Pith. sign in

REVIEW 22 cited by

Data Engineering for Scaling Language Models to 128K Context

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 2402.10171 v1 pith:5R3MAAJG submitted 2024-02-15 cs.CL cs.AI

Data Engineering for Scaling Language Models to 128K Context

classification cs.CL cs.AI
keywords datacontextmodelspretrainingtextitcontinuallanguagescaling
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

We study the continual pretraining recipe for scaling language models' context lengths to 128K, with a focus on data engineering. We hypothesize that long context modeling, in particular \textit{the ability to utilize information at arbitrary input locations}, is a capability that is mostly already acquired through large-scale pretraining, and that this capability can be readily extended to contexts substantially longer than seen during training~(e.g., 4K to 128K) through lightweight continual pretraining on appropriate data mixture. We investigate the \textit{quantity} and \textit{quality} of the data for continual pretraining: (1) for quantity, we show that 500 million to 5 billion tokens are enough to enable the model to retrieve information anywhere within the 128K context; (2) for quality, our results equally emphasize \textit{domain balance} and \textit{length upsampling}. Concretely, we find that naively upsampling longer data on certain domains like books, a common practice of existing work, gives suboptimal performance, and that a balanced domain mixture is important. We demonstrate that continual pretraining of the full model on 1B-5B tokens of such data is an effective and affordable strategy for scaling the context length of language models to 128K. Our recipe outperforms strong open-source long-context models and closes the gap to frontier models like GPT-4 128K.

discussion (0)

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

Forward citations

Cited by 22 Pith papers

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

  1. RULER: What's the Real Context Size of Your Long-Context Language Models?

    cs.CL 2024-04 accept novelty 8.0

    RULER shows most long-context LMs drop sharply in performance on complex tasks as length and difficulty increase, with only half maintaining results at 32K tokens.

  2. RoPE Distinguishes Neither Positions Nor Tokens in Long Contexts, Provably

    cs.CL 2026-05 conditional novelty 7.0

    Proves that RoPE attention loses locality bias and token distinction in long contexts, approaching random behavior independent of content.

  3. MLVU: Benchmarking Multi-task Long Video Understanding

    cs.CV 2024-06 conditional novelty 7.0

    MLVU is a new benchmark for long video understanding that uses extended videos across diverse genres and multi-task evaluations, revealing that current MLLMs struggle significantly and degrade sharply with longer durations.

  4. Leave No Context Behind: Efficient Infinite Context Transformers with Infini-attention

    cs.CL 2024-04 conditional novelty 7.0

    Infini-attention combines compressive memory with masked local attention and long-term linear attention inside each Transformer block to support infinite context length with bounded resources.

  5. Jet-Long: Efficient Long-Context Extension with Dynamic Bifocal RoPE

    cs.LG 2026-07 accept novelty 6.0

    Dynamic bifocal RoPE with adaptive group size recovers the base model at short context and outperforms fixed zero-shot long-context methods up to 128K with near-FA2 cost.

  6. Test-Time Training with Next-Token Prediction

    cs.CL 2026-06 unverdicted novelty 6.0

    TTT-NTP adapts pretrained LLMs at test time by training fast weights to match next-position hidden states from the forward pass, yielding consistent gains on long-context benchmarks across Llama, Mistral, and Qwen models.

  7. HPP: Hierarchical Programmatic Probing for Long Video Understanding by Decoupling Perception and Reasoning

    cs.CV 2026-06 unverdicted novelty 6.0

    HPP decouples perception from reasoning in long-video VLMs by having an LLM run iterative programmatic probes on hierarchically segmented video, reporting gains on LongVideoBench, EgoSchema, VideoMME, and MLVU.

  8. Training Long-Context Vision-Language Models Effectively with Generalization Beyond 128K Context

    cs.CV 2026-05 unverdicted novelty 6.0

    Continued pre-training with balanced long-document VQA data extends a 7B LVLM to 128K context, improving long-document VQA by 7.1% and generalizing to 512K without further training.

  9. Where Does Long-Context Supervision Actually Go? Effective-Context Exposure Balancing

    cs.CL 2026-05 conditional novelty 6.0

    EXACT re-allocates training supervision by inverse frequency of long effective-context targets, improving NoLiMa and RULER scores by 5-18 points on Qwen and LLaMA models without degrading standard QA or reasoning.

