yifan jiang @yifanji24618785

PhD student at ISI, USC Joined May 2023

Tweets

8
Followers

43
Following

96
Likes

11

Dong-Ho Lee @Dongho_Lee_

7 months ago

🤖 Even wonder why chatbots feel off? We unveil REALTALK: 21 days of REAL human chats showing what AI misses: emotions, shifting personas, memory gaps. arxiv.org/abs/2502.13270 github.com/danny911kr/REA… @Snap @SaharaLabsAI @USC_ISI @nlp_usc #LLM #EmotionalIntelligence

1 9 25 3K 7

Download Image

Thank you, @rohanpaul_ai, for the great summary of Red Queen! 🚀 We’ve just released our codebase at GitHub. Feel free to explore it and create your own Red Queen attacks! Git link: github.com/kriti-hippo/re…

Rohan Paul @rohanpaul_ai

a year ago

5 4 13 8K 17

Download Video

0 0 1 130 0

yifan jiang @yifanji24618785

a year ago

RED QUEEN ATTACK construct a multi-turn scenario to conceal malicious intent and mislead models! To our surprise, larger models are more susceptible under our attack😮😮😮

kriti aggarwal @0xKriti

a year ago

RED QUEEN ATTACK construct a multi-turn scenario to conceal malicious intent and mislead models! To our surprise, larger models are more susceptible under our attack😮😮😮

1 0 4 615 1

Download Image

0 0 3 129 0

yifan jiang @yifanji24618785

a year ago

How good are MLLM at solving IQ (abstract visual reasoning) problems? Check our new benchmark paper! MARVEL: Multidimensional Abstraction and Reasoning through Visual Evaluation and Learning Paper: arxiv.org/pdf/2404.13591… Website: marvel770.github.io

1 0 3 590 0

yifan jiang @yifanji24618785

2 years ago

Great pressure to present our cool work at EMNLP 2023 !!! BrainTeaser: a lateral thinking task in a multiple-choice QA format on which large language models struggle to identify puzzle solutions, distracted by surface commonsense associations.

0 0 9 457 0

Download Image

yifan jiang @yifanji24618785

2 years ago

🔊 📝 Announcing our new paper at the EMNLP 2023 main conference! 💡 TLDR： Can LLMs solve complex brain teasers? We introduce BRAINTEASER, a multiple-choice framework aimed at evaluating lateral thinking in language models. (arxiv.org/abs/2310.05057)