82k%AhufJke z5?)EcVS=U2g4>x<&$f53#Via_a{9-{?h=b8O7-IwviKf28I)T?5q=7U(3!KD4Pd=9 z)I$&9>>Q)Fo>FTaYCoQ{73&pa^l!@pn^U{w%o33W6yUCJ=8 zEhE~I&)F@dXV2M~q8GG8D5CV)LrPO*o$+YvWQB}UGjltJACoFQ8lS2UY1Z- zm0=}6^D-@EDct4)HixoOgEE1=XB#=KEK^f`gUzA_lF?L+KuNPevE1ojUo*WesXGw4>=pa zeTnei@iTcd�Z@ddC04#oh*GdUCyVaD$J#q@o|}gK`j=;bmY8m&uDYm|*=5;njoW z=`CNF+Q{F>q@C-#4k@*qFxg z3per!1M$h;7#>{m)z)9WLE6bmlONQ|-s;oVldJSQXbDNP? zH>P3u2ux+H>os@QG1N_WdnES_ZDjj#7}Cq5B5fWs0v%Fko0M4bf|u$)8Ng(s8#(RE zHy)`PJY0v&wOrkRHi+C4T!q-ubZ^s7TxsG$C4g3}&LS$VrHkDN(I(eK+_}Jph{!Pw zW)@Q$C5zOxxLynaZLp;*99{0VyGU=2-=KQ=i;a+RwcJr2U4kwGDo{&$3lo1((;lB$~y zk?LwSPd~vCEh-(bhUK{TwKsi0m(W88zOz({5p7N7&j5L$r?1=N8@j3g^bA{4Q6b$k z#Hh6AH;bl(sR@CX=Xmwcvd<@Q^;~cd4u_IlsPb#$919&AXLQ;#q7!}Q)+12xb&!6We|w_$#%}B zJH8KN&v}x1UpHCRR4>gPlljnEX#Uwd@WsKfBS$ey<~7!(t?AWg5A&_G(5x5vh}@MZ zv(oVs+U)Q0wxy!c)1l--G0PIF_IK$tYUeW+5PAEh=P4P(($;G<2M5Mf@)#Ec5^_}< z+wI^}tG#bo)){fGPTP;yb41LnYL?PNZYQ7I4_kegb>-C`lU^9QadI>7$px4iTkg6} z!g;0-wVl?f8_PGhc(VWacXY&Lfs!gkJ zYo>_(@D8?B_oQ8+L5gOdcl6ba*W>0d+!V!V z;_E$xYR}eTrU&mA`yTT|dhtfK`PIEY_csrf)xGE&W)r0bodGxzRv9nRybwe0?RZ7b zBuJNtbU2Doqd_9(+z!2B=opQ3@+UQxYs>6M^9GGqX@H%h{MMM#D%3Zf{+{Md{LGg( z?n3ntHLvlJZ9978vkM&=aGs}9fMukZmh}FBz7R%-yw^V2-#09=XQ@*<(eeW`guXWa z3BAbH^fN4N7O!bW}`>N9It_?jTRQrc^bR=bRiuERkN+aW{R_viapL~QLwy<1A{Tt)BiFl+2CUzFsKhq&v8?gz^T(#R zo!Lyq!voeNyK>1bJKklB==n#cd*+P|_AnOoT$VgK-J#56~^>i^A&E8(QhnL;9?dA}Ar7 z^1l)z{K-){KnC>UPf{nwTkMyb^dj5&6WFfptL{BInHg-muK2YI@C%qht9ItzTQ}CE zcHIR>lOlnf7IQiq79hcwt^gy9SkZm(<2d-$*ADg<%K`*X zQq%5jUPX9?O1ZfC>e_oNd>W(CPnFz9Rhb5fTYU<|OJVBG=;q$DqeRBoA81S6>Ihzt z?gV16GMWMZk{BB$GT{4FTfod{R}!J974bPf98~p&?!~d=m4D%%tkcqgV67BKbm!FU zJ4XpAdb*hU={)g>18&~Gv@fUj(qk!Q+>-kt2yO{|aa_%+bN%0e>%|3`VW}lqADvB+ z-kCL-;6I zx$*j)UC>|ZSfC~0Hsck6$`d+Cg#$a3Hg(`iAS4>%P+I(NNXsk%~+r-P;>{$Bk!XqFEtHgV%J z!+(l3VEx=ZW!7eU!myTm!nx|Lu^BzANV_Vnh-hci&1bmAZ-`-%c5YU;&j$IL1xRgM zLk(C!Vc#`Km$nD=!>4=m`E@NWY8_y|>K$jvGBq&KgF>n^pG$_7|8OB~hwE`_1_ zLBj?0TTwmJS^B>}dL6cPS^ZM4puSB~hxjT8#bHe)R6!D@Q4fmElV%8eGVC0lOwY3} zR0M&YZ=g;p@b{4M>we+D8u`lBTgy(l>T=Nq{S<`98E5=9DBqX=k&*)(}KyBoz|l zt$Ap>aax-j_?