LESS: Selecting Influential Data for Targeted Instruction Tuning (2024 · 178 citations)
Instruction tuning usually merges many heterogeneous datasets, but real deployments often need only a narrow set of skills.
Fine-Tuning Language Models with Just Forward Passes (2023 · 153 citations)
Fine-tuning large language models with backpropagation is memory-hungry. Activations, gradients and Adam's optimizer state can need roughly an order of magnitude more GPU memory than inference alone. The authors revisit zeroth-order optimization, which estimates gradients from loss differences rather than backward…
CharXiv: Charting Gaps in Realistic Chart Understanding in Multimodal LLMs (2024 · 119 citations)
Benchmarks commonly used to evaluate chart understanding in multimodal LLMs, such as FigureQA, DVQA and ChartQA, rely on synthetic or visually uniform charts and on template-generated questions.

All works

9 works of theirs in our library, the most cited first; 3 with a page here, the others link their original until we summarise them.

LESS: Selecting Influential Data for Targeted Instruction TuningarXiv:2402.04333Mengzhou Xia, Sadhika Malladi, Suchin Gururangan, et al.2024 · 178 citations
Fine-Tuning Language Models with Just Forward PassesConference on Neural Information Processing Systems (NeurIPS)Sadhika Malladi, Tianyu Gao, Eshaan Nichani, et al.2023 · 153 citations
CharXiv: Charting Gaps in Realistic Chart Understanding in Multimodal LLMsConference on Neural Information Processing Systems (NeurIPS)Zirui Wang, Mengzhou Xia, Luxi He, et al.2024 · 119 citations
MUSE: Machine Unlearning Six-Way Evaluation for Language Models ↗arXiv (Cornell University) · no summary yetWeijia Shi et al.2024 · 65 citations
Preference Learning Algorithms Do Not Learn Preference Rankings ↗neural information processing systems · no summary yetAngelica Chen et al.2024 · 26 citations
A Kernel-Based View of Language Model Fine-Tuning ↗arXiv (Cornell University) · no summary yetSadhika Malladi et al.2022 · 22 citations
Unintentional Unalignment: Likelihood Displacement in Direct Preference Optimization ↗arXiv (Cornell University) · no summary yetNoam Razin et al.2024 · 14 citations
Metadata Conditioning Accelerates Language Model Pre-training ↗Qeios · no summary yetTianyu Gao et al.2025 · 9 citations
Overtrained Language Models Are Harder to Fine-Tune ↗arXiv (Cornell University) · no summary yetJacob M. Springer et al.2025 · 8 citations

Co-authors here

12 co-authors with a page, the most papers together first.

Where their works here were published, the most works first, with the editions of those that have a page.

Papers of our library cite their works 470 times (of the 149,439 papers whose bibliographies we read); 110 citing papers with a page here →.

Updated 2 November 2026