Parikshit Ram

Principal Researcher@International Business Machines·美国·OpenReview

研究方向

Hyperdimensional Computing · Compositional Generalization · Meta-learning · Bilevel Optimization · Automated Machine Learning · Large scale learning · Decision trees · Density estimation · Clustering · Nearest neighbour search

8.2

Invariance Makes LLM Unlearning Resilient Even to Unanticipated Downstream Fine-Tuning

ICML 2025Poster

5.3

Learning positional encodings in transformers depends on initialization

ICLR 2025Rejected

5.3

Deep Clustering with Associative Memories

ICLR 2025Rejected

通讯

6.5

WAGLE: Strategic Weight Attribution for Effective and Modular Unlearning in Large Language Models

NeurIPS 2024Poster

6.0

Dense Associative Memory Through the Lens of Random Features

合作者 (20)

Parikshit Ram

Dense Associative Memory with Epanechnikov Energy

Transformers Learn Faster with Semantic Focus

On Learning Representations for Tabular Dataset Distillation

Mitigating Forgetting in LLM Supervised Fine-Tuning and Preference Learning

Invariance Makes LLM Unlearning Resilient Even to Unanticipated Downstream Fine-Tuning

Learning positional encodings in transformers depends on initialization

Deep Clustering with Associative Memories

WAGLE: Strategic Weight Attribution for Effective and Modular Unlearning in Large Language Models

Dense Associative Memory Through the Lens of Random Features