<-T82}R{b$tG-OCsIMM$vg=F=6`Xb^1MFq}Mlk6WT%(qml~%*T?4` z4F<~AK^uQEgvyd%Pm!uOxTVVLVfhUTRa>;aR>*8v@r`Paj}{= z5ZR`})$}opRTX4ZG6o1>YFHQ4COmFTGaRR{nI3b^^o+Rp<2PW@jBZ2(h}kCg?-W(H zEk^2dG*n#t=V^Oe)?5(65;(rfUDR=60dZ8Q0vw<2AD+o5ks6;Fu0JRKp)4oYXuE8} zv5+`*nBAitPrdTrs1LLeiGLw`ec^RGpceE8Ooa=F{N%*oyhno3)=Ud}7e};V)*-*e zEWNES+=|)IZNbG^SzEf@dUzd>{XHl`x@K9UM;LE$?UHfAyr~WGCG#6$2%zeGWSUb6jRda38UtrNRE7HvT3N4URh}!IwCRY!s$V z-E9_^#HI?c==Y5f;ZDUm8za%aEmlO_j$kM)VM|7CXZ9g@>FM?~-4XMQ6+8o}RSkYY zr_yd;&{JzL(GI!_Ee3sfF^!d>zHrWfe9gU58m23R<#$c?u_3eSJychUzY!du7P z!AlwvBtfeiUZ7fOp|~#wr`*NRItpwWJqLJWd@pzjLgWxAPNfV+2!Oo6nUJ|BD-$1r zK;qB}d#FJtX?3>z^3qQhifmoRj!pt(oGY_qU-)Q8(8?kp#U^X1M#lfrIhh%>sdBIv z$W-@;sY&&ZL%tjLhRK7dPrt(PVfZ18<>f7-DoGJOzqEq;&sY%^9%gjk9<@Yu4`Tr2 z+#aOY8P^~Q0Ego}@aX-2w-BMEN#eml7jz!Qsv)%*8C*w$dK1KMrh7{Zlo9gE9R0M{ z7af5afDAQ*H$RjvI3eH7UP4|yp{be-78tIsq@F$bNevF<8r!UR)Q1VpX1!A4_@f$3 zDkRvJD-l{&VR{Tl3g7>!pKI_}dQx8JguNLUz6=U)K9e7d3tKiYnz%d_4{GCXvH|}N zX7Dp!7JVd0*DoVvbe{**sVHOw9K^7JlCRT=53{a}fu0` zMY%IX8!RFtWSV>yOy@RFy`!Am61||#@>eIzJa9hmxAQnG~o;0Jg;(A>-U2v{0OsKlUEIgcH z%syPW?pc?<{axWS25Y12QlCIjNz|pfA1$qjPKa5zKWm-P+oOCOr<##6->P(&uD`y7 z8X;R$59EL?4L+M;1ORH-hpwRq2*iV$R0|~E&(`HOL25Psqw1vGN4pbMc@$rvxF~f6 znz0*{MkB3kevlPyGVWSvCR#D0L(Gq7Y^qbp&v34Dl(hkT!WzlOTf9XMabdwkpwnOP zXPI8my=5h-f*As>f=t}LI<>?O9z!-|k-Hyli2S}cGz5r;=uX#bq1X9hy%nQq>#ly( zFq!~DyOnVurXKBxyXLB{Jrv4l<1ct6YMcF;8ib;`R~d7A79PG9%QA_y1@I8^&9f3@ zW2l(kQS@mTub?3GMOmlgJ?a;2KNGgpIt3hUrdrQZlPb>_$<^SJ!8~{oDcln`jJe(a zuoU{GVuT;$`}?z`$@Z%ybRTA7hb=xn(!urmA2Q_Mz>G?p1^=O^BS3>5U+8F?0jOYM9me0){?ne2M45 zxjPMqMVm&cc|lgIDZ1GK@NBXdehKH{KywRfa1biUFSC5f1Y(lK>dJ0I8=bmKZ%UnJ zj(i`)Zl(JW^yOVaT8BuXt2L>$G}e#E(I+BC^`cj6H2EF#W1Y4Zw;ZYVQTgw+!$XBS z*JaTHJw~Lx7M5{>0p+!`XQ zQ1u&oD9H!$a9Px70$+*E7YxyFu=|!%7bSRi%LBoKtS`PmX4J_stuI3a-}VOF);Cax zB@hbK{EWVZ`QW>!Mf?if82ofLeSqv>ziS(c{rz`$82tXN4+{uQ=9+ZBFzgu*_GQo* ziY1oms6Yp4{aF11@L^XG{)xZ0)k$=vu@CMD0lQyZf_?rNmU(aw-#ZxCPsa+nPwD!n zSPsg86?oL51F@Iph2%ydO}S{=zS5_qRwg! z)ze)UJntzF9_(IuCB@jId?DB#Mx5Ls=FWN&CuY_$A8y^;oVKAGlNE})ofg{InO8C( zvUzc=@W7=<^E^mEPB({n`W#%3PKgQEdvX@J)PHy2fLuAqf)-|bRyuz>0vO7N9e?4H&SLD-<|tb!g3y2aC!