Curated by
More in Master AI & LLM Context: From Basics to Production
See all 15 →What Is a Context Window? — IBM Think
ibm.com
LLM Context Windows Explained: A Developer's Guide — Unstructured
unstructured.io
Context Windows — Official Claude / Anthropic Docs
docs.anthropic.com
Parallel Context Windows for LLMs — arXiv Seminal Paper
Parallel Context Windows (PCW) is a proposed method that extends the effective context window of off-the-shelf large language models without additional training by splitting long contexts into chunks, restricting attention within each window, and reusing positional embeddings. The paper demonstrates substantial improvements on in-context learning, multi-hop questions, and retrieval-augmented QA tasks across models ranging from 750 million to 178 billion parameters, and was presented at ACL 2023.
Built for AI agentsACO · 1105 tokens
Summary
Parallel Context Windows (PCW) is a proposed method that extends the effective context window of off-the-shelf large language models without additional training by splitting long contexts into chunks, restricting attention within each window, and reusing positional embeddings. The paper demonstrates substantial improvements on in-context learning, multi-hop questions, and retrieval-augmented QA tasks across models ranging from 750 million to 178 billion parameters, and was presented at ACL 2023.
Tags
parallel-context-windows · large-language-models · context-window · in-context-learning · attention-mechanism · positional-embeddings · natural-language-processing
Key entities
Parallel Context Windows (PCW) (concept, 0.99) · Nir Ratner (person, 0.95) · Yoav Levine (person, 0.95) · Yonatan Belinkov (person, 0.9) · Ori Ram (person, 0.9) · Amnon Shashua (person, 0.9) · Kevin Leyton-Brown (person, 0.9) · Yoav Shoham (person, 0.9) · Large Language Models (technology, 0.98) · ACL 2023 (event, 0.97) · arXiv (organization, 0.95) · in-context learning (concept, 0.92) · retrieval-augmented question answering (concept, 0.88) · attention mechanism (concept, 0.9)
Classification
reference · language en · status final
Provenance
claude-opus-4-6 via @stacklist/mcp-server@2.0.0, confidence 0.85, 2 Jul 2026