Decoding Data Quality via Synthetic Corruptions: Embedding-guided Pruning of Code Data
This addresses data quality issues for code generation models, but it is incremental as it builds on existing embedding-based pruning methods.
The paper tackled the problem of low-quality code data in datasets like Stack, which affects LLM performance, by developing a synthetic corruption informed pruning (SCIP) method that uses embeddings to identify and remove such data, resulting in up to a 3% performance improvement on HumanEval and MBPP benchmarks.
Code datasets, often collected from diverse and uncontrolled sources such as GitHub, potentially suffer from quality issues, thereby affecting the performance and training efficiency of Large Language Models (LLMs) optimized for code generation. Previous studies demonstrated the benefit of using embedding spaces for data pruning, but they mainly focused on duplicate removal or increasing variety, and in other modalities, such as images. Our work focuses on using embeddings to identify and remove "low-quality" code data. First, we explore features of "low-quality" code in embedding space, through the use of synthetic corruptions. Armed with this knowledge, we devise novel pruning metrics that operate in embedding space to identify and remove low-quality entries in the Stack dataset. We demonstrate the benefits of this synthetic corruption informed pruning (SCIP) approach on the well-established HumanEval and MBPP benchmarks, outperforming existing embedding-based methods. Importantly, we achieve up to a 3% performance improvement over no pruning, thereby showing the promise of insights from synthetic corruptions for data pruning.