#iXDoqV`TnH4wvx?SRbW z{gVsZNAf^!lWoWX${3uBS`1WZbzIZw5IV1!ECD2QrXE`%Gqv2D#(`~YpL1UdYC@O@tl2=D3V~FAEH>)dxB9 zG-Jck-ve=BstcBK=hzHOqE4j~S0BI@x z*z3BM9UU^PM&f?JZ1HzSPzLx^&^(2%PQkh!SxhIC4(uo|+^V<9M3O@o-O}*@^4kBY zH18~QhEZn*hJQ??=`zG=Li&_x1oSkx@+)FY5rXNUP6nM~)5+oceBy zzhJ)wu#g7|o99W*!x+<%cyKMKFROBh&4?QnA5kFE}U@SAN5 zcjGGXrrGnV$$W26!K2`D)4Q?=fhy!JE%s&j2)g~vo=uAxu$N1T307ukqCZCW&#uFL z@#yECASAuu5~ImorRPLuZ}IBzGemBR*|Ncb5* zY~OeJfuC&f#vu0L??s=(AstsT=M(vUrK_4N$O~cL-;1g`bl&D;X!CeMeUk@6cbDu4 zK-wCB-dktgGNai*!iu~m>U+slZsI+=ik$IRl1q}z@Ptb{FU|k?XyTFEf$&P)m@al70><;B8SjjX3C9HAQ zKH#tQusY2KjI}#@ccy^i`$5I`_18mI%vd8fo7au*fbEC8wW9@XLD%8tO$O}ciN1Tt z^(NF#@C~d6MHcf`LNxXs?s1AJE){a1lOlW^EkcAAMXG zoSbwVoR&~lEXk%H zP_hWJ_JSv(W=C*3l4dx8AGCE|E4`SLJ+|>f4M9kJ{qv!D#EdixQu9TR(bCrc-2rg( z!;?`!v4bv%6w_EiCz=Zow>b}249-iD^4FjzIghcollSw4(_=jF(u4D(Ao?Qc## zkfKGM=|i7Py&oW~DkGJqjqg-y_GuUvdwV1k_zWD3w|Wa+{1nEz44l`&r_Zi9Ac_Eu zB>a_L+7buua6Bdtee$6Fp@K8OR`5=^B~*0mEkGxKZ&UpW;L1HMriIa11r%BPCJWl& zA&7AIb6BJi)Bp&P!y>*={GksRH5PVs9jx@l3V6C5tqH)!65pFPF909(;qcT4E{WiX z5`bf9!P4>hQx)_bF8tLKoK|XCLemZgVog;CdWxC9Xfp%&2R8WZeUnhqy6{eo!MEC& zjJ?Ied6hvxBd1%;Q})|~;M{qgXxgQ%A?%n@r0BGfbCm|?OLFp_Zv>N$GiJR(s2SXFGCv5{R&7o00=~bY zGbm}Vpffnp2N;Cemro}c0_u6EQwo5=>8dK|boIR6?4ZbkAo5MS|JEkI5i-VZAaJT} zSAf4_<32e;;G(#dcd3+2&dK3_!>>%Y?43k$RH*qXx-5qloR^`qf{JE0LKLPRlaCKi z`^MaR=7Tm>{GZ;<^aX*~`o3|v6K|$&P;xri(J0Ph0L$yu8~VyWpzNjq07^5F&wL62 z^B6rg0S;78MXUUKI*AA6@_@FZy3MFxZm}7N0@PW9zt#J~jsP?mmrS$`HSG&oYh}Lp zpQ3!i=L>p<{DeE7@1Ot)$toLBJ*XM~3EZV8fMSY*g~{H97NNrWtddYpPunguZV<3o zrBVUW|7Lcb3Mo#meXK(h!{i8DCf|9{=YPaRk@PZm0|x z*uErSz?Xor_NS~KYto;}ed+>sffUbQSpR=KBmf>T{K4S7D){25izouj7kCm59AW8m z74}$2AMF<6!5?vlwt z=XIejPX)$Nk)1OeF{4P(|Jd}GJ?_6w!;YKD7?Z*!V(;lR`_2x|m%H|D>YTc^IUT2i zTTszy;C_e0ZF65UeMVUn7=zXK^~p4vx2-*&jgBCCs;{9*Xr1y&pG-69{ zWID8$AJ4m_S*DeqbLvd?bxwk^1T@oap$10bJFMVe>pIbffYBlHIj#ZV7~b52pjt_Q zt$+Ad1R%`y2IA4FjVCpNUooH=-hiK61x)K04_VR-0W|PmJXwOWX$;M--zhkiGvhUQ z^Lk{&t?`V|PbvE@c5yM{q?%Kt@wW|a>r%4M#!^jg5F`j%%b!*TO9C!|P97bL>vF)g zyJjjMwE{siEB3ApAes79i91mmlP5yU&Y3tlpr-o5UmC#oC-d=}U8~L5)`y~nd z4fIHzcXsm8MFmX;KS7`dQ;~F^G=;xTJD>QThSCVANU6RR{S#RFR#xK@IWmPRcSL@N zmBdcV#ofVi+Kf-742)=jsf?