The paper critiques the common practice of rehearing all key‑value cache entries before compression, showing that under tight budgets this strategy actually harms performance. Experiments on RULER and LongBench reveal that full rereading retains far fewer useful entries than methods that rehearse nothing. The authors propose Attic‑KV, a training‑free rehearsal that quizzes the model with context‑based question‑answer pairs, which outperforms existing methods and scales better as compression budgets shrink.
arXiv:2609.36435v1 Announce Type: new Abstract: An assistant that serves the same user over a long horizon has to answer from what that user has revealed: which preferences still hold, which were rev...
Latent Compaction is a test‑time training method that learns a compact, document‑specific key–value cache for large‑context language models. It initializes cache entries using reconstruction importance, teacher attention, and key‑space diversity, then fits attention outputs and log‑normalizers to the full cache before jointly optimizing the entire cache via end‑to‑end distillation. Experiments on RULER, LongBench, and Phonebook Lookup show significant accuracy gains over budget‑matched compression baselines, achieving 96.7% accuracy at 32× compression on RULER and transferring well across model scales and families.
arXiv:2610.33025v1 Announce Type: cross Abstract: Letting a language model take gradient steps on the text it is reading, known as dynamic evaluation, lowers its loss on the rest of the document. Its...
The paper introduces REM (Reusable Experience Memory), a framework that manages the lifecycle of experience for large language model agents performing long-horizon coding tasks. REM consolidates completed trajectories into compact, phase-structured documents, injects them at task start, and evicts them after a single exposure, making the memory system responsive to the agent’s needs. Experiments on repository-level bug fixing show REM outperforms state‑of‑the‑art memory systems in retrieval accuracy and task resolution while reducing cost.
arXiv:2610.14754v1 Announce Type: new Abstract: Hybrid language models interleave Gated DeltaNet (GDN) layers, whose recurrent memory has a fixed size, with full-attention layers, whose key-value (KV...
The paper introduces Declarative Attention (DA), a protocol that lets language models explicitly declare which parts of their context to focus on during generation. By partitioning decoding into full-context, region-specific, and recent-output-only modes, the inference engine can skip large portions of the KV cache, dramatically reducing attended tokens. Experiments on 15 long-context tasks with off-the-shelf models show significant savings (52.0% and 31.1% reductions) with only modest accuracy drops that diminish as model size increases.
arXiv:2610.15715v1 Announce Type: new Abstract: Linear attention promises the efficiency needed for long-context language modelling, yet its fixed-size recurrent state must simultaneously support ong...
arXiv:2609.36835v1 Announce Type: new Abstract: Long-context large language model inference is bottlenecked by KV caches that grow linearly with sequence length. This burden is especially severe for...
arXiv:2610.22381v1 Announce Type: new Abstract: When a fact becomes harder to elicit from a language model, has the model lost it, or has it merely become less accessible? Parameter-updating test-tim...
arXiv:2606. 28876v2 Announce Type: replace-cross Abstract: 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.
The paper introduces LRE (Learned Relevance Eviction), a lightweight, CPU‑only, language‑model‑free scorer that learns which parts of an agent’s interaction history are task‑critical and preserves them verbatim. In experiments, LRE matches or surpasses baseline eviction policies on accuracy‑cost trade‑offs, recovers 93% of full‑history accuracy, reduces worst‑case prompt size by 52%, and outperforms dense and token‑pruning encoders in conversational memory while being 295–1569× smaller. The method also achieves superior budgeted answer quality on LoCoMo reading and can be trained annotation‑free, recovering 95% of supervised scorer performance.
arXiv:2607. 08032v1 Announce Type: new Abstract: Large language models, and the agents built on them, spend an ever-growing share of their compute and memory on remembering: caching attention keys and values, carrying long prompts, maintaining recurrent state, and storing what happened in previous turns and sessions.