NishiSearch

Web This site Trusted

30 result(s) for “okapi bm25 site www eyespy cy site www abclinuxu cy site dlmultimedia epa in site:arxiv.org” of 33 matched · 443 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: newest result fetched 24 d ago · median age of these results 24 d · 1 of 30 carry a fetch time

CY) We show that knowledge distillation (KD) in small instruction-tuned language models has asymmetric effects on bias, and that measuring them correctly requires accounting for where refusal mass moves and what the parser can legitimately score
from the top result below

Computation and Language

https://arxiv.org/list/cs.CL/new
LG) Half a million cuneiform clay tablets survive in museums worldwide, yet modern users can neither read nor write in the world's oldest writing system, leaving a 4,000-year cultural barrier that existing NLP tools have only partially addressed

Artificial Intelligence

https://arxiv.org/list/cs.AI/new
In practice, many field descriptions remain ambiguous or incomplete, as much of the essential context (e

%PDF-1.7 % 1 0 obj > stream x[[s۶~?tbwpә؎/Iܸ:Ai[

https://arxiv.org/pdf/2608.03358
tbwpә؎/Iܸ:Ai[,&(%;nf,(^Xp)"[" (EPA(ܺR(%Jhʠ j%ß B-PeRheneᴨ&ve(+2ƍ@kP3&KQO"-G oՔ22

%PDF-1.5 % 7 0 obj > stream xP(endstream endobj 9 0 obj >

https://arxiv.org/pdf/2407.00215
stream xP(endstream endobj 11 0 obj > stream xP(endstream endobj 17 0 obj > stream xP(endstream endobj 20 0 obj > stream xP(endstream endobj 23 0 obj > stream xP(endstream endobj

%PDF-1.7 % 1 0 obj > stream xZr6}߯ࣵXJV7dɊ$n做%

https://arxiv.org/pdf/2608.00426
gHÑ|v A**Gn>hఄ$OJ:#$]rd KeѵL!"p 8 H&+) ,)KAY++\Jt 0D `IwGHm4=pW&Z D>60!DJ W!HB K~OR"U y@4(RRRTm@O'h2C#AhC !TCp=5

%PDF-1.7 % 1 0 obj > stream x\rFr}WQ`IihʫRZ

https://arxiv.org/pdf/2608.03411
{~l F )=PH# ӑHton> stream DUD: Decoupled Update Dynamics for Reliable Uncertainty Quantification in Large Language ModelsYixin BuRunze XiaGuanyun ZouYupeng JiHaodong LiuPiji Lihttp://creativecommons

%PDF-1.7 % 1 0 obj > stream xiSѮG**&ر dUd

https://arxiv.org/pdf/2607.26327v3
zcYF-k&XKM^jqXJ^Z0klvTJ>K6hz2*免Pu_w/'OuUvqpaR

%PDF-1.7 % 1 0 obj > stream xƖ=3DΘ>gZ0gЉ;P

https://arxiv.org/pdf/2608.03421
~k oLP=H"wBd|vob7ct3`׿CfkvrY'G7svU6{>X*_\r^\|X\-sD6?{^~C /OOFmQjNGU-`=-n0w t+aÛu{7'ux;9n E^$/hu^pm

%PDF-1.7 % 1 0 obj > endobj 2 0 obj > endobj 3 0 obj > stream Cache

https://arxiv.org/pdf/2510.03215
7 % 1 0 obj > endobj 2 0 obj > endobj 3 0 obj > stream Cache-to-Cache: Direct Semantic Communication Between Large Language ModelsTianyu FuZihan MinHanling ZhangJichao YanGuohao DaiWanli OuyangYu Wanghttp://arxiv

%PDF-1.7 % 1 0 obj > stream xN0E|,#oʎ+MF

https://arxiv.org/pdf/2608.03833
SJ%V{ǖ!!8D e1"B Q@1oV+` rf^ I !lkmǩf0KXbp$Q #t3SBF'=o2ehvUzk]p+|)g{&61yny!sykZS?==ZU*\>zhj쁶gO-͔D`+m\

1 Introduction

https://arxiv.org/html/2501.13956v1
com Abstract We introduce Zep, a novel memory layer service for AI agents that outperforms the current state-of-the-art system, MemGPT, in the Deep Memory Retrieval (DMR) benchmark

Report GitHub Issue

https://arxiv.org/html/2608.02613v1
7 Extrapolation across readers for cache-replay backends TTFT rows in Table 3

Report GitHub Issue

https://arxiv.org/html/2608.03048v1
To address these issues, we propose PI-Mem (Parallel-Iterative Memory), a mechanism that processes all chunks in parallel and iteratively refines a shared memory over a bounded number of turns

