pith. sign in

arxiv: 2401.18058 · v1 · pith:UEAXDXOCnew · submitted 2024-01-31 · 💻 cs.CL · cs.LG

LongAlign: A Recipe for Long Context Alignment of Large Language Models

classification 💻 cs.CL cs.LG
keywords longcontextdatalongalignlengthmodelstasksalignment
0
0 comments X
read the original abstract

Extending large language models to effectively handle long contexts requires instruction fine-tuning on input sequences of similar length. To address this, we present LongAlign -- a recipe of the instruction data, training, and evaluation for long context alignment. First, we construct a long instruction-following dataset using Self-Instruct. To ensure the data diversity, it covers a broad range of tasks from various long context sources. Second, we adopt the packing and sorted batching strategies to speed up supervised fine-tuning on data with varied length distributions. Additionally, we develop a loss weighting method to balance the contribution to the loss across different sequences during packing training. Third, we introduce the LongBench-Chat benchmark for evaluating instruction-following capabilities on queries of 10k-100k in length. Experiments show that LongAlign outperforms existing recipes for LLMs in long context tasks by up to 30\%, while also maintaining their proficiency in handling short, generic tasks. The code, data, and long-aligned models are open-sourced at https://github.com/THUDM/LongAlign.

This paper has not been read by Pith yet.

discussion (0)

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

Forward citations

Cited by 8 Pith papers

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

  1. EndPrompt: Efficient Long-Context Extension via Terminal Anchoring

    cs.CL 2026-05 conditional novelty 7.0

    EndPrompt induces reliable long-context generalization in LLaMA models from sparse positional supervision via a two-segment short-sequence construction with terminal anchoring.

  2. Long-Text-to-Image Generation via Compositional Prompt Decomposition

    cs.CV 2026-04 unverdicted novelty 7.0

    PRISM lets pre-trained text-to-image models handle long prompts by breaking them into compositional parts, predicting noise separately, and merging outputs via energy-based conjunction, matching fine-tuned models whil...

  3. EndPrompt: Efficient Long-Context Extension via Terminal Anchoring

    cs.CL 2026-05 unverdicted novelty 6.0

    EndPrompt induces long-context generalization in LLaMA models via a two-segment short-sequence construction with terminal positional anchoring, outperforming full fine-tuning and prior methods on RULER and LongBench w...

  4. 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.

  5. CogVideoX: Text-to-Video Diffusion Models with An Expert Transformer

    cs.CV 2024-08 unverdicted novelty 6.0

    CogVideoX generates coherent 10-second text-to-video outputs at high resolution using a 3D VAE, expert adaptive LayerNorm transformer, progressive training, and a custom data pipeline, claiming state-of-the-art results.

  6. A Decomposition Perspective to Long-context Reasoning for LLMs

    cs.CL 2026-04 unverdicted novelty 5.0

    Decomposing long-context reasoning into atomic skills, synthesizing targeted pseudo-datasets, and applying RL improves LLM performance on long-context benchmarks by an average of 7.7%.

  7. Context Collapse: Barriers to Adoption for Generative AI in Workplace Settings

    cs.CY 2026-04 unverdicted novelty 5.0

    Expert interviews demonstrate that context in generative AI workplace use collapses or rots over time, limiting tool effectiveness and revealing pitfalls in computational context approaches.

  8. SmolLM2: When Smol Goes Big -- Data-Centric Training of a Small Language Model

    cs.CL 2025-02 unverdicted novelty 5.0

    SmolLM2 is a 1.7B-parameter language model that outperforms Qwen2.5-1.5B and Llama3.2-1B after overtraining on 11 trillion tokens using custom FineMath, Stack-Edu, and SmolTalk datasets in a multi-stage pipeline.