mlsanityData to 14 Oct 2026

Interpreting time-series and weather models

14 papers0 this weekHigh confidence in this grouping: the papers form one clear topicHigh confidence in this grouping: the papers form one clear topic · About

Keywords: time series, series, time, interpretability, forecasting

Key facts. Interpreting time-series and weather models: 14 papers on arXiv since 2024 (data through 14 Oct 2026). 2 of its 14 mapped papers (14%) cite “Interpretability for Time Series Transformers using A Concept Bottleneck Framework” (2024, arXiv 2410.06070). 2 new papers 24 Sep–14 Oct.

What’s new

24 Sep – 14 Oct 2026

2 new since 24 Sep

  1. Mechanistic Interpretability of Atmospheric Rivers in GraphCast
  2. TSAE: Structured Sparse Autoencoders for Interpreting Time-Series Forecasting Models

Foundations

Most cited by papers here

  1. Interpretability for Time Series Transformers using A Concept Bottleneck Framework2024cited by 2 here
  2. Mechanistic Interpretability for Transformer-based Time Series Classification2025cited by 2 here
  3. Dissecting Chronos: Sparse Autoencoders Reveal Causal Feature Hierarchies in Time Series Foundation Models2026cited by 1 here

Lineage

The papers this direction builds on, oldest first.

Each paper on the line cites the one before.

  1. 2022Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 smallcited by 4 here
  2. 2023Progress measures for grokking via mechanistic interpretabilitycited by 1 here
  3. 2023A Toy Model of Universality: Reverse Engineering How Networks Learn Group Operationscited by 1 here
  4. 2023Towards Automated Circuit Discovery for Mechanistic Interpretabilitycited by 2 here
  5. 2023Sparse Autoencoders Find Highly Interpretable Features in Language Modelscited by 6 here
  6. 2023Towards Best Practices of Activation Patching in Language Models: Metrics and Methodscited by 1 here
  7. 2024Patchscopes: A Unifying Framework for Inspecting Hidden Representations of Language Modelscited by 2 here
  8. 2024Mechanistic Interpretability for AI Safety -- A Reviewcited by 1 here
  9. 2024A Practical Review of Mechanistic Interpretability for Transformer-Based Language Modelscited by 1 here
  10. 2024Interpretability for Time Series Transformers using A Concept Bottleneck Frameworkcited by 2 here
14 more it builds on, not in this chain
  1. 2022Toy Models of Superpositioncited by 2 here
  2. 2023Causal Abstraction: A Theoretical Foundation for Mechanistic Interpretabilitycited by 1 here
  3. 2023Towards Vision-Language Mechanistic Interpretability: A Causal Tracing Tool for BLIPcited by 1 here
  4. 2024Rethinking Interpretability in the Era of Large Language Modelscited by 1 here
  5. 2024Towards Uncovering How Large Language Model Works: An Explainability Perspectivecited by 1 here
  6. 2024Improving Dictionary Learning with Gated Sparse Autoencoderscited by 1 here
  7. 2024Scaling and evaluating sparse autoencoderscited by 3 here
  8. 2024Interpreting Attention Layer Outputs with Sparse Autoencoderscited by 1 here
  9. 2024Jumping Ahead: Improving Reconstruction Fidelity with JumpReLU Sparse Autoencoderscited by 1 here
  10. 2024Discover-then-Name: Task-Agnostic Concept Bottlenecks via Automated Concept Discoverycited by 1 here
  11. 2024Gemma Scope: Open Sparse Autoencoders Everywhere All At Once on Gemma 2cited by 1 here
  12. 2025Sparse Autoencoders Learn Monosemantic Features in Vision-Language Modelscited by 2 here
  13. 2025Mechanistic Interpretability for Transformer-based Time Series Classificationcited by 2 here
  14. 2026Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnetcited by 2 here

The 24 most cited of 31.

Browse

11 of 14, ranked 6–16

  1. 4Prompting Underestimates LLM Capability for Time Series Classification20260
  2. 5TimeSAE: Causal Sparse Decoding for Faithful Explanations of Black-Box Time Series Models20260
  3. 6Mechanistic Interpretability Tool for AI Weather Models20260
  4. 7Superposition Is Not Necessary: A Mechanistic Interpretability Analysis of Transformer Representations for Time Series Forecasting20260
  5. 8OceanCBM: A Concept Bottleneck Model for Mechanistic Interpretability in Ocean Forecasting20260
  6. 9Beyond Linear Superposition: Discovering Climate Features in AI Weather Models with KAN-SAE20260
  7. 10One Lens, Many Worlds : A Capability-Typed Interface for World-Model Interpretability20260
  8. 11Emergent Latent-State Computation under Stochastic Volatility20260
  9. 12SAE-Xplainers: Rule-Based Feature Interpretation for Extreme Earth Events20260
  10. 13Exploring the Effectiveness and Interpretability of Texts in LLM-based Time Series Models20250
All 11
  1. 14Integrating Time Series into LLMs via Multi-layer Steerable Embedding Fusion for Enhanced Forecasting20250