Computers and Society Aug 2026

https://arxiv.org/list/cs.CY/current
251-256 Showing up to 50 entries per page: fewer | more | all [1] arXiv:2608

Pt^| o?px8:Fߓ)IĊ,2wStN}

https://arxiv.org/src/2006.02085
@z/|h LT;QU؃HqK)S½$iNjMUbiI

%PDF-1.5 % 91 0 obj > stream xڭ[YƑ~ׯG#) 6h$Hv4P݄h

https://arxiv.org/pdf/2006.01402
G_΄_a~29 0LO ӁIN2

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

https://arxiv.org/pdf/2608.01785
s],+Y BVC:⳰wWbdR)Ȇ(P:"e$3Ūlvr;s5- (hAhmy;a崻G0$HO2FѿIaT$s0v6sn7_&oU82kELiKBgm DIn

Zzc\msFguT$#oYo'sξ홮x Yb3

https://arxiv.org/src/2211.10435v1
HIķR&8ק_e]&K{n>{'~O޹s;O&'>Q>$Q.tzӺ}W;8Y]+s ߿߽~OǓO#qSoʺ|L C3fzf-|ȶ jZ|17vDٝ59@"0l3b0GYs 3 Wt ->

!YfZ TS瞏J}jkGn@@jE $-)h)VRE@iL $

https://arxiv.org/src/2405.20232
~ 'O IWpf)YP@lغr3,^";odVMq%@Е,_9ڍ#kN4}IN~tz{gr~]uԤgO5{h=]'@Ua{/ų@R lGf(ȴjxjvI6߸5p6oG͛ #%lu>G1):}&x J3R0m˾ߝ0VD^G_`

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;

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

2Cd[ioȖɯ(4tc,vł%ٽWse˜*AKN I S

https://arxiv.org/src/2212.07841
-/ .CdN?ZOw ?~R",MtE-gW0؃ѐw# 1E{p*"r)g"iߌ8QcH$gC]R>BMW)v gdaygcB+XLA(r^Ha?(`~X^ Gh#YH_ү[u6h XUx/``

%PDF-1.5 % 212 0 obj > stream xڝZKw۶Wh)c1nVI7mvvAI>T>

https://arxiv.org/pdf/2211.01910
~ -E7&8|wշWbo7O7I{q6wۏ*squo^8}ͱK~):Y1~uww7(\(9Q{io&^fY j7>,DI%*bEY ͔NjM~~w6B~qh}O&2Oc~Xoơ

ߍ^[kƕgNT%9rW¬l>.v47*T0#Y+>Di:1@[vU

https://arxiv.org/src/2006.02155
wtԌN-Q0/]}ղ\_e8 " " ۻ葨^GQSMjMcQlL"kZC6Yqd;Z|GcYz੉uNy˘,-lZ6@cD؟fEi,8\gd-F㸭 FH##

k1a}kwh^ \&>璴޶L6ÖlIJI L?$S6">LDY2et"ڌG

https://arxiv.org/src/2104.08691
>6JfHx@OFm1Id4L$3~8q=vtz3D w9Hpր `uy[z[(VD8\I,R/(~ǺzYSΫӑQ #SgKI!p 8 , #X6Lp;8@$"ZLU67^B+h!CT[-8y# E;ܟdA!\O}ܟ(

^񰏽 NkzJrYY^hBB۷'ucRʲ

https://arxiv.org/src/2006.01331
>'"J=&_[:$@0՛^Ajj> [AQt`}~Q?z(Rx{>2J`һA;2?;_ڱ퀂.OVcбn{¾c){ƞ_ oKf?7o-{.؏'3c&͘\ٜ-ؒy7v| ,f KYn-[5 Cdj"F?&8C%(E|C$

drj[Fo=EqlDuWq'@H3FD$)KAy}}9UHj

https://arxiv.org/src/2608.04003
$Y 0"~sͫ/޽''ϧtɒxߺ:J'yRfɟe׆n>)42&zm^VVlu .8vK0\If'Gz4Yd&҅ $×U[-Ǯqo{$*N /+$\5l]5Fa͉2w|: *Uڹ2t*~ =0oW-BH-VI*0+]F~Q'{

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

https://arxiv.org/abs/2608.03883
All submitted systems follow a staged retrieval design in which BM25 retrieves candidates at scale and semantic or LLM rerankers refine the final rankings

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

https://arxiv.org/abs/2104.08663
Benchmark for Zero-shot Evaluation of Information Retrieval Models, by Nandan Thakur and 4 other authors View PDF HTML (experimental) Abstract:Existing neural information retrieval (IR) models have often been studied in homogeneous and narrow settings, which has considerably limited insights into their out-of

[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