REVIEW 19 cited by
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
Cartridges: Lightweight and general-purpose long context representations via self-study
read the original abstract
Large language models are often used to answer queries grounded in large text corpora (e.g. codebases, legal documents, or chat histories) by placing the entire corpus in the context window and leveraging in-context learning (ICL). Although current models support contexts of 100K-1M tokens, this setup is costly to serve because the memory consumption of the KV cache scales with input length. We explore an alternative: training a smaller KV cache offline on each corpus. At inference time, we load this trained KV cache, which we call a Cartridge, and decode a response. Critically, the cost of training a Cartridge can be amortized across all the queries referencing the same corpus. However, we find that the naive approach of training the Cartridge with next-token prediction on the corpus is not competitive with ICL. Instead, we propose self-study, a training recipe in which we generate synthetic conversations about the corpus and train the Cartridge with a context-distillation objective. We find that Cartridges trained with self-study replicate the functionality of ICL, while being significantly cheaper to serve. On challenging long-context benchmarks, Cartridges trained with self-study match ICL performance while using 38.6x less memory and enabling 26.4x higher throughput. Self-study also extends the model's effective context length (e.g. from 128k to 484k tokens on MTOB) and surprisingly, leads to Cartridges that can be composed at inference time without retraining.
Forward citations
Cited by 19 Pith papers
-
Continual Learning Bench: Evaluating Frontier AI Systems in Real-World Stateful Environments
CL-Bench is the first expert-validated benchmark for continual learning in frontier LLMs across six real-world domains, showing limited gains and that naive in-context learning outperforms dedicated memory systems.
-
Self-Study Reconsidered: The Hidden Fragility of Learning from Self-Generated QA
Self-generated QA supervision for language models is fragile due to non-uniform question selection and instruction compliance during answering, with mitigations that reduce compliance from 88% to 13%.
-
Doc-to-Atom: Learning to Compile and Compose Memory Atoms
Doc-to-Atom decomposes documents into composable micro-LoRA adapters selected by a query router for efficient long-context QA.
-
Do Language Models Need Sleep? Offline Recurrence for Improved Online Inference
A sleep mechanism with N offline recurrent passes consolidates context into fast weights, improving performance on reasoning tasks where standard transformers fail.
-
MeMo: Memory as a Model
MeMo encodes new knowledge into a separate memory model for frozen LLMs, achieving strong performance on BrowseComp-Plus, NarrativeQA, and MuSiQue while capturing cross-document relationships and remaining robust to r...
-
SnapStream: Efficient Long Sequence Decoding on Dataflow Accelerators
SnapStream deploys sparse KV attention in a production inference system on dataflow accelerators, delivering 4x on-chip memory savings for DeepSeek-671B at 128k context with up to 1832 tokens/sec and minimal accuracy ...
-
MIRIX: Multi-Agent Memory System for LLM-Based Agents
MIRIX introduces a modular multi-agent architecture with Core, Episodic, Semantic, Procedural, Resource, and Knowledge Vault memories that outperforms RAG baselines by 35% on ScreenshotVQA and reaches 85.4% on LOCOMO.
-
What to Keep, What to Forget: A Rate--Distortion View of Memory Compaction in LLMs and Agents
KV-cache eviction, prompt compression, recurrent state bounding, and agent memory consolidation are unified as one rate-distortion problem with a shared lower bound, shared failure mode, and transferable mechanisms.
-
End-to-End Context Compression at Scale
LCLMs are scaled 0.6B-encoder 4B-decoder compressors pre-trained on over 350B tokens that improve the Pareto frontier for general-task performance, compression speed, and peak memory in long-context language model inference.
-
Still: Amortized KV Cache Compaction in a Single Forward Pass
Still is an amortized per-layer Perceiver that synthesizes compact KV caches in one forward pass, outperforming selection and per-context baselines on RULER, HELMET, and LongBench at 8-200x compression.
-
Language Models Need Sleep: Learning to Self-Modify and Consolidate Memories
Language models can use a two-stage sleep process of upward distillation for memory consolidation and RL-based dreaming for unsupervised self-improvement to enable continual learning.
-
Knowledge Offloading: Decomposing LLMs into Sparse Backbones and Memory Modules
KOFF prunes LLMs to ~12% sparsity while adding LoRA and learned KV memories, preserving performance where plain pruning fails across 3B-8B Llama and Qwen models.
-
Parallel Context Compaction for Long-Horizon LLM Agent Serving
Parallel compaction for LLM agent context management provides predictable volume control and reduces wall time versus sequential baselines on HotpotQA and LoCoMo.
-
PreFT: Prefill-only finetuning for efficient inference
Prefill-only adaptation of LLMs yields 1.9x higher throughput for 512 adapters on Llama 3.1 70B with near-parity performance on RL tasks and recoverable loss on SFT.
-
The Verbose Context Problem in Medical Records
Presents PopMedQA benchmark and shows domain-independent LLM methods fail on token-inefficient longitudinal medical records, leaving room for domain-specific approaches.
-
Amortizing Federated Adaptation: Hypernetwork Driven LoRA for Personalized Foundation Models
HyperLoRA amortizes federated LoRA adaptation via hypernetwork-generated initializations and product-space aggregation to fix structural bias and initialization lag.
-
MeMo: Memory as a Model
MeMo encodes new knowledge into a separate memory model that integrates with frozen LLMs, showing strong performance on QA benchmarks while avoiding catastrophic forgetting and working without access to model weights.
-
Context Distillation as Latent Memory Management
Context distillation is reformulated as latent memory management via a bank of LoRA adapters, retrieval, routing, and self-gating to outperform retrieval baselines and improve robustness.
-
Back to Basics: Let Conversational Agents Remember with Just Retrieval and Generation
A minimalist retrieval-and-generation framework using turn isolation and query-driven pruning outperforms complex memory systems by directly addressing signal sparsity and dual-level redundancy in dialogues.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.