Data Engineering for Scaling Language Models to 128K Context

• URL: http://arxiv.org/abs/2402.10171v1
• Date: Thu, 15 Feb 2024 18:19:16 GMT
• Title: Data Engineering for Scaling Language Models to 128K Context
• Authors: Yao Fu, Rameswar Panda, Xinyao Niu, Xiang Yue, Hannaneh Hajishirzi, Yoon Kim and Hao Peng
• Abstract summary: We study the continual pretraining recipe for scaling language models' context lengths to 128K. We find that naively upsampling longer data on certain domains like books, a common practice of existing work, gives suboptimal performance. Our recipe outperforms strong open-source long-context models and closes the gap to frontier models like GPT-4 128K.
• Score: 98.41554785106902
• License: http://creativecommons.org/licenses/by/4.0/
• 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.

Related papers

• Scaling Pre-training to One Hundred Billion Data for Vision Language Models [23.437750698345067]
  We investigate the potential of pre-training vision-language models on an unprecedented scale: 100 billion examples. We find that model performance tends to saturate at this scale on many common Western-centric classification and retrieval benchmarks. tasks of cultural diversity achieve more substantial gains from the 100-billion scale web data, thanks to its coverage of long-tail concepts.
  arXiv  Detail & Related papers  (2025-02-11T15:05:33Z)
• Predicting Large Language Model Capabilities on Closed-Book QA Tasks Using Only Information Available Prior to Training [51.60874286674908]
  We focus on predicting performance on Closed-book Question Answering (CBQA) tasks, which are closely tied to pre-training data and knowledge retention. We address three major challenges: 1) mastering the entire pre-training process, especially data construction; 2) evaluating a model's knowledge retention; and 3) predicting task-specific knowledge retention using only information available prior to training. We introduce the SMI metric, an information-theoretic measure that quantifies the relationship between pre-training data, model size, and task-specific knowledge retention.
  arXiv  Detail & Related papers  (2025-02-06T13:23:53Z)
• The interplay between domain specialization and model size: a case study in the legal domain [8.653321928148547]
  We investigate the interplay between domain and model size during continual pre-training under compute-constrained scenarios. Our goal is to identify a compute-efficient training regime for this scenario. As model size increases, the compute-effectiveness gap between specialized and general models widens.
  arXiv  Detail & Related papers  (2025-01-03T19:28:53Z)
• Untie the Knots: An Efficient Data Augmentation Strategy for Long-Context Pre-Training in Language Models [21.90388980448712]
  Training models to handle long contexts presents significant challenges. We introduce Untie the Knots (textbfUtK), a novel data augmentation strategy employed during the continue pre-training phase. We conduct extensive experiments on models with 7B and 72B parameters, trained on 20 billion tokens, demonstrating that UtK achieves 75% and 84.5% accurracy on RULER at 128K context length.
  arXiv  Detail & Related papers  (2024-09-07T09:28:55Z)
• Training-Free Long-Context Scaling of Large Language Models [114.53296002607993]
  We propose Dual Chunk Attention, which enables Llama2 70B to support context windows of more than 100k tokens without continual training. By decomposing the attention for long sequences into chunk-based modules, DCA manages to effectively capture the relative positional information of tokens.
  arXiv  Detail & Related papers  (2024-02-27T12:39:23Z)
• Delving Deeper into Data Scaling in Masked Image Modeling [145.36501330782357]
  We conduct an empirical study on the scaling capability of masked image modeling (MIM) methods for visual recognition. Specifically, we utilize the web-collected Coyo-700M dataset. Our goal is to investigate how the performance changes on downstream tasks when scaling with different sizes of data and models.
  arXiv  Detail & Related papers  (2023-05-24T15:33:46Z)
• Pre-Training to Learn in Context [138.0745138788142]
  The ability of in-context learning is not fully exploited because language models are not explicitly trained to learn in context. We propose PICL (Pre-training for In-Context Learning), a framework to enhance the language models' in-context learning ability. Our experiments show that PICL is more effective and task-generalizable than a range of baselines, outperforming larger language models with nearly 4x parameters.
  arXiv  Detail & Related papers  (2023-05-16T03:38:06Z)
• Efficient Speech Translation with Pre-trained Models [13.107314023500349]
  We investigate efficient strategies to build cascaded and end-to-end speech translation systems based on pre-trained models. While the end-to-end models show superior translation performance to cascaded ones, the application of this technology has a limitation on the need for additional end-to-end training data.
  arXiv  Detail & Related papers  (2022-11-09T15:07:06Z)
• Grounded Compositional Outputs for Adaptive Language Modeling [59.02706635250856]
  A language model's vocabulary$-$typically selected before training and permanently fixed later$-$affects its size. We propose a fully compositional output embedding layer for language models. To our knowledge, the result is the first word-level language model with a size that does not depend on the training vocabulary.
  arXiv  Detail & Related papers  (2020-09-24T07:21:14Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.