AI Paper Atlas

Research map / Large language models / Retrieval and knowledge

Panini: Continual Learning in Token Space via Structured Memory

Shreyas Rajesh, Pavan S Holur, Mehmet Yigit Turali, Chenda Duan, Vwani Roychowdhury

ICML 2026 · Large language models › Retrieval and knowledge · approach: RAG robustness and evaluation · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
query only the structured GSW QA-pair memory with chain retrieval instead of retrieving verbatim chunksPANINI · Six QA benchmarks averageF1: 53.3 -> 56.1Improved
feed short QA pairs from GSW to the answer model instead of document chunksPANINI · Six QA benchmarks averageanswer-context tokens: 705 -> 320Improved
use PANINI memory in place of chunk retrieval on PopQAPANINI · PopQAF1: 59.9 -> 57.6Hurt
use PANINI instead of dense retrieval on Platinum answerable questionsPANINI · Platinum average, answerableF1: 61.7 -> 79.9Improved
use PANINI instead of HippoRAG 2 on unanswerable questionsPANINI · Platinum average, unanswerablerefusal accuracy: 58.5 -> 72.8Improved
reduce beam width from 5 to 3 (ablation)PANINI RICR · six QA benchmarksaccuracy and tokens: about 25% fewer tokensNo clear effect
use single beam (B=1) in chain retrieval (ablation)PANINI RICR · six QA benchmarksaccuracyMixed
replace geometric-mean chain scoring with similarity-only, combined or greedy last-hop selection (ablation)PANINI RICR · six QA benchmarksaccuracyImproved
swap Search-R1 chunk retrieval for the GSW index without extra training (ablation)Search-R1 · six QA benchmarksperformance: improves on 4 of 6 datasetsMixed
replace proprietary components with open-source models (Qwen3 decomposer and answerer) (ablation)PANINI · six QA benchmarksF1Mixed
build GSW with smaller open-source models (Qwen3-8B/14B, GPT-OSS-120B) (ablation)PANINI · six QA benchmarksF1Mixed

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy