跳到正文
HuggingFace Daily Papers(社區熱門論文)·· 24 天前AI 評分50

WMIP:受工作記憶啟發的大語言模型累積狀態追蹤探針

WMIP: A Working-Memory-Inspired Programmable Probe of Cumulative State Tracking in Large Language Models

AI 導讀

WMIP 是一套以程式計算答案、用於測試大語言模型累積狀態追蹤能力的探針,涵蓋 3 至 100 次更新及不同查詢和輸出協議。18 個模型在單步算術的準確率為 84.9%,但在需處理三至七次更新的核心項目上只有 31.7%;更新達 50 次後,沒有模型的準確率超過 10%。加入書寫步驟並要求答案置於末行後,核心準確率升至 88.0%,研究亦公開生成器、測試項目、評分程式碼及模型回應。

正文

Published on Oct 30

Authors:

,

,

,

,

Abstract

Large language models struggle with cumulative state tracking across multiple updates despite strong single-step arithmetic, and their measured accuracy depends heavily on output protocols.

Human working memory lets people maintain and update information. Large language models (LLMs) face a related demand when tracking updates within a single prompt without writing out intermediate states, yet existing tests differ in their constraints on written steps. We introduce a working-memory-inspired probe of cumulative state tracking with program-computed answers, 3 to 100 updates, and final-state, intermediate-state, or multi-target queries. It compares protocols on identical items. Across 18 LLMs, accuracy is 84.9% on single-step arithmetic but 31.7% on core items requiring one number after three to seven updates. From 50 updates on, no model exceeds 10% accuracy. Core accuracy reaches 88.0% under a larger budget and instructions to write steps and place the answer on the last line, mainly due to the instruction changes rather than the budget. Core accuracy also correlates positively with public Berkeley Function Calling Leaderboard scores. These findings show that strong single-step arithmetic does not ensure reliable cumulative tracking and that output protocols substantially change measured accuracy. We release the generator, items, scoring code, and responses for reproducible comparisons across depths, query types, and protocols.

View arXiv page View PDF Project page GitHub 1 Add to collection

Get this paper in your agent:

hf papers read 2603.27343

Don't have the latest CLI?

curl -LsSf https://hf.co/cli/install.sh | bash

Models citing this paper 0

No model linking this paper

Cite arxiv.org/abs/2603.27343 in a model README.md to link it from this page.

Datasets citing this paper 1

dengzhe-hou/WMIP

Viewer •

Updated 9 days ago

•

1.5k

•

200

Spaces citing this paper 0

No Space linking this paper

Cite arxiv.org/abs/2603.27343 in a Space README.md to link it from this page.

Collections including this paper 0

No Collection including this paper

Add this paper to a collection to link it from this page.

來源:HuggingFace Daily Papers(社區熱門論文) · huggingface.co