mlsanityData to 7 Oct 2026

Future Lens: Anticipating Subsequent Tokens from a Single Hidden State

arXiv 2311.04897 ↗First posted 8 Nov 2023cs.CL cs.LGcited by 116

Not on the map. Why?

Abstract (arXiv)

We conjecture that hidden state vectors corresponding to individual input tokens encode information sufficient to accurately predict several tokens ahead. More concretely, in this paper we ask: Given a hidden (internal) representation of a single token at position t in an input, can we reliably anticipate the tokens that will appear at positions ≥ t + 2? To test this, we measure linear approximation and causal intervention methods in GPT-J-6B to evaluate the degree to which individual hidden states in the network contain signal rich enough to predict future hidden states and, ultimately, token outputs. We find that, at some layers, we can approximate a model's output with more than 48% accuracy with respect to its prediction of subsequent tokens through a single hidden state. Finally we present a "Future Lens" visualization that uses these methods to create a new view of transformer states.

Key facts. Cited by 116 papers in the map (data through 7 Oct 2026). Builds on 22 of its 34 references that are in the map. First posted on arXiv 8 Nov 2023.

Builds on

22 of 34 references.

References · 22

  1. 2017Attention Is All You Needcited by 59,932
  2. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,204
  3. 2021Prefix-Tuning: Optimizing Continuous Prompts for Generationcited by 3,878
  4. 2022Locating and Editing Factual Associations in GPTcited by 2,609
  5. 2020The Pile: An 800GB Dataset of Diverse Text for Language Modelingcited by 2,346
  6. 2020Extracting Training Data from Large Language Modelscited by 2,144
  7. 2020Transformer Feed-Forward Layers Are Key-Value Memoriescited by 1,307
  8. 2022Mass-Editing Memory in a Transformercited by 976
  9. 2018The Secret Sharer: Evaluating and Testing Unintended Memorization in Neural Networkscited by 847
  10. 2019Universal Adversarial Triggers for Attacking and Analyzing NLPcited by 709
  11. 2023Eliciting Latent Predictions from Transformers with the Tuned Lenscited by 579
  12. 2020What Neural Networks Memorize and Why: Discovering the Long Tail via Influence Estimationcited by 492
10 more
  1. 2023Finding Neurons in a Haystack: Case Studies with Sparse Probingcited by 328
  2. 2023Linearity of Relation Decoding in Transformer Language Modelscited by 190
  3. 2020Accelerating Training of Transformer-Based Language Models with Progressive Layer Droppingcited by 93
  4. 2021Does BERT Pretrained on Clinical Notes Reveal Sensitive Data?cited by 91
  5. 2023Jump to Conclusions: Short-Cutting Transformers With Linear Transformationscited by 89
  6. 2022A Survey on Non-Autoregressive Generation for Neural Machine Translation and Beyondcited by 64
  7. 2023Evaluating the Zero-shot Robustness of Instruction-tuned Language Modelscited by 63
  8. 2022Understanding Transformer Memorization Recall Through Idiomscited by 62
  9. 2021Non-Autoregressive Text Generation with Pre-trained Language Modelscited by 36
  10. 2022LM-Debugger: An Interactive Tool for Inspection and Intervention in Transformer-Based Language Modelscited by 28

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 1; 2024: 40; 2025: 38; 2026: 37.

116 papers.

Sort:

By area:

