# Measuring the Partial-Credit Gap: A Strict Benchmark on Vietnam's 2025 Convex Marking Scheme

arXiv: https://arxiv.org/abs/2608.18336 · HTML: https://mlsanity.com/paper/2608.18336/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 12 of its 14 references that are in the map.
- First posted on arXiv 18 Aug 2026.

## Builds on (most cited first, 12 of 12)

- [ArabicMMLU: Assessing Massive Multitask Language Understanding in Arabic](https://mlsanity.com/paper/2402.12840/) (2024)
- [Can multiple-choice questions really be useful in detecting the abilities of LLMs?](https://mlsanity.com/paper/2403.17752/) (2024)
- [Large Language Models Only Pass Primary School Exams in Indonesia: A Comprehensive Test on IndoMMLU](https://mlsanity.com/paper/2310.04928/) (2023)
- [Psychometric Alignment: Capturing Human Knowledge Distributions via Language Models](https://mlsanity.com/paper/2407.15645/) (2024)
- [VNHSGE: VietNamese High School Graduation Examination Dataset for Large Language Models](https://mlsanity.com/paper/2305.12199/) (2023)
- [CMPhysBench: A Benchmark for Evaluating Large Language Models in Condensed Matter Physics](https://mlsanity.com/paper/2508.18124/) (2025)
- [SinhalaMMLU: A Comprehensive Benchmark for Evaluating Multitask Language Understanding in Sinhala](https://mlsanity.com/paper/2509.03162/) (2025)
- [RadSEM: A Finding-by-Finding Metric for Clinical Consistency in Radiology Reports](https://mlsanity.com/paper/2606.17062/) (2026)
- [LaoBench: A Large-Scale Multidimensional Lao Benchmark for Large Language Models](https://mlsanity.com/paper/2511.11334/) (2025)
- [PsyScore: A Psychometrically-Aware Framework for Trait-Adaptive Essay Scoring and ZPD-Scaffolded Feedback](https://mlsanity.com/paper/2606.20287/) (2026)
- [SteuerLLM: Local specialized large language model for German tax law analysis](https://mlsanity.com/paper/2602.11081/) (2026)
- [Assessing the Pedagogical Readiness of Large Language Models as AI Tutors in Low-Resource Contexts: A Case Study of Nepal's K-10 Curriculum](https://mlsanity.com/paper/2604.09619/) (2026)

## Abstract (arXiv)

When evaluating language models on human exams, benchmarks typically score each response as right or wrong and report the overall accuracy. This approach assumes that partial knowledge is worth proportional credit, an assumption that fails when an examination uses a non-additive grading scheme. The 2025 reform of Vietnam's National High School Graduation Examination demonstrates the cost of this substitution. In Part II of the exam, candidates evaluate four true/false statements per question. The grading is convex: the number of correct statements earns 0, 0.10, 0.25, 0.50, or 1.00 points. Identifying three statements correctly pays 0.50 points, not the 0.75 points that standard accuracy metrics would award. Because Part II accounts for 4.00 of the exam's 10.00 points, reporting accuracy inflates the score by rewarding partial knowledge that the state explicitly penalizes. We introduce THPT-Ladder, a benchmark of 632 items from 21 official exams across 11 subjects, graded exactly as the ministry grades its students. The ministry publishes the marks of over a million candidates, allowing us to place models directly into the human cohort. Across eight models, the official rubric pays 0.020 to 0.159 points less per Part II question than proportional credit. This shortfall changes a model's apparent competence. For Qwen3.5-27B on the 2025 History exam, a 0.042-point shortfall drops its standing from the 90th to the 77th percentile among 481,293 candidates. A model's accuracy does not predict this penalty. At Claude Sonnet 5's accuracy level, different distributions of errors yield scores varying from 0.869 to 0.932 points per question. Official marks depend on how correct statements are grouped, meaning standard benchmarks report a competence the institution would not certify.