In this reading group, Shreyash Kale presents an interesting paper where a multi-agent interaction between two LLMs tries to elicit untoward behavior from one of the models. The paper is called Red Teaming Language Models with Language Models by Perez et al....
In this paper, Jeff Schulman presents a mind expanding paper on how knowledge graphs and agents can help identify structures that suggest fruitful investigation in that topic area. The paper is called Accelerating Scientific Discovery with Generative Knowledge...
In this paper, Sourav Panda covers one of the foundational papers in starting the scientific community’s ability to create agents that play StarCraft. The paper is called StarCraft II: A New Challenge for Reinforcement Learning, by Vinyals et al. Presentation...
In this reading group, Jeff Schulman presents a paper leveraging the active inference framework for XAI purposes, making it an important one to think about for PLAINTEXT affiliates. The paper is called Designing explainable artificial intelligence with active...
In this paper, Sourav Panda covers an interesting work combining reinforcement learning and natural language processing by providing a reward function that incorporates a variety of objectives (e.g., information, non-repetition, etc.). The paper is called Deep...
In this reading group, Jeff Schulman presents Emergent representations of program semantics in language models trained on programs by Jin and Rinard. This paper offers interesting results, showing that despite the narrow task of next-token prediction, LLMs can...
In this reading group, Jeff Schulman covers an interesting paper on how Active Inference can assist HCI researchers. The paper is called Active Inference and Human-Computer Interaction, by Murray-Smith et al. Presentation Link:...
In this paper, Sourav Panda covers a neat paper that approaches the VERY hard problem of playing StarCraft, but without the massive effort found in AlphaStar. The paper is called TStarBots: Defeating the Cheating Level Builtin AI in StarCraft II in the Full Game, by...
In this reading group, Shikha Soneji covers An NLP Analysis of ChatGPT’s Personality Simulation Capabilities and Implications for Human-centric Explainable AI Interfaces by Zylowski and Wolfel. This is an interesting paper because it shows how one can use...
In this reading group meeting, Rupak Kumar Das presents an interesting paper on the current state of the art for textual semantic similarity. The paper is called BERTScore: Evaluating Text Generation with BERT by Zhang et al. Presentation Link:...