  10. Learning to Route Queries to Heads for Attention-based Re-ranking with Large Language Models

    cs.IR 2026-04 conditional novelty 6.0

    RouteHead trains a lightweight router to dynamically select optimal LLM attention heads per query for improved attention-based document re-ranking.

  11. Stacked from One: Multi-Scale Self-Injection for Context Window Extension

    cs.CL 2026-03 unverdicted novelty 6.0

    SharedLLM stacks two copies of a short-context LLM so the lower one compresses context into query-aware multi-grained tokens that are injected only at the lowest layers of the upper one, enabling generalization from 8...

  12. HeteroCache: A Dynamic Retrieval Approach to Heterogeneous KV Cache Compression for Long-Context LLM Inference

    cs.CL 2026-01 unverdicted novelty 6.0

    HeteroCache dynamically allocates KV cache space to attention heads based on their temporal stability and uses hierarchical asynchronous retrieval to achieve state-of-the-art long-context performance with up to 3x fas...

  13. TurboQuant: Online Vector Quantization with Near-optimal Distortion Rate

    cs.LG 2025-04 unverdicted novelty 6.0

    TurboQuant achieves near-optimal vector quantization distortion for both MSE and inner products via random rotation and per-coordinate scalar quantization, with a formal proof that it matches lower bounds within a fac...

  14. Jet-Long: Efficient Long-Context Extension with Dynamic Bifocal RoPE

    cs.LG 2026-07 accept novelty 5.0

    Jet-Long is a tuning-free bifocal RoPE method that dynamically sets remote group size from sequence length, recovering the base model within the pretrained window and beating prior zero-shot extenders on RULER, HELMET...

  15. Shuffle the Context: RoPE-Perturbed Self-Distillation for Long-Context Adaptation

    cs.CL 2026-04 unverdicted novelty 5.0

    RoPE-Perturbed Self-Distillation improves positional robustness during long-context fine-tuning of LLMs by training models to produce consistent outputs across RoPE-perturbed views of the input.

  16. Self-Describing Structured Data with Dual-Layer Guidance: A Lightweight Alternative to RAG for Precision Retrieval in Large-Scale LLM Knowledge Navigation

    cs.CL 2026-03 unverdicted novelty 5.0

    SDSR places human metadata at file primacy and combines it with prompt routing rules to reach 100% primary category accuracy on a 119-category benchmark, far above the 65% no-guidance baseline.

  17. Smarter, Better, Faster, Longer: A Modern Bidirectional Encoder for Fast, Memory Efficient, and Long Context Finetuning and Inference

    cs.CL 2024-12 unverdicted novelty 5.0

    ModernBERT is a new bidirectional encoder model achieving SOTA performance on diverse classification and retrieval benchmarks while offering superior speed and memory efficiency for long-context inference.

  18. Unlock the Potential of Large Language Models for Predictive Tabular Tasks in Data Science with Table-Specific Pretraining

    cs.LG 2024-03 unverdicted novelty 5.0

    Table-specific pretraining of Llama-2 yields significant gains on zero-shot, few-shot, and in-context tabular prediction tasks over prior benchmarks.

  19. Tokalator: A Context Engineering Toolkit for Artificial Intelligence Coding Assistants

    cs.SE 2026-04 unverdicted novelty 4.0

    Tokalator is a toolkit with VS Code extension, calculators, and community resources to monitor and optimize token usage in AI coding environments.

  20. Yi: Open Foundation Models by 01.AI

    cs.CL 2024-03 unverdicted novelty 4.0

    Yi models are 6B and 34B open foundation models pretrained on 3.1T curated tokens that achieve strong benchmark results through data quality and targeted extensions like long context and vision alignment.

  21. WorkBench Revisited: Workplace Agents Two Years On

    cs.AI 2026-06 unverdicted novelty 3.0

    Workplace agent success on WorkBench rose from 43% to 98% over two years with sharply reduced harmful actions.

  22. The Hitchhiker's Guide to Agentic AI: From Foundations to Systems

    cs.AI 2026-06 unverdicted novelty 2.0

    A comprehensive reference book organizing existing techniques for agentic AI systems across LLM substrate, reasoning, agent design patterns, inter-agent coordination, and production deployment.