mlsanityData to 14 Oct 2026

ParaScopes: What do Language Models Activations Encode About Future Text?

arXiv 2511.00180 ↗First posted 31 Oct 2025cs.CL cs.LGcited by 2

Not on the map. Why?

Abstract (arXiv)

Interpretability studies in language models often investigate forward-looking representations of activations. However, as language models become capable of doing ever longer time horizon tasks, methods for understanding activations often remain limited to testing specific concepts or tokens. We develop a framework of Residual Stream Decoders as a method of probing model activations for paragraph-scale and document-scale plans. We test several methods and find information can be decoded equivalent to 5+ tokens of future context in small models. These results lay the groundwork for better monitoring of language models and better understanding how they might encode longer-term planning information.

Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Builds on 34 of its 42 references that are in the map. First posted on arXiv 31 Oct 2025.

Builds on

34 of 42 references.

References · 34

  1. 2017Attention Is All You Needcited by 60,361
  2. 2024The Llama 3 Herd of Modelscited by 14,715
  3. 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arenacited by 8,626
  4. 2022Locating and Editing Factual Associations in GPTcited by 2,691
  5. 2023G-Eval: NLG Evaluation using GPT-4 with Better Human Alignmentcited by 2,276
  6. 2024Gemma 2: Improving Open Language Models at a Practical Sizecited by 1,998
  7. 2025Gemma 3 Technical Reportcited by 1,824
  8. 2023Sparse Autoencoders Find Highly Interpretable Features in Language Modelscited by 1,420
  9. 2025Qwen3 Embedding: Advancing Text Embedding and Reranking Through Foundation Modelscited by 1,261
  10. 2022Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 smallcited by 1,165
  11. 2024The FineWeb Datasets: Decanting the Web for the Finest Text Data at Scalecited by 1,160
  12. 2022In-context Learning and Induction Headscited by 964
22 more
  1. 2021Probing Classifiers: Promises, Shortcomings, and Advancescited by 876
  2. 2022Discovering Latent Knowledge in Language Models Without Supervisioncited by 767
  3. 2023Prometheus: Inducing Fine-grained Evaluation Capability in Language Modelscited by 575
  4. 2024Scaling and evaluating sparse autoencoderscited by 553
  5. 2019What do you learn from context? Probing for sentence structure in contextualized word representationscited by 438
  6. 2024Patchscopes: A Unifying Framework for Inspecting Hidden Representations of Language Modelscited by 235
  7. 2025Measuring AI Ability to Complete Long Software Taskscited by 156
  8. 2023Future Lens: Anticipating Subsequent Tokens from a Single Hidden Statecited by 121
  9. 2023SONAR: Sentence-Level Multimodal and Language-Agnostic Representationscited by 108
  10. 2021Learning Compact Metrics for MTcited by 93
  11. 2024Large Concept Models: Language Modeling in a Sentence Representation Spacecited by 89
  12. 2025Auditing language models for hidden objectivescited by 83
  13. 2025On the Emergence of Position Bias in Transformerscited by 59
  14. 2024LatentQA: Teaching LLMs to Decode Activations Into Natural Languagecited by 51
  15. 2024Do language models plan ahead for future tokens?cited by 49
  16. 2024Speculative Streaming: Fast LLM Inference without Auxiliary Modelscited by 38
  17. 2024Evidence of Learned Look-Ahead in a Chess-Playing Neural Networkcited by 34
  18. 2024Predicting vs. Acting: A Trade-off Between World Modeling & Agent Modelingcited by 20
  19. 2024Planning in a recurrent neural network that plays Sokobancited by 18
  20. 2024Extracting Paragraphs from LLM Token Activationscited by 9
  21. 2024Semformer: Transformer Language Models with Semantic Planningcited by 9
  22. 2024Quantifying Positional Biases in Text Embedding Modelscited by 4

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 2.

2 papers.

  1. 2026Where's the Plan? Locating Latent Planning in Language Models with Lightweight Mechanistic Interventionscited by 1
  2. 2026Script Choice in LLMs: Evidence for Late-Layer Commitmentcited by 1

Similar, not cited

Close in wording, no citation either way

  1. 2026Measuring Activation Control in Large Language Models
  2. 2025Are language models aware of the road not taken? Token-level uncertainty and hidden state dynamics
  3. 2024Activation Scaling for Steering and Interpreting Language Models
  4. 2025Internal Planning in Language Models: Characterizing Horizon and Branch Awareness
  5. 2026Learning to Read Out: Unembedding Dynamics in Language Model Pretraining
  6. 2024Towards Unifying Interpretability and Control: Evaluation via Intervention
  7. 2025Explainability of Large Language Models: Opportunities and Challenges toward Generating Trustworthy Explanations
  8. 2026Do Vision-Language-Action Models Understand Instructions? A Mechanistic Interpretability Study on Language Grounding