24 upvotes
New benchmark lets shopping agents practice long-term user dialogue
What is UserArena: Benchmarking Interactive User Simulation for Long-Horizon Shopping Agents?
UserArena is a simulated shopping environment that models how users react over many steps. It lets researchers test recommendation agents that need to plan across long interactions.
Who it is for
If you build or evaluate interactive recommender systems, you can use UserArena to measure how well your agent pursues shopper goals.
How to try it
Open the Hugging Face repo at https://huggingface.co/datasets/UserArena