Cocktail

AI & ML interests

Cocktail, a comprehensive benchmark designed to evaluate Information Retrieval (IR) models amidst the evolving landscape of AI-generated content (AIGC). In an era dominated by Large Language Models (LLMs), the traditional IR corpus, previously solely composed of human-written texts, has expanded to include a significant proportion of LLM-generated content. Cocktail emerges as a valuable resource to response to this transformation, aiming to provide a robust framework for assessing the performance and bias of IR models in handling mixed corpora in this LLM era.

Recent Activity

KID-22  authored a paper about 2 months ago

DR-Venus: Towards Frontier Edge-Scale Deep Research Agents with Only 10K Open Data

Yuqi-Zhou  authored a paper about 2 months ago

Length-Induced Embedding Collapse in Transformer-based Models

Yuqi-Zhou  authored a paper about 2 months ago

Source Echo Chamber: Exploring the Escalation of Source Bias in User, Data, and Recommender System Feedback Loop

View all activity

Team members 2