mlsanityData to 7 Oct 2026

Entity Tracking in Language Models

arXiv 2305.02363 ↗First posted 3 May 2023cs.CLcited by 44

Not on the map. Why?

Abstract (arXiv)

Keeping track of how states of entities change as a text or dialog unfolds is a key prerequisite to discourse understanding. Yet, there have been few systematic investigations into the ability of large language models (LLMs) to track discourse entities. In this work, we present a task probing to what extent a language model can infer the final state of an entity given an English description of the initial state and a series of state-changing operations. We use this task to first investigate whether Flan-T5, GPT-3 and GPT-3.5 can track the state of entities, and find that only GPT-3.5 models, which have been pretrained on large amounts of code, exhibit this ability. We then investigate whether smaller models pretrained primarily on text can learn to track entities, through finetuning T5 on several training/evaluation splits. While performance degrades for more complex splits, we find that even when evaluated on a different set of entities from training or longer operation sequences, a finetuned model can perform non-trivial entity tracking. Taken together, these results suggest that language models can learn to track entities but pretraining on text corpora alone does not make this capacity surface.

Key facts. Cited by 44 papers in the map (data through 7 Oct 2026). Builds on 35 of its 71 references that are in the map. First posted on arXiv 3 May 2023.

Builds on

35 of 71 references.

References · 35

  1. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,204
  2. 2020Language Models are Few-Shot Learnerscited by 32,199
  3. 2023GPT-4 Technical Reportcited by 18,132
  4. 2022Training language models to follow instructions with human feedbackcited by 15,712
  5. 2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,491
  6. 2019BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehensioncited by 5,874
  7. 2019BERT Rediscovers the Classical NLP Pipelinecited by 1,046
  8. 2019What Does BERT Look At? An Analysis of BERT's Attentioncited by 994
  9. 2022Emergent World Representations: Exploring a Sequence Model Trained on a Synthetic Taskcited by 417
  10. 2015Towards AI-Complete Question Answering: A Set of Prerequisite Toy Taskscited by 309
  11. 2018TextWorld: A Learning Environment for Text-based Gamescited by 304
  12. 2017End-to-end Neural Coreference Resolutioncited by 240
23 more
  1. 2022Neural Theory-of-Mind? On the Limits of Social Intelligence in Large LMscited by 220
  2. 2021Implicit Representations of Meaning in Neural Language Modelscited by 179
  3. 2022Language Models of Code are Few-Shot Commonsense Learnerscited by 160
  4. 2020How Can We Accelerate Progress Towards Human-like Linguistic Generalization?cited by 114
  5. 2015The Goldilocks Principle: Reading Children's Books with Explicit Memory Representationscited by 108
  6. 2019Coreference Resolution as Query-based Span Predictioncited by 91
  7. 2021Chess as a Testbed for Language Model State Trackingcited by 69
  8. 2019An Annotated Dataset of Coreference in English Literaturecited by 57
  9. 2021Provable Limitations of Acquiring Meaning from Ungrounded Form: What Will Future Language Models Understand?cited by 55
  10. 2017Simulating Action Dynamics with Neural Process Networkscited by 53
  11. 2016Tracking the World State with Recurrent Entity Networkscited by 51
  12. 2022Can language models handle recursively nested grammatical structures? A case study on comparing models and humanscited by 41
  13. 2016Simpler Context-Dependent Logical Forms via Model Projectionscited by 33
  14. 2019Everything Happens for a Reason: Discovering the Purpose of Actions in Procedural Textcited by 31
  15. 2018PreCo: A Large-scale Dataset in Preschool Vocabulary for Coreference Resolutioncited by 30
  16. 2023Can Large Language Models Play Text Games Well? Current State-of-the-Art and Open Questionscited by 30
  17. 2020Learning to Ignore: Long Document Coreference with Bounded Memory Neural Networkscited by 29
  18. 2022Unobserved Local Structures Make Compositional Generalization Hardcited by 29
  19. 2022When a sentence does not introduce a discourse entity, Transformer-based models still sometimes refer to itcited by 22
  20. 2017Dynamic Entity Representations in Neural Language Modelscited by 19
  21. 2019Tracking Discrete and Continuous Entity State for Process Understandingcited by 15
  22. 2019Effective Use of Transformer Networks for Entity Trackingcited by 11
  23. 2019What do Entity-Centric Models Learn? Insights from Entity Linking in Multi-Party Dialoguecited by 4

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 4; 2024: 6; 2025: 17; 2026: 17.

