Author

Harshvardhan Saini

Observed name: Saini, Harshvardhan

Institutions: National University of Singapore, Indian Institute of Technology Dhanbad

Observed institutions: National University of Singapore, Indian Institute of Technology Dhanbad

OpenAlex ID: A5134518221

Masking Frequent Tokens Sharpens Direct Preference Optimization

2026 · Saini, Harshvardhan, Jha, Samyak, et al. · arXiv (Cornell University)

Understanding and Exploiting Anisotropy in Post-Training

2026 · Jha, Samyak, Saini, Harshvardhan, et al. · arXiv (Cornell University)

Capsule Lens: Locating and Tracking Concept Geometry in Model Representations

2026 · Tang, Yiming, Saini, Harshvardhan, et al. · arXiv (Cornell University)

Emergent Hierarchical Monosemantic Neurons from the Group-Contrastive Forward-Forward Algorithm

2026 · Tang, Yiming, Qi, Qinglin, et al. · arXiv (Cornell University)

When Language Overwrites Vision: Over-Alignment and Geometric Debiasing in Vision-Language Models

2026 · Saini, Harshvardhan, Jha, Samyak, et al. · arXiv (Cornell University)

Bridging Mechanistic Interpretability and Prompt Engineering with Gradient Ascent for Interpretable Persona Control

2026 · Saini, Harshvardhan, Tang, Yiming, et al. · arXiv (Cornell University)

Works count: 6

Citation count:

H-index:

I10-index: