Pith. sign in

REVIEW 4 cited by

LLoCO: Learning Long Contexts Offline

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

arxiv 2404.07979 v2 pith:AAASTQIJ submitted 2024-04-11 cs.CL cs.AIcs.LG

LLoCO: Learning Long Contexts Offline

classification cs.CL cs.AIcs.LG
keywords llococontextduringlongapproachcontextslearningtimes
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Processing long contexts remains a challenge for large language models (LLMs) due to the quadratic computational and memory overhead of the self-attention mechanism and the substantial KV cache sizes during generation. We propose LLoCO, a novel approach to address this problem by learning contexts offline through context compression and in-domain parameter-efficient finetuning with LoRA. Our method enables an LLM to create a concise representation of the original context and efficiently retrieve relevant information to answer questions accurately. Our approach extends the effective context window of a 4k token LLaMA2-7B model to handle up to 128k tokens. We evaluate our approach on several long-context question-answering datasets, demonstrating that LLoCO significantly outperforms in-context learning while using $30\times$ fewer tokens during inference. LLoCO achieves up to $7.62\times$ speed-up during inference and $11.52\times$ higher throughput during finetuning, substantially reduces the cost of long document question answering. This makes it a promising solution for efficient long context processing. Our code is publicly available on https://github.com/jeffreysijuntan/lloco.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 4 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Language Models Need Sleep: Learning to Self-Modify and Consolidate Memories

    cs.LG 2026-06 unverdicted novelty 6.0

    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.

  2. Language Models Need Sleep: Learning to Self-Modify and Consolidate Memories

    cs.LG 2026-06 conditional novelty 6.0

    Sleep-time Knowledge Seeding plus Dreaming lets LLMs expand capacity, distill fragile in-context memories into stable parameters, and self-improve without human labels.

  3. E2LLM: Encoder Elongated Large Language Models for Long-Context Understanding and Reasoning

    cs.CL 2024-09 unverdicted novelty 5.0

    E2LLM uses encoder-based soft prompt compression for long contexts to improve LLM reasoning on tasks like summarization and QA while maintaining efficiency.

  4. Parameter-Efficient Fine-Tuning for Large Models: A Comprehensive Survey

    cs.LG 2024-03 accept novelty 4.0

    A comprehensive survey of PEFT algorithms for large models, covering their performance, overhead, applications, and real-world system implementations.