# Diagnosing Capability Gaps in Fine-Tuning Data

arXiv: https://arxiv.org/abs/2604.27547 · HTML: https://mlsanity.com/paper/2604.27547/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 21 of its 26 references that are in the map.
- First posted on arXiv 30 Apr 2026.

## Builds on (most cited first, 21 of 21)

- [Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena](https://mlsanity.com/paper/2306.05685/) (2023)
- [Finetuned Language Models Are Zero-Shot Learners](https://mlsanity.com/paper/2109.01652/) (2021)
- [Self-Instruct: Aligning Language Models with Self-Generated Instructions](https://mlsanity.com/paper/2212.10560/) (2022)
- [Learning to summarize from human feedback](https://mlsanity.com/paper/2009.01325/) (2020)
- [G-Eval: NLG Evaluation using GPT-4 with Better Human Alignment](https://mlsanity.com/paper/2303.16634/) (2023)
- [WILDS: A Benchmark of in-the-Wild Distribution Shifts](https://mlsanity.com/paper/2012.07421/) (2020)
- [PubMedQA: A Dataset for Biomedical Research Question Answering](https://mlsanity.com/paper/1909.06146/) (2019)
- [LIMA: Less Is More for Alignment](https://mlsanity.com/paper/2305.11206/) (2023)
- [Length-Controlled AlpacaEval: A Simple Way to Debias Automatic Evaluators](https://mlsanity.com/paper/2404.04475/) (2024)
- [Teaching Models to Express Their Uncertainty in Words](https://mlsanity.com/paper/2205.14334/) (2022)
- [Prometheus 2: An Open Source Language Model Specialized in Evaluating Other Language Models](https://mlsanity.com/paper/2405.01535/) (2024)
- [Dataset Cartography: Mapping and Diagnosing Datasets with Training Dynamics](https://mlsanity.com/paper/2009.10795/) (2020)
- [Confident Learning: Estimating Uncertainty in Dataset Labels](https://mlsanity.com/paper/1911.00068/) (2019)
- [GPTScore: Evaluate as You Desire](https://mlsanity.com/paper/2302.04166/) (2023)
- [From Quantity to Quality: Boosting LLM Performance with Self-Guided Data Selection for Instruction Tuning](https://mlsanity.com/paper/2308.12032/) (2023)
- [Identifying Mislabeled Data using the Area Under the Margin Ranking](https://mlsanity.com/paper/2001.10528/) (2020)
- [Data-centric Artificial Intelligence: A Survey](https://mlsanity.com/paper/2303.10158/) (2023)
- [Generative Judge for Evaluating Alignment](https://mlsanity.com/paper/2310.05470/) (2023)
- [BillSum: A Corpus for Automatic Summarization of US Legislation](https://mlsanity.com/paper/1910.00523/) (2019)
- [Grading Scale Impact on LLM-as-a-Judge: Human-LLM Alignment Is Highest on 0-5 Grading Scale](https://mlsanity.com/paper/2601.03444/) (2026)
- [SibylSense: Adaptive Rubric Learning via Memory Tuning and Adversarial Probing](https://mlsanity.com/paper/2602.20751/) (2026)

## Abstract (arXiv)

Fine-tuning large language models (LLMs) for domain-specific tasks requires training datasets that comprehensively cover the target capabilities a practitioner needs. Yet identifying which capabilities a dataset fails to support, and doing so before an expensive fine-tuning run, remains a largely unsolved problem. We introduce GoalCover, a framework that helps practitioners systematically detect capability gaps in fine-tuning datasets through interactive goal decomposition and automated coverage assessment. GoalCover guides a practitioner through structured decomposition of a high-level goal into atomic, independently evaluable subgoals; assigns each training sample an LLM-based alignment score against every subgoal; and surfaces missing capabilities through automated analysis of low-scoring sample explanations. We validate the framework along two complementary axes. First, through controlled corruption experiments across three domains (medical QA, legal summarization, code generation), we show that GoalCover reliably distinguishes targeted from non-targeted capability impacts: target subgoals degrade by 25.6% on average versus 2.1% for non-target subgoals (Cohen's d=1.24). Second, we demonstrate downstream utility on a financial-summarization Reinforcement Fine-Tuning (RFT) task with Qwen-3-14B: training on GoalCover-filtered data improves the LLM-judge reward from 3.77 to 4.12 (out of 5) over the unfiltered baseline, and combining filtered data with goal-conditioned synthetic samples yields the strongest result (4.20). The two results together show that GoalCover works as a practical pre-fine-tuning diagnostic: it detects capability gaps and produces concrete signal for closing them.