Reshmi Ghosh's picture

2 7 4

Reshmi Ghosh

reshmighosh

·

https://reshmighosh.github.io

AI & ML interests

LLM, Responsible AI, NLP, HCI

Recent Activity

authored a paper about 1 month ago

VLMGuard: Defending VLMs against Malicious Prompts via Unlabeled Data

authored a paper about 1 month ago

Quantifying reliance on external information over parametric knowledge during Retrieval Augmented Generation (RAG) using mechanistic analysis

liked a model about 1 month ago

withsecure/Llama3-8B-PromptInjectionHardened

View all activity

Organizations

None yet

reshmighosh's activity

commented a paper about 1 month ago

SAGEval: The frontiers of Satisfactory Agent based NLG Evaluation for reference-free open-ended text

Paper • 2411.16077 • Published Nov 25, 2024 • 1 •

commented a paper 4 months ago

VLMGuard: Defending VLMs against Malicious Prompts via Unlabeled Data

Paper • 2410.00296 • Published Oct 1, 2024 • 4 •