  1. 2024Mechanistic Interpretability for AI Safety -- A Reviewcited by 415
  2. 2024Better & Faster Large Language Models via Multi-token Predictioncited by 344
  3. 2024Semantic Entropy Probes: Robust and Cheap Hallucination Detection in LLMscited by 246
  4. 2024Patchscopes: A Unifying Framework for Inspecting Hidden Representations of Language Modelscited by 225
  5. 2024Let's Think Dot by Dot: Hidden Computation in Transformer Language Modelscited by 186
  6. 2024The pitfalls of next-token predictioncited by 156
  7. 2024A Practical Review of Mechanistic Interpretability for Transformer-Based Language Modelscited by 113
  8. 2024Grokked Transformers are Implicit Reasoners: A Mechanistic Journey to the Edge of Generalizationcited by 78
  9. 2024Defending Large Language Models Against Jailbreak Attacks via Layer-specific Editingcited by 77
  10. 2025Do Multilingual LLMs Think In English?cited by 75
  11. 2025LLMs Encode Harmfulness and Refusal Separatelycited by 75
  12. 2024Transformers represent belief state geometry in their residual streamcited by 73
38 more
  1. 2024Knowledge Mechanisms in Large Language Models: A Survey and Perspectivecited by 50
  2. 2024LatentQA: Teaching LLMs to Decode Activations Into Natural Languagecited by 49
  3. 2024Do language models plan ahead for future tokens?cited by 47
  4. 2024From Tokens to Words: On the Inner Lexicon of LLMscited by 46
  5. 2024Who's asking? User personas and the mechanics of latent misalignmentcited by 40
  6. 2023Neuron-Level Knowledge Attribution in Large Language Modelscited by 37
  7. 2024Speculative Streaming: Fast LLM Inference without Auxiliary Modelscited by 37
  8. 2024Tracing Representation Progression: Analyzing and Enhancing Layer-Wise Similaritycited by 36
  9. 2024Investigating Layer Importance in Large Language Modelscited by 35
  10. 2024Evidence of Learned Look-Ahead in a Chess-Playing Neural Networkcited by 33
  11. 2025Sycophancy Is Not One Thing: Causal Separation of Sycophantic Behaviors in LLMscited by 32
  12. 2024Localizing Paragraph Memorization in Language Modelscited by 30
  13. 2025Pixels Versus Priors: Controlling Knowledge Priors in Vision-Language Models through Visual Counterfactscited by 30
  14. 2024Diffusion Lens: Interpreting Text Encoders in Text-to-Image Pipelinescited by 28
  15. 2025LLM Probability Concentration: How Alignment Shrinks the Generative Horizoncited by 26
  16. 2024AdvPrefix: An Objective for Nuanced LLM Jailbreakscited by 25
  17. 2024On Behalf of the Stakeholders: Trends in NLP Model Interpretability in the Era of LLMscited by 24
  18. 2025The Dual-Route Model of Inductioncited by 22
  19. 2026Locate, Steer, and Improve: A Practical Survey of Actionable Mechanistic Interpretability in Large Language Modelscited by 22
  20. 2024Unlocking the Future: Exploring Look-Ahead Planning Mechanistic Interpretability in Large Language Modelscited by 21
  21. 2024Predicting vs. Acting: A Trade-off Between World Modeling & Agent Modelingcited by 20
  22. 2025Emergent Response Planning in LLMscited by 17
  23. 2024Linking In-context Learning in Transformers to Human Episodic Memorycited by 15
  24. 2024Investigating Mysteries of CoT-Augmented Distillationcited by 15
  25. 2025AdaptiVocab: Enhancing LLM Efficiency in Focused Domains through Lightweight Vocabulary Adaptationcited by 15
  26. 2025How Do LLMs Use Their Depth?cited by 15
  27. 2024All or None: Identifiable Linear Properties of Next-token Predictors in Language Modelingcited by 14
  28. 2024Transformers Use Causal World Models in Maze-Solving Taskscited by 13
  29. 2025Is the Reversal Curse a Binding Problem? Uncovering Limitations of Transformers from a Basic Generalization Failurecited by 13
  30. 2025LLM Microscope: What Model Internals Reveal About Answer Correctness and Context Utilizationcited by 11
  31. 2025Interpretation Meets Safety: A Survey on Interpretation Methods and Tools for Improving LLM Safetycited by 10
  32. 2024Does Transformer Interpretability Transfer to RNNs?cited by 9
  33. 2024Extracting Paragraphs from LLM Token Activationscited by 9
  34. 2025Language Models Can Predict Their Own Behaviorcited by 8
  35. 2025Mechanistic Understanding and Mitigation of Language Confusion in English-Centric Large Language Modelscited by 8
  36. 2024InversionView: A General-Purpose Method for Reading Information from Neural Activationscited by 7
  37. 2025Unboxing the Black Box: Mechanistic Interpretability for Algorithmic Understanding of Neural Networkscited by 6
  38. 2026Eliciting Numerical Predictive Distributions of LLMs Without Autoregressioncited by 5

50 most cited of 116 All 116 (JSON)