NishiSearch

Web This site Trusted

30 result(s) for “okapi bm25 site hotchips pp site www uscourts gov site www news nil gov site:arxiv.org” of 61 matched · 877 ms

Scope: Nishi’s own index (our crawl + Common Crawl) — not the whole live web. Some sites block all automated crawlers (anti-bot walls), so their pages may be absent here even though we never filter lawful content.

Freshness: no result on this page carries a fetch time yet (indexed before the fetch epoch was stored beside the doc)

The query asks the model to recall the topic of an earlier conversation about Tesla; the gold answer is anchored in one specific session about Pooja’s dig site
from the top result below

Report GitHub Issue

https://arxiv.org/html/2608.02613v1
We evaluate five open-weight readers with Vanilla context, BM25-RAG, Oracle retrieval, Memobase, and MemSearch as memory backends

%PDF-1.5 % 152 0 obj > stream xmTˎ0+$$0  a#A%

https://arxiv.org/pdf/2402.03300
߯jD岻fc;Z̫MfG} q]/ޭmޯo⣩0Z^x]fkn{E+{*ʧypg6;5PVpH8$hmڢ*߄zR:")󨺠3qXysO'H)-"}[˺s 3 4{pYdrK+ a }ѫW{ Fvm7344AGc

%PDF-1.5 % 181 0 obj > stream xZKs8Wp dfbOav*[p!

https://arxiv.org/pdf/2005.14165
{괙֭ƭ[ux~t Y;i;w~~\^^puWWW^__ggg{/_;-5S;wzrrAR

1 Introduction

https://arxiv.org/html/2501.13956v1
2 Knowledge Graph Construction 2.1 Episodes 2.2 Semantic Entities and Facts 2.2.1 Entities 2.2.2 Facts 2.2.3 Temporal Extraction and Edge Invalidation 2.3 Communities 3 Memory Retrieval 3.1 Search 3.2 Reranker 4 Experiments 4.1 Choice

Report GitHub Issue

https://arxiv.org/html/2608.03048v1
× Title: Content selection saved. Describe the issue below: Description: Submit without GitHub Submit in GitHub arXiv is now an independent nonprofit! Learn more × Back to arXiv Why HTML? Report Issue Back to Abstract Download PDF

%PDF-1.5 % 2 0 obj > stream xmTˎ0+$$0  a#A%

https://arxiv.org/pdf/2407.02402
߯jD岻fc;Z̫MfG} q]/ޭmޯo⣩0Z^x]fkn{E+{*ʧypg6;5PVpH8$hmڢ*߄zR:")󨺠3qXysO'H)-"}[˺s 3 4{pYdrK+ a }ѫW{ Fvm7344AGc

%PDF-1.5 % 152 0 obj > stream xڭ[Yܸ~ׯЃ:Fh8^X?

https://arxiv.org/pdf/2102.13654
48#vBB҈ mt={v ֳĉ{Wkv_rORu>YS\+V,LqO8--gϞMqJLLTwv͛7O߷o_pp8Mҥ >uּSN +V,]kVre ; y)񄿀pW[Y'0A=zxX,={VfE$0Q[JH*4Z|do߾

w_z߯x;ã'x|p ãGsJĊ(2w[

https://arxiv.org/src/2006.02230
2+45(@;KQ&5HS+DD~:Q7X0]""n@h|P_a&%Bqr^݊'|V}W4 nGo>h(F?)SxCƗz^G_A +|CHurX$=ҜnR ߵ-TSZ^a } /_SS`]2f|m0Ui@S75Ј4

rj;srY9靟OO߰ޟ4'}Ʊڅ'o1d{l

https://arxiv.org/src/2608.03883
lއxG&JWXԴ&[v @V[uԇF7vz%N`pP \>$kb {K/05 U0[|ʋXcc S-bQEAE

^ |ɒKl  v?q|rr0

https://arxiv.org/src/2006.01354
TdIjǖI,D ?=x`],Ă3֯v;!픻֭.t-W8I{參&-{s/}+Yp"X&v*i; b"^ |aǶ >!ZQ}N?}=c߳? b? {.؏'3'ѱ^[ͦa.ll X{,c7-ي}

%PDF-1.5 % 1 0 obj > stream xY]o6}_fuo ( ,omE}ZADږ

https://arxiv.org/pdf/2608.02851v1
%6G"8/1tx493C`LԒ)c9S18YecܨՌ;4LCς eQ0F3pLVVһL -ѳc8zt4h(SdSLՂ5\ҳaZ*LkC|r*1)3I83Ҥ) hL1F2aYebt,qU1+(nřG.|-;tϾvcynЏ

%PDF-1.7 % 1 0 obj > stream xVM6W.`[CQIZT

https://arxiv.org/pdf/2608.03233
T+jq8 M\c~wD`%::P sC` zzTrH`${0F{kʆƄ endstream endobj 411 0 obj > stream On the Diversity of Analogy Making in Large Language ModelsYuanhao ShenDaniel Xavier de SousaCaio César Sifuentes BarcelosHongyu GuoXiaodan Zhuhttp://arxiv

8kd[[S ՙ9 ;n_2e !!/n٭

https://arxiv.org/src/2211.01910v2
-wBS3Cb---뷖Dr5GO ?]󉟵^}v[~j9N},?ipOZ[HLdL|1@3ZV Cy')տ8[F:0/10IQi|}/OnP0 E EGZi ž0>ދ 6Ў[h}RXc25

rj[rFϿ#!Y|X=J9W+Cr,@0I\S$8

https://arxiv.org/src/2608.03882
I[[٭Z#35==5Hү~Oo9o|_>?8}/i|JmB,3?g!,";P\>3MӥG&;&y+NכDBLBFw?v-\FB,%]; DJ{jt"l,\%T'&9 $X7o}~ BҎnoOyMq