+YD|=;tBAR-%6gm)>RKCGQ*FJxLQZwE8QRk|U$f@tn z-sd}!JppV&q4JOt!(&rkm^|w5;S%iRKX3`OD3v~E{2Dk(5kTRBe@}gUdxo5&=6bfH zan(ot=x3`@Ny?o~+e_1-^pqKQj!bqlI8$ZUM%T*wb2u%9oG|Dy9WaJ41O2^=YTex? z!e9kj))FT`1QvAg=F(tVE?r&usRFy5C^LEL0QBNwt^^i_gW{odh~TkvaG?hOmvESQ z;V*M2DLh6{p$mPCNygowaV`~}9GwPVDp=ZVD4?dr9B+6P_9UC~)dmW8vNwGO5)$pI z{O~FunW)Scrj1S*wd+c{`om~y3K$1@#BKMC0q+lYP0B}yT23X#4$WCO%{nIQ3*;l` zW}_0ZrhL=ghp7}>6=Mmp=gT@KGO{kg-)_JQdMy2}IE%0#Yd!|3j)0oDDS6rjs8&R3 zrYAnVVlgGymnV{-gVqP8L5~z>Pr&}J=T=MkGaD6(rV5=^q7$3&ffuT4AekS`X@CnZg+!Bw&n;QEp8Pi#!f4evwLlpEQNn;4Xc576Jz9x~RsKmsVO0A!$B9(F`qiJ8oD+KC zBvoevpIw9!T+?wf+TPZyPoJBCpm>hV2F%nz@h-t%B#FlqMw_17KZh{m-+q4^K#Wb0 znIq4%Dorz6Yw9GjxB!Q2whCMOy7O0@ySlS=qEkM=bsGrOE4*KLk-xSVi2bwNIVWW8 z`PIQA+09(V_qQtN>>ZV$O==giFGRSGtHQ@!$M0~jlI*YS^X|1nBls+!Cz#;Mp{>Vi zFT6cp9fpEDToNaA25I!xG@|)8IX#$TaSUn=F&cY)ohxTuQ=j>6&V?#8kD}dhlKiRs z&m`e&@u8lMl?nmf>jR%?_q^~CR3JZgvx|ooe>#8o?4;<;n-Vz{+<=qSo_P^J`%*r1 zu4#s^ix#Z(-ZiQ!4L+BmZQIOM8Eb3iuW}csyBW2z1&QG&iVb-G8SjC(Istv*2#iYn z?RF_yXiPwN*(efMo(yhG2(GQd5%`AGd zpak%^%c{EBOqF|<-S1TfqL}Ek8)U9YPBhJyhVXroowN`3uXytV=WsnzYp)PGci?Rx zC_<0z%D>WNw`#pEQac4N4A-#f8T%9uZ`^wQtP^vbRXO5#5W~l49@>4-iq>Ea?EVId zW@k6?rSh<6CS5P=b0KI+iIz3e5&WA~;>*7_+765k!mYDRFp93hp582oY)wVb!^Wa}d|mFJGyXXpL!Zb~=0+829GQ zd&9(D?wBA0xg|OR&cZFWB_3!gB(?;&{kbdG~MN^`++e z0s4)UmzbF3Sz$UD;nk~Q)&-F3FMjqud&nFmO(kCPcO^Y_Hf7ZMzH2aI@Y-@#y|`NQ ztk$-Lpu;2I^#EC)!0rQ$6D8btI8X6@hxbAX^wLS~t6}Rjyl*~Qqo_y!EAm1%s-MfG zrmBAaYv=D;fyt63dZ+$CEAyIbXZuum%zo6e8Ts%Q_bZD%;ggj#(EG zfl6(9XZ+I2PfsRTPj^{Tvp3V7M+m~(P?YM)Z%vNnzI+_rzZJYxo!C+;E+Rjpc}0)!ogG$K}m#a2$Ut<@gE#(Ou9bW>w1*vaGq35niMUQJxBoi zqehC=S6jh2*1tcUQrG)#g?o`E?h54#W;~pG5!yF>pXIYvVYqA5)7-q5q1y96(F{<4 z@wU(R_U(Wy*{#C>1Hn)lf)X>To3AHmdmy@VoNx7A?-U%;l$Q4`r!Ni86nGXVLY&B( zqh2$piU6*GGSC`j^CDdni3cT9w=EMD%r-rO1fV(s>}G_z=?<`e0T;W8iW4Wh7g?E! z!tpO$QETEtmHom)*Ym84cLTOt(C`>u_MW$dAifTL(QKgL?&T}1o&fP^j?{n6#IbfK zeO9b{F2eG`@vS=Y7=#x?ot{hE@ZtYa^!f|tZ+d9Dd4qhJ zu;*Nr5m1zefgB>NQ?aL^LGJjGjpa4Vv#t>VNV60yL7ydZQg~nEl4u%sdf+~D7u8VC zRrQWP@LB}@_RS+VlUArY4MF+IILKCWXiEy26*um@m2P+AN?sgQ2D-v~9T;m&r3*f`^T(iy} z3(U}kJ9oo7$tfQ&^2b~JNykZ>-P;; z->{pUzgH=pahCZ3%=gmK@M8RVa9Gqu(QKHSfSu8>c3Q5-UKALYf}eT)6W_8BNR$vA ztMy2_bC(?l+nvvIU55xWWA|@>aaGtGkN$Qm$3`B?V1wRn5p-%twtNqnl)A8N8H$qR zE%4SxzO;?!M@EWz%hpY1&(A=4t=cA8@!R|k7jSdl@Y!I^CtxEi2ETOtsyeQRzN}c~r5ig^x2}-I2WFxndfBS`&_5 z`HZm)q6FZ5k@s%tTt+Xz=Dz1FH zUDB-$aP*V-?#**ZqhCyae4go=mDs(;t{ymp+AdwFNr!!YA9G7*0Bd?D(!Coc`6Yv* zPtm`Z(#X`LF1jd%wSTn#0gVdpRe$8l=@~jBw$ZGudZFw)J`=q4%cP@?3IgJ7q|qs1 zFWw7@OD(K%5+hH!z~>*T&W?4Qt;~=myBTB;YXFZ;k|AIiDPhn9~Z|+#G^US zs|3c%X1JrcY7qEDaqTPLyd${p$3p5qPuFb-nkCjqhr&{&gbTmrAijPd;{Q`a051?b zcQ-Frh`yK*uO6T&ExK4x$t^(?*opy{J#Fk;2?Q&ds5?XpoUw)qSoCEbo0TZUgJZ-u zG+RY4HdsG$?af1Y(_fSX$aVm-i>@to&c#ZLZ^rfn=i(K|VE18!AJ!db&g7W3Y7{mt zt=i4queKoQKbsJ3uzkmONJ}`<=ykJ-aQ;L!{?gtxl?YZ+M?2GY2bte8XF;ayyucHU zR@MyWx{n3$+EBM-Q>!siA1AANaT`pOQegj6JdCoB_Cc%$1SUTBp9DGaow4S@IX)Y5 zJIqfGdUpeDN6@t$cNuyZ=a}f3YgA+Yu)7a3x6!S}{#4CR^+a*YF3&CPZeWD0e;}_K zL1yDk&gc|j!0K#8{vE8{!6Xo~9rrlmT`Sto%_a5T7_JhbB^0!d+j460Qh8kU)Xda# zED1>NU!XL`l+`aUfW_c`P;f(Q$Pal1`MW;o0V2|b?(<)buuOFI)&7FVu`1niiK+PW zFHxOmioXi(A3|My5RQHsUUoxlNtNYC7cZ)1XlW!6J^n0VaH*0$nsZ7Rf7SUBu#&!^ z!}Z__Y%PiEYTm`q)trJ4`In4h?-bx2*KiG?S#^Bx`6MU_d;M)_UJ8qp(La}44s02r z`dJ)=wL6=mDN1JNSKT6hM05%^k6h`{6aRiLxo1BdMfA1^+Tcj@uA8`=;H6>pT#V;&-WqD(A{^tpfY@&Id(k`HQI?BMlCP(*{{ogyN?YXpI~y!mQ@rSv^1c5m z1Y+%5E?9AEL+cZxY-+^SD(~Npjb^BGL^up!bN@^kkc%2ZnNAeHPW@h|1yTGBez71U zzR5%8`Pp;?&)8e!F|>(ayqfW};p4jukJyK`xDIWY@+JDk>U*S7f#<>|; z5w^5yK92Slix;IaLgyax4SK2c=yKZM<>EMmkQP2}8Ow{TI2I9Fq>NYYdAT<3txL}^ zr0sAMf{$k`rrn1@hp^8l%l-alm)#PVZ+2>ilJKj7TYK{RocXUKXmE$Lf1mN;_b~oR zCGXE|xt*}c1tNa89DIkD3US!0*M%aG14(c(Ri;X=lGuXl{XEpvW@yj5Xyf+V9e~We z_(IIq?z5i{5?|b&*&z6Aj971+?|7;ud$A)h*3GM1Kxk`E!Ff=q>6tnvH_ zVS8gsoO|2bRT%wcAmn*)?z!oX1wKSIiP8XdqEEq<1Lbx+f|c0uR;+*-+3~(xn1OhE z95IW*mnp55&xBQZtYg7GZ$6vREQ~uhX1s~G(Xfrn{Tp3Y%Ivv_NypjNnEDQy}6d= zFrXFkWNXj11=9$OV2DxLlUE0MNnq{ng`EefXDn(@-FWU&6PxIr5+iy!1#0&=cQ*Mk zvDG+Mu8|v80GE78`1}5Iy|&ex%aH?hbdoF#++X!{3R4 zMJo}J0grNR#qXE56>ZC@#mF?W;4XccrYu?B|8qrVZIz%S*f}IbvGVm6>%QM>T#|Qh zKb53$m>qaQMa&v_#}aRG?QrV&?xUtw^bBp!zeO5Ok?-66F`T9@%dFEs1S8xmZl=@C(9u@}d3`faVkBryb`Z5}Wbak_FzpjGz zS;VJLdU;fO6jYV7n`roodj%uXg>buFmTbomM7j|!x**7mCZhK_IBA^SWY|RZu$t`@A#$EFyz~tv|l`A#>WWiVhW9 z@5JyT&r@(emGgo$>)9mN`&VFlUaJm>$X8xZ4(z!