CORTEXA
← Browse
arxivcs.CLcs.LG2026-06-26

Turn-Averaged SAEs for Feature Discovery and Long-Context Attribution

Kevin Der, Harish Kamath, Ben Thompson

Sparse autoencoders (SAEs) have become a useful tool for extracting interpretable features in language models. However, standard SAE architectures operate on individual token activations, meaning that the number of active features scales linearly with context length, and studying long model transcripts becomes difficult. We introduce turn-averaged SAEs, which represent a single Human or Assistant turn with a fixed number of features by learning to reconstruct the average model activation across the turn. We find that turn-averaged features describe a single turn's high-level characteristics more completely than per-token features when judged by an LLM. We also demonstrate that turn-averaged SAEs greatly simplify common downstream uses of SAEs like attribution graphs. Broadly, turn-averaged SAEs make interpretability techniques practical at long context lengths.

View free PDFSource page

Related papers

arxivcs.CLcs.AIcs.LG2026-06-29

MATCH: Modulating Attention via In-Context Retrieval for Long-Context Transformers

Linrui Ma, Chun Hei Lo, Xinyu Wang, Peng Lu, Xihao Yuan, Hanting Chen, et al.

The quadratic computational cost of traditional attention mechanisms poses a major bottleneck to the scalability and practical deployment of large language models (LLMs), particularly in long-context scenarios. To improve efficiency, existing approaches often enforce rigid struct…

View free PDFSource page
arxivcs.CLcs.LG2026-06-27

Memory-Managed Long-Context Attention: Bounded Editable Memory with a Hard Lifecycle and Calibrated Sparse Fallback

Junyi Zou, Avrova Donz

We study memory-managed long-context attention: explicit bounded memory with a learned query-independent writer, lifecycle control, query-aware reading, calibrated sparse fallback, and frozen-LLM generation from raw evidence. Track A is a controlled versioned-variable task where…

View free PDFSource page
arxivcs.CLcs.LG2026-07-16

VarRate: Training-Free Variable-Rate KV Cache Compression for Long-Context LLMs

Shahrzad Esmat, Dhawal Shah, Ali Jannesari

The key-value (KV) cache is the main memory bottleneck in long-context large language model (LLM) inference. Two leading training-free families are both structurally limited: token-selection methods (SnapKV, Ada-KV) score importance from an observation window and evict low-scorin…

View free PDFSource page
arxivcs.LGcs.CL2026-06-30

RaBitQCache: Rotated Binary Quantization for KVCache in Long Context LLM Inference

Wenhao Li, Jinhao Dong, Hailin Zhang, Wenhang Shi, Wei Lu, Xiaoyong Du

Long-context Large Language Model inference is severely bottlenecked by the massive Key-Value (KV) cache, yet existing sparse attention methods often suffer from static fixed-budget (Top-k) retrieval or rely on proxy scores that are computationally expensive and biased. To addres…

View free PDFSource page
arxivcs.LGcs.AIcs.CLcs.IR2026-07-11

Context by Distinct Information: An Auditable Dirichlet-Process Working Memory for Long, Redundant Context Streams

Siddharth Pal, Viktoria Rojkova

Context engineering decides what information a model carries forward, and current designs meter it in tokens: compressing the past into a bounded recurrent state, keeping a key-value entry for every token, or imposing a fixed budget through a window or eviction rule. All three ma…

View free PDFSource page
arxivcs.LGcs.AIcs.CL2026-07-02

Training Hybrid Block Diffusion Language Models with Partial Bidirectionality

Pranshu Chaturvedi, Parth Shroff, Tarun Suresh, Hangoo Kang, Kaiyue Wen

High-throughput long-context generation is one of the central challenges for large language models. Generation is typically memory-bandwidth-bound rather than compute-bound: each decoding step must stream the accumulated key/value (KV) cache from memory, so bandwidth demand grows…

View free PDFSource page