6 papers
Localizing Safety Alignment: MLP Layers and Mid-Network Blocks Encode Refusal Behavior in Large Language Models
Mingyu Zong, Sampad Mohanty, Bhaskar Krishnamachari
Safety alignment in large language models is often treated as a distributed property of the entire network, yet its practical brittleness suggests that refusal behavior may be conc…
On the Convergence of Adam, Revisited
Steven Heilman, Sampad Mohanty
We show that projected Adam for online optimization with arbitrary moment decay parameters can have average regret bounded away from zero. A similar result of R…
MLM: Multi-Layer Moire -- A Python Package for Generating Commensurate Supercells of Twisted Multilayer Two-Dimensional Materials
Anikeya Aditya, Sampad Mohanty
Moire superlattices formed by stacking atomically thin two-dimensional materials with a relative twist angle have emerged as a versatile platform for engineering quantum electronic…
SATORIS-N: Spectral Analysis based Traffic Observation Recovery via Informed Subspaces and Nuclear-norm minimization
Sampad Mohanty, Bhaskar Krishnamachari
Traffic-density matrices from different days exhibit both low rank and stable correlations in their singular-vector subspaces. Leveraging this, we introduce SATORIS-N, a framework…
LEAP -- Live Experiments for Active Pedagogy
Sumedh Karajagi, Sampad Bhusan Mohanty, Bhaskar Krishnamachari
Interactive computational environments can help students explore algorithmic concepts through collaborative hands-on experimentation. However, static and instructor controlled demo…
DISPATCH -- Decentralized Informed Spatial Planning and Assignment of Tasks for Cooperative Heterogeneous Agents
Yao Liu, Sampad Mohanty, Elizabeth Ondula +1
Spatial task allocation in systems such as multi-robot delivery or ride-sharing requires balancing efficiency with fair service across tasks. Greedy assignment policies that match…