#(%;W?!5UnFH>GM^;)Ndd4Pw!u?TKy8Jh5i%8eCdjGFKOD3O&(r<xI)9yUc`J%sw^=Wq*zV_>6ZUMMWVJq19=*otDnU&96fb-#8 zXAv#2)uKd_qk3rR%~)nvaQ;+nBk|~iC^g{qrWifE+Vcfw0qCU86nl1Nkqk+ z0>y1uHh7&5Ply{ok5j{UN1s!fRke@4-m(>4=dpIkFz1AIXRCUSjWR5ABeP|c3C2>* z8nMdu_toa7N+SN1KIbLQ>y``+bW{!P*y2MHYb{{D5y&eJ8_q{8Kyjrum-Ea;l1Jj4 zzKs7`><0xUefadLA1haI-5We{-@x1dbsugsRS-L_>w)1166bu&eqp^k3tDjJpsSyv zo6F#^ab$DW-@vO^L7msmP#3jOG$0VK;_3ed;Nk4q*p)pRN9ixTJUp!4sqLiL+5d3(_^>|UPEj1)sH3O$^BqMS-{23*GrAYJZFcq;Os{Y#>*R$jUm22}6Fw8!)cfB4ZmM$m!#eidr!>MiMO3suj4 zsQTd!TI6Z8*X;dXDA6ZH@kaRPLU%V(JwEmsHL?|V4W-H7$d$;KwpM$@0+;SmH40Nv z8gCBUB3BQAH=1yjzxaA{fyN zg)3UHPPz%$mNunnznQ7YHB9sbok#@!vf}dM7^y}fH}J9M`18FbsM$7yj;%&ZB25pc z89w_?3(na=$~6zBYQEXog7(H(JoQz?-tW62_u8XI_Z+L}lrZi7Qy+^*Khi$O9A9Ua zIM$-2=6IdoU9nm9avWaoZsesHbeOq#RSnBKhW)iIbCGsn>+VKrkD}xhsm*W;%jKcc z&H$CLP=a=LP#>>I*|avICXN$&4I&E~0q;4RqBQd03y0*vu ztXtogjzoER2kozgZ{k1Tt`6l9J9Z_^#pQ_xR8}7F% zoApVDn)RaV>YNGi2D1Cz?n}J(pn6^PAP-ZAEf!qFd*n5Xw~`GbC)v z4L|s^rW*-z;AM|U-2Oj>TxU}gN*7KDRY96a6G5tU=^#RAQbjQI-UL)ax%3{I7Z8I; z3DQA{5PFRiAyJVoy+m4~G^I-~33+pW!8@~Gc6Mg>oU^Cw?40LO-7}C<+E8yC3<2#Q zeK_=1J+2NkBRJvenr%#+zPfx%-SslTL*O;Ix6HE1|7JRQYrWfY2^C0M-U{D;X zp-0JC9SCDqHOqO}vo8pJ`2+v2Y10iqsMzdPnx4qRE~Ms zAdt>^j0-Agzma>yfm^p2RK^V-NrMbuzKVPxOq*}^ zBMg3>E$7sGbs?TpnemkMkBgVtX=CBc9*9`;8KDAn!%(xl??B+jWg`Lf|G?+t?saS5gRBZMaCVCyAKW{aI{LIm+tqaT}JPFu{ah# z#=2qo#~kC?Jb?jf0rY3ef4KYYS4&U}(7qh`amZ8#jWC>ut| zbm@_=FK9M!im3+VX4cEM(Qc;K=PSjXxc!wbjYis?$n!^ey)Mi5nOC8!6T*|)-tGsv zSN#?xo*Zvb>Z7#-i^4y(i|k}ZYt8_wvsfD0FTJtVwsLWvDhs^xBtQ98$;poF{Hozp_5*Q@Rpe~t~Y zAqD+lJmm`Dj@UJu9ty5G`7MMMP3-QBU<7Gubw|i*;CCSmuQmiFu0$%F4 zN5;hEMwx)C5t5lv@w2OKZ30`%5~Zn}XE8U-xd&78db}tzAYl(QY0c|QXwQxeT{Xo{u)?T@^iE|rG+w&`_ePPk zfy4cwI`6Der^!Bby9@$T2&!onRtVKbR%-w}ndTP|n z2Tprt%VGPGeTPydyaQY?{m$&iClZfMt00l-&0m(jA}%ju6EF0yr?>wY?9)2%fcGR zfWf*)Vofuj&rZS2BDBxhjfoZ->aLjHAj}h9+nvLqemwi_=~h0phb!uK&Va4HX=^W{S8+oZ?