LLM-AutoDP: Automatic Data Processing via LLM Agents for Model Fine-tuning

Dimensions overall score 8.0

GitHub Code Pulse

No public code linked for this paper yet.

• LLM-AutoDP achieves over 80% win rates against models trained on unprocessed data.result 90%
• LLM-AutoDP achieves approximately a 65% win rate compared to AutoML baselines based on LLM agents.result 90%
• The acceleration techniques in LLM-AutoDP reduce the total searching time by up to 10 times.technical 90%
• LLM-AutoDP enables automated data processing without exposing raw data, addressing privacy concerns in sensitive domains.technical 85%
• The framework uses an iterative in-context learning mechanism that enables convergence toward high-quality processing pipelines without direct human intervention.method 85%
• The system assumes availability of representative datasets for initial strategy formulation.limitation 80%
• LLM-AutoDP was tested on five medical datasets across three model architectures.method 85%
• The framework introduces Distribution Preserving Sampling to reduce data volume while maintaining distributional integrity.technical 90%

Competitive landscape

Competitor map is still being generated for this paper. Enable generation or check back soon.

Startup potential card