Research
Synthetic Data Characterization via Training Dynamics
arXiv:2609.39447v1 Announce Type: new Abstract: Interpreting properties of LLM-generated data is important for understanding its utility and limitations across learning tasks. In this work, we charact
arXiv:2609.39447v1 Announce Type: new Abstract: Interpreting properties of LLM-generated data is important for understanding its utility and limitations across learning tasks. In this work, we characterize synthetic data through sample-level learnability, studying variation among LLM families and scales, alongside human-written data as a reference. We first generate synthetic datasets spanning single- and multi-label classification, labeling, and tree prediction tasks. We then derive empirical data distributions from encoder training dynamics for both machine and organic data, and estimate the robustness of these distributions across encoders. Finally, we evaluate how data selection strategies based on these learnability signals affect both data sources differently.
Related
- Geometric Filtering of LLM-Generated Samples for Few-Shot Text Classification
- Measuring Task-Agnostic Training Data Influence Across Language Model Pretraining
- How Can We Synthesize High-Quality Pretraining Data? A Systematic Study of Prompt Design, Generator Model, and Source Data
Source: arXiv cs.CL | 2026-10-01