Bhuwan Dhingra
@bhuwandhingra
Followers
1K
Following
172
Statuses
96
Natural Language Processing / Machine Learning research. Assistant Professor @dukecompsci, @duke_nlp; Research Scientist @Apple
Durham, NC
Joined May 2014
0
0
1
New benchmark for multimodal information extraction from materials science papers!
π’ New preprint on a benchmark for multimodal information extraction! Structured data extraction from long documents consisting of interconnected data in text, tables, and figures remains a challenge. MatViX aims to fill this gap.
1
1
24
1
0
2
RT @ghazalkhn: π Excited to share that IsoBench has been accepted at @COLM_conf! IsoBench features isomorphic inputs across Math/Graph proβ¦
0
5
0
Exciting new work from Roy on membership inference for LLMs. TL;DR: conditional likelihood is a strong predictor of membership.
π¨ Breaking: >90% AUC on the WikiMIA dataset for membership inference! Want to know if your data is in LLM's training set?π Check out our latest work "ReCaLL: Membership Inference via Relative Conditional Log-Likelihoods" β¨ π§΅1/6
1
1
10
@florian_tramer @ruoyuxyz @daphneipp @ChunyuanDeng @armancohan @WeijiaShi2 @niloofar_mire @iamgroot42 @jeff_cheng_77 @alexsablay @aterzis @srxzr @kandpal_nikhil @Chris_Choquette Though your work definitely suggests we should try out our idea on more datasets. Stay tuned for that!
0
0
5
New paper on prompt extraction from LLMs by @JunlinWang3! TL;DR: GPT-4, Gemini, Llama-2 and Mixtral are all susceptible to prompt extraction -- but simple defenses can go a long way.
π¦Excited to announce our work on robustness & security of LLM systems! πππππ¨π¨π§: Prompt Extraction Benchmark of LLM-Integrated Applications Prompt extraction from LLM-integrated apps like GPT-s is a critical security concern. βΌοΈ
0
1
4
RT @ghazalkhn: π How well can large language models extract structured data from materials science literature? Accepted at #ACL2024nlp Finβ¦
0
1
0
Welcome Shuyan! Really excited to work with you soon :D
I am thrilled to announce that I will be joining @DukeU
@dukecompsci as an Assistant Professor in summer 2025. Super excited for the next chapter! Stay tuned for the launch of my lab π§ π€
1
0
11
Check out @RichStureborg 's NAACL paper on tailoring vaccine messages to common opinions for (hopefully) more effective interventions!
Can we tailor LLM responses to better establish common-ground in high-stakes communication? Happy to announce our recent #NAACL findings paper where we investigate a new perspective on ~LLM personalization~ through a task for vaccine messaging. 1/3...
0
0
5
RT @DeqingFu: Do multimodal foundation models treat every modality equally? Hint: Humans have picture superiority. How about machines? Inβ¦
0
25
0