Ever Wondered If Chatbots Truly Understand You? The journey of perfecting conversational AI is a thrilling one. The Commonsense Persona-grounded Dialogue (CPD) Challenge 2023 is your chance to delve deep into the intricacies of LLM, and perhaps, change the future of AI dialogues!
- Participate In CPD Challenge:
https://www.aicrowd.com/challenges/commonsense-persona-grounded-dialogue...
- Total Prizes:
$35,000 Prize pool
- Challenge Deadline:
March 1, 2024
- Join the Dialogue Evolution:
The quest for truly intuitive and context-aware dialogue systems has been a long one. While we've made impressive strides with Transformers and Large Language Models, the path to perfection is still being paved. The CPD Challenge 2023 offers you a distinct platform to explore, innovate and redefine the boundaries of conversational AI.
Delve into the high-quality human-human dialogues and a newly curated Commonsense Persona-grounded Dialogue dataset, aimed at evaluating models that can truly simulate and generate genuine conversation.
- Unveiling the Unique Dataset:
Grounded in commonsense persona knowledge, challenge dataset stands unparalleled in the arena of conversational AI. Notably, the PeaCoK: Persona Commonsense Knowledge for Consistent and Engaging Narratives (https://aclanthology.org/2023.acl-long.362/) has been honored with the ACL2023 Outstanding Paper Award.
- Challenge Tasks:
This challenge features two unique tasks:
1. Commonsense Dialogue Response Generation: Challenge your model to generate responses that harmonize with a persona. Gauge its capabilities on the specially prepared persona-grounded dialogue dataset. And if you need a starting point, we've got a baseline model (https://github.com/Silin159/PersonaChat-BART-PeaCoK) ready for you. Challenge your model to generate responses that harmonize with a persona.
2. Commonsense Persona Knowledge Linking: Evaluate your model's adeptness at integrating knowledge coherently within dialogues. And to get started, we recommend checking out the baseline model (https://github.com/Silin159/ComFact-Relation-Agnostic) as well.
- Prizes For CPD Challenge
While the collective prize pool stands at a commendable 35,000 USD, the true essence lies in achieving recognition and paving the way for a groundbreaking scientific evolution in the field of LLM.
Participating teams are eligible to win prizes in multiple leaderboards spread across both the tracks.
Task1: Commonsense Dialogue Response Generation
* First place 15,000 USD
* Second place 7,000 USD
* Third place 3,000 USD
Task2: Commonsense Persona Knowledge Linking
* First place 5,000 USD
* Second place 3,000 USD
* Third place 2,000 USD
- Timeline:
The challenge will take place across in 3 Rounds which differ in the evaluation dataset used for ranking the systems.
* Warm-up Round : 2nd November, 2023
* Round 1 : 1st December, 2023
* Round 2 : 15th January, 2024
* Challenge End: 1st March, 2024
We believe that our community is brimming with talent and innovation. Whether you're a seasoned researcher, an AI hobbyist, an industry professional, or an LLM enthusiast, we are eager to see the marvels you'll achieve in this challenge!
- Make your first submission with ease by following this starter-kit: https://www.aicrowd.com/challenges/commonsense-persona-grounded-dialogue...
and join the conversation with other participants on Discord Server:
https://discord.gg/xgTsY3CbJX
Looking forward to seeing your groundbreaking solutions!
All The Best,
Sony, AIcrowd and EPFL