🌗 大型語言模型能否以 TLA+ 建立真實系統模型?
➤ 從「教科書式復述」到「精確建模」的挑戰
https://www.sigops.org/2026/can-llms-model-real-world-systems-in-tla/
本文探討了大型語言模型(LLM)在進行系統形式化驗證時的侷限性。研究團隊發現,LLM 雖然能流暢地編寫出符合語法的 TLA+ 規範,但往往只是在「背誦」教科書上的演算法原型,而非真正反映如 Etcd 或 ZooKeeper 等真實軟體的實作細節。為了區分兩者,團隊開發了 SysMoBench 基準測試框架,透過語法、運行、一致性及不變量檢查四個階段,精確定位模型與實際程式碼間的邏輯偏差,揭示了 LLM 在抽象化實作邏輯上的核心挑戰。
+ 這篇文章點出了 LLM 的一個盲點:它們很會模仿常見模式,但在面對需要深入理解特定實作細節的系統架構時,很容易產生幻覺或忽略關鍵的業務邏輯。SysMoBench 是一個非常務實的評估工具。
+ 對於分佈式系統工程師來說,這非常有價值。我們常說 TLA+
#形式化驗證 #大型語言模型 #系統程式設計 #TLA+
Can LLMs model real-world systems in TLA+? 

Editors’ note: AI has been actively pushing the frontier of applied formal methods for computing systems. In this article, the Specula team wrote about their experience of evaluating LLMs on modeling system code, the basic capability for agentic model checking, using TLA+, a specification language f

ACM SIGOPS

LLMs y TLA+: el 54% falla en modelar sistemas reales

¿Pueden los LLMs modelar sistemas TLA+ correctamente? Explorá este análisis profundo sobre especificación formal e inteligencia artificial en la práctica.

https://blog.donweb.com/llms-modelar-sistemas-tla-plus-conformidad-verificacion-form/

#tla #verificaciónformal #llms #sistemasdistribuidos #especificacionesformales

LLMs modelar sistemas TLA+: solo 46% conformidad

¿Pueden los LLMs modelar sistemas TLA+ correctamente? Explorá este análisis profundo sobre especificación formal e inteligencia artificial en la práctica.

Blog Donweb
Can #LLMs model real-world systems in TLA+? 🤔 Oh sure, because who wouldn't want a linguistically confused #AI applying #logic it doesn't truly comprehend to complex systems modeling? 🙄 Next up: teaching your cat to do your taxes! 🐱💼
https://www.sigops.org/2026/can-llms-model-real-world-systems-in-tla/ #TLA+ #complexsystems #catstaxes #HackerNews #ngated
Can LLMs model real-world systems in TLA+? 

Editors’ note: AI has been actively pushing the frontier of applied formal methods for computing systems. In this article, the Specula team wrote about their experience of evaluating LLMs on modeling system code, the basic capability for agentic model checking, using TLA+, a specification language f

ACM SIGOPS
Can LLMs model real-world systems in TLA+? 

Editors’ note: AI has been actively pushing the frontier of applied formal methods for computing systems. In this article, the Specula team wrote about their experience of evaluating LLMs on modeling system code, the basic capability for agentic model checking, using TLA+, a specification language f

ACM SIGOPS

Rijndael (@rot13maxi)

TLA+의 작동 원리를 설명하는 트윗입니다. 전체 시스템을 상태 기계로 모델링하고, 변수와 상태를 업데이트하는 액션으로 시스템 동작을 표현한다는 점을 강조합니다. 다음에 취할 수 있는 액션은 현재 상태에 따라 달라지며, TLA+의 흥미로운 기능 중 하나로 액션이 (트윗에서 이어지지 않지만) 특정 선택이나 결정을 할 수 있음을 언급합니다. 형식적 사양 및 검증 관점에서 유용한 설명입니다.

https://x.com/rot13maxi/status/2032794746918252677

#tla+ #formalmethods #statemachine #specification

Rijndael (@rot13maxi) on X

@KingBootoshi The way TLA+ works is you write up your whole system as a state machine. There are basically variables and then actions that update those variables. The actions that can be taken next depend on the current state. One of the (many) cool things about tla+ is an action can pick a

X (formerly Twitter)

BOOTOSHI (@KingBootoshi)

작성자가 TLA+를 접하고 에이전틱(agentic) 코딩에 매우 유용하다고 감탄하고 있습니다. TLA+로 설계 상태(state)의 모든 가능한 시나리오를 수학적으로 검증해 버그와 충돌을 예방할 수 있으며, 문제가 발견되면 에이전트가 즉시 피드백을 받아 반복적으로 수정한다고 설명합니다.

https://x.com/KingBootoshi/status/2032775953479520306

#tla+ #formalverification #agentic #verification

BOOTOSHI 👑 (@KingBootoshi) on X

HOLY FUK I JUST LEARNED ABOUT TLA+ AND IT'S SO GOOD FOR AGENTIC CODING ur telling ME that i can mathematically fact check every possible scenario of my design STATE to prevent bugs and crashes AND IF IT FINDS SOMETHING THE AGENTS GET INSTANT FEEDBACK AND LOOP FIXING IT TILL IT

X (formerly Twitter)

Применяем TLA+ на практике

Привет, Хабр! Меня зовут Сергей, я работают в компании InfoWatch разработчиком на продукте ARMA Стена (NGFW). Подробнее о том, что такое ARMA Стена, можно прочитать тут . В этой статье я хочу поделиться опытом применения метода формальной верификации в решении практической бизнес-задачи. Сразу оговорюсь, что в статье используется TLA+, без введения в инструмент, чтобы не увеличивать объём статьи. Подробнее про инструмент вы можете почитать на сайте создателя , тут и тут . Необходимые объяснения даются по ходу изложения. Статья состоит из двух частей: 1) Что такое формальная верификация и где она применятся 2) Решение бизнес-задачи в NGFW Верифицировать статью

https://habr.com/ru/companies/infowatch/articles/1003382/

#формальная_верификация #tla+ #python #ngfw #ARMA

Применяем TLA+ на практике

Привет, Хабр! Меня зовут Сергей, я работают в компании InfoWatch разработчиком на продукте ARMA Стена (NGFW). Подробнее о том, что такое ARMA Стена, можно прочитать  тут . В этой статье я хочу...

Хабр
Pathological Digital Affection Personal Display Avoidance Public Demand Assistant #PDA #TLA

Learn TLA+ by example, from the Transcendental Hillel Wayne (which is how I’d imagine his Marvel series would be titled).

#tla #formalMethods #math

https://tlabyexample.com/

TLA+ By Example

Learn TLA+ specifications through interactive examples in your browser