00:00:00 / 00:00:00

Synthetic Data – Friend or Foe in the Age of Scaling?

By Julia Kempe

Appears in collection : Mathematics for and by Large Language Models

As AI and LLM model size grows, neural scaling laws have become a crucial tool to predict the improvements of large models when increasing capacity and the size of original (human or natural) training data. Yet, the widespread use of popular models means that the ecosystem of online data and text will co-evolve to progressively contain increased amounts of synthesized data. In this talk we ask: How will the scaling laws change in the inevitable regime where synthetic data makes its way into the training corpus? Will future models, still improve, or be doomed to degenerate up to total (model) collapse? We develop a theoretical framework of model collapse through the lens of scaling laws. We discover a wide range of decay phenomena, analyzing loss of scaling, shifted scaling with number of generations, the ''un-learning" of skills, and grokking when mixing human and synthesized data. Our theory is validated by large-scale experiments with a transformer on an arithmetic task and text generation using the LLM Llama2.

Information about the video

  • Date of recording 23/05/2024
  • Date of publication 25/05/2024
  • Institution IHES
  • Language English
  • Audience Researchers
  • Format MP4

Last related questions on MathOverflow

You have to connect your Carmin.tv account with mathoverflow to add question

Ask a question on MathOverflow




Register

  • Bookmark videos
  • Add videos to see later &
    keep your browsing history
  • Comment with the scientific
    community
  • Get notification updates
    for your favorite subjects
Give feedback
Loading…
Loading the web debug toolbar…
Attempt #