Abhimanyu Hans

@ahans30

cs phd candidate advised by @tomgoldsteincs at @umdcs, prev math @univofdelhi trying not taking myself so seriously ¯\_(ツ)_/¯

Washington, DC

ahans30.github.io

Joined July 2011

365 Following

262 Followers

Replies Reposts Media

Abhimanyu Hans
@ahans30
Aug 7, 2025
zoom bombing is lame guys, especially in 2025, especially in someone's PhD proposal talk totally unrelated but guess who's a PhD candidate now 👀
4
Abhimanyu Hans
@ahans30
Feb 6, 2025
the real problem is that current retrievers are not instruction-friendly that's why you google keywords likely to be found in the answers/docs than try to "prompt" your query or you know...just ask your LLM
will brown
@willcb
Feb 6, 2025
LLMs know way too many facts. you can ask Llama 1B about the history of pizza and it gives a decent answer. it shouldn't be able to do that. it should just do a google search. use those weights for something else
Abhimanyu Hans
@ahans30
Jan 31, 2025
Never more true than on conference deadlines
Abhimanyu Hans
@ahans30
Dec 9, 2024
poster sent for print 😮‍💨 are you concerned your prod LLM might regurgitate exact training data to your users? join me and my co-authors at #NeurIPS2024 on wednesday's 1st poster session & learn how goldfish loss can help you. eager to meet friends from past and future!
Abhimanyu Hans
@ahans30
Jun 17, 2024
Litigators hate this one simple loss function 📉 We present goldfish loss, an effective objective to prevent your LLM from regurgitating training data verbatim.
Abhimanyu Hans
@ahans30
Sep 26, 2024
Super happy to announce that our work has been accepted at #NeurIPS2024. tl;dr: our simple yet effective goldfish loss mitigates memorization even after training a 7B LLM for 100 epochs by masking tokens during loss computation. See you all in Vancouver in December! :)
Abhimanyu Hans
@ahans30
Jun 17, 2024
Litigators hate this one simple loss function 📉 We present goldfish loss, an effective objective to prevent your LLM from regurgitating training data verbatim.
3