$gu2hD%U zCU;Aouk!gDypQ)c*L-~$gbk})XVbCk_Kn+RYkoLval=DldK&l8=jQRwajY}{cfPhB zwdA+S$6hU$;AFlyb@7XY9r2HVUG2r@5HuT_>bLbEtc1tbIoU?I%N=;JAoMSQ$5w_G zg>qp%*;`8D0je?fdFdLoEC~%l2~yBZgcpqsp2byV>zhWwKE8w&5r3F91(wy})3d&0 zf-*EtO{t{DWK7fddwW1JClj)}ORKB|#+nM9s`ppt*^W1Dof9VTSxc*Iis^+W!K--e z1`7%m#zNMrX3d!ONgY^bak|wTnmDh#h#UeqUyFXRpJjL<8TQ^NRt;H6sa^fM@I8@t|P%GmxGLi#8ZWFqXUT5 z<)|BbHnR7T@b|>J)c7+v-AFl)+Fq{J8qfwL=E~sJW! ze{3P;QH^#x7LlZbAMTH(@!_t5=%b)fu*z3?$vS(fDpeIg3C*SBy^L#UwmpLh=voL! zdEoE`^?F@71>qCN11xg&C#bse&mE|1rg8F8zxGfV^>&5-TMRqSRvW=6(N=%5*|f6W zwsQWG#&Ibzg0XtQRP#j*w5gpPXnAImGHxrW`G7N1l-9ZrflE)kdI#_8*d)~x$yEF; zlUpOnw=a5rVjnv}-hSguUi@;U$Ej(!PgL-Epo>y@;ob>nO_?O0UJ%v2K5A$*rF}#4 zM$YqGI~-&(B%^BVEk?MkIBrCqwje1Kyei3)^E~Gwj`_Vyg4KJM%gpRZs(X{6IOw5T zc73$vVD=eqBdzP%XuDuY1!eUGMKK=^smTS*9hcjv74Z}IJRTd}D&ri?6XZ|851NA0 z9CtE(PSdIyG>uS{QDyNFbG@t>0OU#)Sy5vyxNhCCPpRa2X(B22sG`uw^TQ#Rmk=SX zNZxO0@OTM-{rB$3Fl2qv!p4IfjO7KjgL_qedXqFS4ZGA;5t&v*4!tLTJBlqvbOyfye7)tnvx~0y-rMH^OpDy;L=ai(A z5!xZ|rDsSqFM3@|jOX#aIBJRj5~lFD)%okE@&TOax{mnO3Jmo;XPL#UP5o6$vR2+R z+p_WbW85M`;h9n2ldgF@jj}uIXgCwg1(~^@c(Dek1W2IAAw{VDTzhjN0&M$5*V_)e z%cDj8sI8IpRyMTS_4(SN?`nADbay_>OqIN9wB${T$5H z%Qb3S&J~Av%;elrhzIxV%n*Md68^2X;dSr+D8+;ItvhrcS=|7Tdqua3;z%7%lQEq0 zLvuJ1cct6n$A$Qq*Vg^~fkTRGvaM4n4PZ5}1Xzp>gO7qrN;UfA-!sxE(=Gv`Sx$KZ zSZ72orB)^X#t$}Q!HC}I-U**@(u5BWPk71PcSC~3DQ}z^paj`rckO5hqjN7Ldd7os zW*Z%q5QKjPF94Si77`n|_gD!5?l3zd-F)`?>-hDp>mKo`*mGRZSF8BxC{h&uS8Xt- zN6z`H*g=J^%NU8f+$M#>wLkbzX&c-KYU^%8R+veU)ur)91uP!@qr<;pwo; z97J7_%DPYViAMhKd->8=8*|@@zA);vXEJe~YIa0hMm&FZ)2E`#Q7Z@2(mwSSi0UR3 zxN+<(=ruBZzOP3s%iUK7m(}-T<4Pm(VfIz;)6T0?B=->(QD2gLR-0Ja1OiNMd7DF48DmB;Ax_zJ9JWi%k>s8D$ zi3|M)Zs@-W>;sYC5l>pe@|X+LJ=wQ>RRiuE7s>E>JQ6>k-cg(*#M*<)U^Va}17u zh!=9slWRpOC4E(M0a!M|N5;B~Mys!lxl8Z4>qjjnGG|V zB^pMP0PuOFNT@&J$*1o4mmMVJvQ?Rwt1{ZMPVd`H*1+E7jbF7o5o`s7aK*cF=O6mx zX1+vbQt)LR9Z`c{Z8>97Z*pWch@_9UviQHGSWD0(XM-V}+TmPLnIv-UDHo!xqE6h} zzYhbilES~l8laXSU0y?cSoM_M4gOb+lwOfAt(^->6br_@`h4rvO=f~LF0JzA@|5>6AqzY4~G&MrXhe4uC*rd_vgnuF<=FNab~0cE~~GS;MZ#hIXH z+B_9j)m(?DPu^Sj!aakfNbTtJPfm=VBtDZ2Jyo{&-fToztjfyA{np1qp`sV%g`BKV zA%epmd9Jg7rM5~nec7ci&&bWVY(ktYZH>Q7nG_*)e&ZDJe>SkVJixh`fdjr>~qO#!8sb2Vo7w zVefZWaCW9DhuKBhnDnBXc;M@ax);7C3?~`55UOjJkleM-yA)Ipa6j7K(wIL$cJGFp z7-m9+e3h>c)t6BJ!QL?M!SkY96$~Xaz5+y3v=#Vgv3yAg75_3p_WO`KK}m*WrmTCg zc4Erdxb=Hw$F zth8|D3#~>8#7`iiX5Sh;x}4qVOvj)`B`N9mURXSFJe?xu$p9E^3kaCY+!5o8?^RWH)rJ< z&qKD*^?6#yqFSZQa{b@y3sPs6Y4m1nU!cMNq!|@hhGk^LhN;|+fMk}ZONlu-{(ZVO zSMXHfFDaTUoW&N=9%=-+s-Z)~Dk8cz*0_bbOTwPi+h-N+$pC+?F%iG~B;MIM) z?Lkb2#N-;O8kzlp-l4SD7I43?6oH;ICTcvBb?bT|Y&1|Kba`}m!o{MPt2xngSvTa= zCX7q$T8WekE>=kt+JSH^t2YoU~=|c#k2TC4kW>suF)HF5S#CX{md+1I#&CeXd_EU4QIo{E}CqljW#0GQ~0sH%{ET_ z%RMLf<1H(b%*bM(_JNAd)2N4dQ=rX?jN|jGoIvA)TnT}vCeh;XX85UZ6doYTJ|$=b znPsHkaHaq)Ph&fUC?^PiM+gc$^*F8?`e7Ii)SQfG&MJW-81Zozy06jU-~p)vN7j`JO(t& z@+DVTgun|BuN5bJpLozR=1j>2< + +

+Today Large Language Models (LLM) and other big Machine Learning (ML) models take the upfront of the stage. These models can now be trained for specific, customized solutions. But running these models, doing inference on a new dataset, still requires access to a big datacenter. +

+What if a company or an organization doesn't have access to a datacenter, or if the input data is too confidential? We propose to run the inference across existing computers in the office, like MacBooks and PCs. This eliminates the need for expensive hardware or cloud services, and keeps data secure. +

+ +

Our Solution

+ +

+We fully automate and optimize the distributed deployment of ML models for training and inference, dynamically leveraging available local machines (MacBooks, PCs, internal servers). Our high-performance, secure solution is ideal for companies seeking local ML usage with sovereignty and scalability. +

+The Unique Selling Points of our solution are: +

+
    +
  1. Simplicity – Clients can focus on their business applications while our solution transparently handles distributed deployment. +
  2. Efficiency – Clients can utilize existing machines, maximizing available computing power—even across heterogeneous hardware. +
  3. Scalability – Large models can be run locally; for example, 4 MacBooks are enough to run a 70B-parameter model +
  4. Privacy – Our solution enables organizations to leverage AI’s power locally without relying on untrusted providers. +