44 papers.

Sort:

By area:

  1. 2024The Illusion of State in State-Space Modelscited by 181
  2. 2024Fine-Tuning Enhances Existing Mechanisms: A Case Study on Entity Trackingcited by 140
  3. 2023What Formal Languages Can Transformers Express? A Surveycited by 120
  4. 2023How do Language Models Bind Entities in Context?cited by 102
  5. 2023BoardgameQA: A Dataset for Natural Language Reasoning with Contradictory Informationcited by 42
  6. 2024Monitoring Latent World States in Language Models with Propositional Probescited by 39
  7. 2025Language Models use Lookbacks to Track Beliefscited by 36
  8. 2025PaTH Attention: Position Encoding via Accumulating Householder Transformationscited by 32
  9. 2024Code Pretraining Improves Entity Tracking Abilities of Language Modelscited by 31
  10. 2025(How) Do Language Models Track State?cited by 31
  11. 2025RefactorBench: Evaluating Stateful Reasoning in Language Agents Through Codecited by 28
  12. 2025Transformers Provably Learn Chain-of-Thought Reasoning with Length Generalizationcited by 12
32 more
  1. 2024Are Language Models More Like Libraries or Like Librarians? Bibliotechnism, the Novel Reference Problem, and the Attitudes of LLMscited by 11
  2. 2024Understanding Synthetic Context Extension via Retrieval Headscited by 10
  3. 2025Positional Biases Shift as Inputs Approach Context Window Limitscited by 9
  4. 2025Mechanistic evaluation of Transformers and state space modelscited by 8
  5. 2025Sequential-Parallel Duality in Prefix Scannable Modelscited by 8
  6. 2026Latent Reasoning with Supervised Thinking Statescited by 8
  7. 2025Benefits and Limitations of Communication in Multi-Agent Reasoningcited by 7
  8. 2025Born a Transformer -- Always a Transformer? On the Effect of Pretraining on Architectural Abilitiescited by 6
  9. 2025Investigating Mechanisms for In-Context Vision Language Bindingcited by 5
  10. 2026Do Language Models Track Entities Across State Changes?cited by 5
  11. 2025A Unifying Framework for Parallelizing Sequential Models with Linear Dynamical Systemscited by 4
  12. 2026WMF-AM: Probing LLM Working Memory via Depth-Parameterized Cumulative State Trackingcited by 4
  13. 2025Is This Just Fantasy? Language Model Representations Reflect Human Judgments of Event Plausibilitycited by 3
  14. 2026Can Vision-Language Models Solve the Shell Game?cited by 2
  15. 2023Advancing Perception in Artificial Intelligence through Principles of Cognitive Sciencecited by 1
  16. 2025MET-Bench: Multimodal Entity Tracking for Evaluating the Limitations of Vision-Language and Reasoning Modelscited by 1
  17. 2026Why Depth Matters in Parallelizable Sequence Models: A Lie Algebraic Viewcited by 1
  18. 2026Barriers to Universal Reasoning With Transformers (And How to Overcome Them)cited by 1
  19. 2025Do Large Language Models know who did what to whom?cited by 0
  20. 2025Semantic F1 Scores: Fair Evaluation Under Fuzzy Class Boundariescited by 0
  21. 2025Circuit Distillationcited by 0
  22. 2026Flash PD-SSM: Memory-Optimized Structured Sparse State-Space Modelscited by 0
  23. 2026Entity Binding Failures in Speech LLM Reasoning: Diagnosis and Chain-of-Thought Interventioncited by 0
  24. 2026Persistent Sparse Autoencoders: Learning Feature-Specific Timescales in Language Model Representationscited by 0
  25. 2026EvolveScaler: Synthesizing Information-Evolution Contexts via Executable State Machines and Natural-Language Renderingcited by 0
  26. 2026PetriBench: Benchmarking LLM Reasoning over Dynamic State Spacescited by 0
  27. 2026Written as a Record, Read as an Address: What a Forward Pass Leaves in an Operation's KV Cachecited by 0
  28. 2026Length-Independent State Tracking Under a Parallel Scancited by 0
  29. 2026When Users Change Their Minds: Measuring and Repairing Intent Drift in LLM Agentscited by 0
  30. 2026Lost with a Map: Conversational State and Behavioral Reliability in Language Modelscited by 0
  31. 2026Transformers Stop Thinking Too Early, and a Tiny LoRA Fixes Itcited by 0
  32. 2026When Context Changes: Understanding Update Failures in LLMscited by 0