Rethinking Contextualization by Reinterpreting Attention Head Channels
cs.LG, cs.CL
Submitted: 2026-09-27
Updated: 2026-09-27
Terminology
Sources
- SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator
- Unveiling Induction Heads: Provable Training Dynamics and Feature Learning in Transformers
- Binary Autoencoder for Mechanistic Interpretability of Large Language Models
- The Pile: An 800GB Dataset of Diverse Text for Language Modeling
- Decoupling the Benefits of Subword Tokenization for Language Model Training via Byte-level Simulation
- Successor Heads: Recurring, Interpretable Attention Heads In The Wild
- The Llama 3 Herd of Models
- How to use and interpret activation patching
- In-Context Learning Creates Task Vectors
- Linguistic Knowledge and Transferability of Contextual Representations
- Olmo 3
- In-context Learning and Induction Heads
- Norm of Word Embedding Encodes Information Gain
- Deep contextualized word representations
- Eliciting In-context Retrieval and Reasoning for Long-context Large Language Models
- Axiomatic Attribution for Deep Networks
- Llama 2: Open Foundation and Fine-Tuned Chat Models
- Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 small
- Label Words are Anchors: An Information Flow Perspective for Understanding In-Context Learning
- Retrieval Head Mechanistically Explains Long-Context Factuality
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: Local-Consistency Optimisation for Bayesian Neural Networks
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks