๐ŸŽฎ Welcome to Commonsense Persona-Grounded Dialogue Challenge 2025

:video_game: Introducing the Commonsense Persona-grounded Dialogue Challenge (CPDC 2025)! :video_game:

:raising_hand_woman: New to the challenge? :thinking: Want to make your first submission?
:gear: Access the Starter-Kit here.

Join this challenge to redefine interactive storytelling in video games with AI-driven, context-aware non-player characters (NPCs). Imagine playing your favourite game where every NPC remembers past interactions, adapts their responses, and creates a dynamic world of gaming experience.

Why CPDC 2025 Matters Today, NPCs often break the immersion with repetitive, disconnected dialogue. The CPDC 2025 aims to change this by fostering NPCs that can engage in realistic, nuanced conversations that reflect their memories and personalities, enhancing the gaming experience to new heights of realism.

:male_detective: About the Challenge The CPDC 2025 builds on the advancements of large language models and dialogue systems to create NPCs that are not only aware of the gameโ€™s context but also capable of task-oriented interactions. This year, weโ€™re taking a giant leap from previous competitions by integrating persona-grounded dialogue capabilities that can handle complex, multi-turn conversations based on in-game developments.

Challenge Tasks Overview

  • Task 1: Task-Oriented Dialogue Agents - Develop NPCs that can perform specific game-related tasks through dialogue.
  • Task 2: Context-Aware Dialogue Agents - Create NPCs that can conduct engaging and relevant small talk, reflecting the gameโ€™s ongoing narrative.
  • Task 3: Integrating Contextual Dialogue and Task Execution - Combine the strengths of Tasks 1 and 2 for a holistic dialogue system that enhances player immersion.

:rocket: Challenge Structure Participants will tackle three main tasks, each designed to test different facets of dialogue system capabilities within a gaming context. For those interested in pushing the boundaries, the integrated Task 3 provides a comprehensive test of handling both contextual chatting and task execution seamlessly.

:date: Key Dates and Timeline

  • Warm-up Round: 9th April 2025
  • Round 1: 20th April 2025
  • Round 2: 25th May 2025
  • Challenge End: 30th June 2025

:trophy: Prizes Galore! Dive into the competition with a total prize pool of $20,000, spread across multiple tracks:

  • Task 1: $4,000 - Focus on task-oriented dialogue.
  • Task 2: $4,000 - Master context-aware interactions.
  • Task 3: $12,000 - Excel at integrating dialogue with task execution.

:date: Challenge Timeline | :trophy: Prizes

:speech_balloon: Join the conversation on Discord โ€“ Connect with fellow participants, stay updated, and introduce yourself! | :busts_in_silhouette:Form a team | :speech_balloon: Have feedback or questions?