Papers for

language model trainers

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

Large language model tutors show varied fairness across student groups

EduFair-Bench: Evaluating Pedagogical Fairness of LLM Tutors Across Student Demographics

Abstract: Large language models (LLMs) are increasingly deployed as tutors, but it is unclear whether they support all students equally well. We introduce \textbf{EduFair-Bench}, a benchmark for auditing the pedagogical fairness of LLM tutors---whether tutoring quality varies systematically with student demographics. EduFair-Bench pairs a multi-domain question bank (mathematics, physics, chemistry) with a controlled simulation in which a fixed LLM student interacts with each tutor across nine demographic levels spanning four dimensions: gender, immigration background, first language, and socioeconomic status (SES). Tutoring quality is scored on five turn-level pedagogical metrics and four conversation-level dimensions, using an LLM judge validated against three-annotator consensus on 180 tutor turns. Bias is measured via paired Wilcoxon signed-rank tests and bootstrap effect-size confidence intervals. Two ablations (demographic cues conveyed through names; conflicting demographic information between tutor and student) disentangle tutor-driven from student-driven bias. Across five tutors, we find that model capability and demographic fairness are largely orthogonal: the smallest model is the most consistent while the four more capable tutors all exhibit wide demographic gaps with no clear capability-to-fairness ordering, pedagogy-specific RL training redistributes rather than removes bias, and language- and immigration-related cues produce larger gaps than gender- and SES-related cues.

Fri 11 SeptArtificial Intelligence
The gist
This work studies how equally large language model tutors help students from different backgrounds. The authors create a testing system that lets a simulated student interact with tutors pretending to be from diverse groups. They find that better model skill doesn’t always mean fairer tutoring, and some types of demographic information create bigger differences in help given. Their method measures fairness using detailed conversation ratings and statistical tests.
Open 2609.12949v1

Multi dimensional educational data scoring improves language model training

Edu-QuRating: Multi-Dimensional Educational Data Curation with Distilled Pairwise Judgements

Abstract: Educational data filters have become a practical way to improve language-model pre-training, but most filters treat educational value as a single scalar property. This may be too broad for some applications, especially if the data set already features a high density of educational material. Useful learning material needs to be accurate, engaging, well structured, and appropriate for the intended audience and application (e.g. learner- vs teacher-facing). Following QuRating (Wettig et al. 2024), we introduce Edu-QuRating: a pipeline for multi-dimensional educational data scoring and curation. Edu-QuRating defines education-specific rubrics, uses an LLM judge to label sampled document pairs and distills those pairwise preferences into reusable Edu-QuRaters, which can score individual text chunks on a set of educational criteria. Across two sequence-classification base models and six educational criteria, the best Edu-QuRater recovers held-out GPT-4.1-mini pairwise judgements with mean accuracy 0.917. We then apply the resulting scorers in two applications. First, we investigate the potential of Edu-QuRaters for corpus filtering to improve pretraining of small language models. We scored 322.25M FineWeb-Edu-Fortified documents to obtain a filtered pre-training mixture. In matched single-run pre-training comparisons, models trained with Edu-QuRating-based mixtures reached higher observed aggregate accuracy across nine benchmarks than the FineWeb-Edu baseline, with gains concentrated in particular tasks. Second, we used Edu-QuRater scores as reward terms for GRPO post-training. In held-out pairwise judge evaluations, combining Edu-QuRater and answer-structure rewards produced responses preferred to the Qwen3-4B base model on both pedagogical quality and instruction following.

Tue 8 SeptComputation and LanguageArtificial Intelligence
The gist
Many language models learn better when trained on good educational content, but not all learning materials are helpful in the same way. The authors created a tool called Edu-QuRating that checks educational texts on several qualities like accuracy and engagement, rather than just one score. This helps pick better study materials for training small language models, making them perform better on some tests. They also used these scores to improve models that give educational answers, making them easier to follow and more helpful.
Open 2609.09425v1