Paper Introduction - Category - DataSci Ocean

Paper Introduction

2026

Skill Files Bloat as Agents Self-Evolve: Inside SkillZip's Fix 10-21

Do Agents Need a Map to Search Documents? Inside CorpusMap 10-17

Can You Trust a PDF Parsing Leaderboard? Inside ParseBench 10-15

Can a Search Index Repair Itself? A Skeptical Look at SELF-INDEX 10-10

Can an LLM Write Its Own Playbook From a Manual? Ctx2Skill 10-07

Cheap LLM Judge First, GPT-6 Only When Unsure: Does JEV Save Money? 10-06

Do Self-Improving Agents Just Memorize the Test? Inside Google's RRSI 10-04

Resource2Skill: Video Alone Beats Code, Articles, Artifacts Combined 09-30

Naive Prompt Optimization: What's Left Once You Remove the Search 09-27

Dream-RSI: Replay History as a Free Simulator, Not a World Model 09-21

More Examples Isn't Better: Rethinking Many-Shot CoT-ICL 09-16

RecursiveMAS: Multi-Agent Systems That Think in Vectors 09-10

WikiSkill: A Skill Memory That Never Gets Rolled Back 09-02

LLM-as-a-Verifier: A Better Alternative to LLM-as-a-Judge 08-26

RAGU: The Paper That Accidentally Measured GraphRAG's Ceiling 08-22

SAG: Multi-Hop RAG via Query-Time SQL, Not Offline Graphs 08-19

SketchVLM: Letting a VLM Draw Its Own Reasoning 08-08

Persona-Aware D2S: One Paper, Four Audience-Tailored Slide Decks 07-29

Cerebras's Enterprise RAG: Narrow-Waist Design, Decoupled Retrieval 07-23

SkillOpt: Treating an Agent's Skill File as a Trainable Weight 07-22

1
2
3
4