Two-Stage Optimizer-Aware Online Data Selection for Large Language Models

1. 1Identify the limitations of existing gradient-based data selection methods in offline settings
2. 2Develop a two-stage optimizer-aware framework for online data selection
3. 3Implement the framework to adapt to sequential data arrival and step-dependent sample utility
4. 4Evaluate the effectiveness of the framework in improving fine-tuning efficiency

ML researchers and engineers working on large language models can benefit from this framework to optimize data selection and improve model performance

💡 Optimizer-aware online data selection can significantly improve the efficiency of large language model fine-tuning

🚀 Improve LLM fine-tuning with two-stage optimizer-aware online data selection!