Kaustubh Ponkshe

12 papers Journal 10Unranked 2
YearRankTypeTitle / Venue / Authors
2026 J jnl
Fed-SB: A Silver Bullet for Extreme Communication Efficiency and Performance in (Private) Federated LoRA Fine-Tuning.
Trans. Mach. Learn. Res.
Raghav Singhal, Kaustubh Ponkshe, Rohit Vartak, Lav R. Varshney, Praneeth Vepakomma
2025 J jnl
A Lightweight Method to Disrupt Memorized Sequences in LLM.
CoRR
Parjanya Prashant, Kaustubh Ponkshe, Babak Salimi
2025 J jnl
ABBA: Highly Expressive Hadamard Product Adaptation for Large Language Models.
CoRR
Raghav Singhal, Kaustubh Ponkshe, Rohit Vartak, Praneeth Vepakomma
2025 J jnl
Fed-SB: A Silver Bullet for Extreme Communication Efficiency and Performance in (Private) Federated LoRA Fine-Tuning.
CoRR
Raghav Singhal, Kaustubh Ponkshe, Rohit Vartak, Lav R. Varshney, Praneeth Vepakomma
2025 — conf
FedEx-LoRA: Exact Aggregation for Federated and Efficient Fine-Tuning of Large Language Models.
ACL (1)
Raghav Singhal, Kaustubh Ponkshe, Praneeth Vepakomma
2025 — conf
GuideQ: Framework for Guided Questioning for progressive informational collection and classification.
NAACL (Findings)
Priya Mishra, Suraj Racha, Kaustubh Ponkshe, Adit Akarsh, Ganesh Ramakrishnan
2025 J jnl
Power Mechanism: Private Tabular Representation Release for Model Agnostic Consumption.
CoRR
Praneeth Vepakomma, Kaustubh Ponkshe
2025 J jnl
Safety Subspaces are Not Distinct: A Fine-Tuning Case Study.
CoRR
Kaustubh Ponkshe, Shaan Shah, Raghav Singhal, Praneeth Vepakomma
2024 J jnl
Exact Aggregation for Federated and Efficient Fine-Tuning of Foundation Models.
CoRR
Raghav Singhal, Kaustubh Ponkshe, Praneeth Vepakomma
2024 J jnl
GUIDEQ: Framework for Guided Questioning for progressive informational collection and classification.
CoRR
Priya Mishra, Suraj Racha, Kaustubh Ponkshe, Adit Akarsh, Ganesh Ramakrishnan
2024 J jnl
Initialization using Update Approximation is a Silver Bullet for Extremely Efficient Low-Rank Fine-Tuning.
CoRR
Kaustubh Ponkshe, Raghav Singhal, Eduard Gorbunov, Alexey Tumanov, Samuel Horváth, Praneeth Vepakomma
2024 J jnl
StructFormer: Document Structure-based Masked Attention and its Impact on Language Model Pre-Training.
CoRR
Kaustubh Ponkshe, Venkatapathy Subramanian, Natwar Modani, Ganesh Ramakrishnan