J^P_.I/"]Mzi"H =[@*H.Z!t{

https://arxiv.org/src/2006.01402
{s|3|3'C M]z'b=~gGk??巐={8BbbbWpqXߔ+++wڏUH艀$jJ:88035+d88ڦ(GlcQůT$de܊ѓNjM_5##,NG)}Rj6T+kC_E}[

m_\ysFϿO[d턄RdTSxW3RMIAfϾqrnI

https://arxiv.org/src/2006.02464
>>!ŧ/h4>!BO28w/ #,l4I~˥ ]Jo ?t&;Iԋiz3f6-O4}z,(L$MD4~\4J7K%(@WubL,4| bG bU\j)cL W^ SKB;

%PDF-1.5 % 1 0 obj > endobj 2 0 obj > endobj 3 0 obj > stream

https://arxiv.org/pdf/2508.16279
AgentScope 1.0: A Developer-Centric Framework for Building Agentic ApplicationsDawei GaoZitao LiYuexiang XieWeirui KuangLiuyi YaoBingchen QianZhijian MaYue CuiHaohao LuoShen LiLu YiYi YuShiqi HeZhiling LuoWenmeng ZhouZhicheng ZhangXuguang

z;c:XSW._U20eKrI$!B\UkQPk+V

https://arxiv.org/src/2205.11916v3
"uՍJ-HP=7!yz$$g9Or%$)VÃכ{{}^>|+߻@`4 y fESܻ>BJ*S9 ЎeljĄ%hR)1bT 9q%J% *T?My !䄘E[a@6U{6i*8.űɤi _ƷLsm

%PDF-1.7 % 1 0 obj > stream x[is۸~?Nޔ bRuV%v8N

https://arxiv.org/pdf/2608.02615
mNJ3mDbwF=Ml;x,qq,ޣ^}ZXy:wWW|

%PDF-1.5 % 106 0 obj > stream xڍ[K WU* G󡽍a

https://arxiv.org/pdf/2407.02510
JXH'>4EQM/j4~z}7O_}jv. ;i(Sq9C>Tmm ~.3Tеx(=)g;>7K}&4秿Ӑʼn|C[ռ? te^W+ $@GP&lmQvZz8%e^cŔ~O0N7uyZn Vz) "M!

LG.T樣ψ4x+j G~dt>`7GL3վˠR9 vS,Mދq

https://arxiv.org/src/2303.12712
#Z-! MCj8 Õ /EtNky>ΉA4Yr'KR.qyKy92fFʕJ.h ݬ VwNc}]X;qf#aH腚,9ޭdұ1Ǝ:%Ztu!*Z:۹;"F. ::Tx.F666{'OUMU' 3`$Q

%PDF-1.5 % 185 0 obj > endobj 186 0 obj > /W [ 1 2 1 ] /Index [ 185

https://arxiv.org/pdf/2508.04671
485 ] /Info 411 0 R /Root 187 0 R /Size 670 /Prev 2921183 /ID [] >> stream xcbdg`b`8 $;A. H0 >X`$$,@ P)$A@{A ayY@* $xM@ D(201* (1J cr]0 -Qb%0T endstream endobj 187 0 obj >

Computation and Language

https://arxiv.org/list/cs.CL/new
We evaluate five open-weight readers with Vanilla context, BM25-RAG, Oracle retrieval, Memobase, and MemSearch as memory backends

ꥷh\sF+RY$%]:)V*%h7 0$1JbP9}N}>

https://arxiv.org/src/2509.02523
:RՋ,5x)mpPɖyx4\xbuT+σyzEc~mW]̴Yw&\uZ2ꧺ=@Mq:U:sRΎIsrյxRtT_AU>68]US16KVj$@)h!*_$:ЇتzEGC;MLT=Z

%PDF-1.7 % 1 0 obj > stream xYێ6}߯cC; gf憽HKߪ

https://arxiv.org/pdf/2608.00785
*"4X]滉` Yd)$w(P\w󲒴e$S IGRǼ%:VͶg{*Ɗъ$vbyxZ1phN"l̀ >=9͟^11>"mS;wn *j[Gx.ڛ7pi

%PDF-1.7 % 1 0 obj > stream xYYܸ~ϯ`fěv,l^Bfw

https://arxiv.org/pdf/2608.01785
B%pp7w}^t3>/nN y@IvonG

Decoupling Generation and Selection for Budget-Constrained Faithful

https://arxiv.org/html/2608.03655v1
org perpetual non-exclusive license arXiv:2608

[2608.03883] DS@GT-ARC at eRisk 2026 Task 3: Sparse, Semantic, and LLM

https://arxiv.org/abs/2608.03883
Our systems combine sparse BM25 retrieval, evidence-aware rescoring for self-referential symptom reports, embedding-based reranking, query-prototype expansion, and LLM-based reranking

[2104.08663] BEIR: A Heterogenous Benchmark for Zero-shot Evaluation of

https://arxiv.org/abs/2104.08663
Our results show BM25 is a robust baseline and re-ranking and late-interaction-based models on average achieve the best zero-shot performances, however, at high computational costs

[2608.03860] SciRet: A Compute-Aware Empirical Study of Retrieval and

https://arxiv.org/abs/2608.03860
The pipeline combines sentence-window chunking, BM25, BGE-M3 dense retrieval, reciprocal rank fusion, optional cross-encoder reranking, and grounded answer generation