🙋‍♀️ Make your first submission with ease: Task 1 and Task 2 FAQs

Task 1: Dialogue Response Generation - Elevating Conversational AI

This task is an invitation for you to advance conversational AI by crafting responses that transcend standard interactions, aiming to improve the response quality and user engagement.

Possess a Preliminary Chatbot Model? If your current model is adept at generating basic responses, you’re on the right trajectory. We invite you to refine it in accordance with our submission standards.

Is Your AI Capable of Persona-Grounded Dialogue? Our challenge calls for models that account for the dialogue history and personas. If your model is equipped for this, we encourage you to adapt it to our specified format and submit.

Aspiring to Integrate Commonsense Knowledge? We are looking for models that seamlessly integrate personal insights with a broader commonsense understanding. If your model is not there yet, consider our baseline model as a foundation. Your innovative solutions could pave the way for more dynamic and engaging AI conversations.

Embarking on Dialogue Generation? For those new to this domain, we suggest exploring our resource section or investigating foundational texts in open-domain dialogue generation. Start with a foundational model and progressively enhance its capabilities.

Task 2: Knowledge Linking - Advancing AI’s Conversational Contextualization

This task involves training AI to adeptly correlate dialogue with pertinent real-world knowledge, thereby enriching the informational and contextual quality of conversations.

Focusing on Binary Classification? If your model is proficient in discerning relevant from irrelevant information, this task is suitable for your research. Utilize our baseline model’s training dataset as a starting point.

Expertise in Knowledge Graphs? We encourage the development of models that can effectively associate knowledge segments with dialogue elements. We have prepared a set of potential knowledge-dialogue pairs for your AI to analyze and link appropriately.

Initiating Research in Knowledge Linking? For newcomers in this field, our “Baseline” section and the ComFact paper serve as comprehensive introductory resources. This task offers an excellent avenue to delve into the intricate interplay between AI and knowledge graphs.

View post 0

💬 Feedback & Suggestions

View post 0

👥 Looking for teammates?

View post 0

✨ Welcome to Commonsense Persona-Grounded Dialogue Challenge (CPDC 2023)

View post 0

✨ Welcome to Commonsense Persona-Grounded Dialogue Challenge (CPDC 2023)

Introducing the Commonsense Persona-grounded Dialogue Challenge (CPDC 2023)

We are thrilled to unveil the CPDC 2023, a collaboration among leading researchers at Sony and EPFL. With the rapid advancements in dialogue systems and the capabilities of Transformers and Large Language Models (LLM), the CPDC 2023 competition beckons you to the cutting edge of conversational AI.

Harness the newly released PeaCoK knowledge graph, delve into the depths of persona-grounded dialogues, and compete on two distinct tracks: Commonsense Dialogue Response Generation (Task1) and Commonsense Persona Knowledge Linking (Task2).

image787×424 27.5 KB

image774×411 28.6 KB

About the Challenge

The goal of this contest is to drive research into truly understanding natural conversation flow, focusing on the incorporation of commonsense persona-grounded knowledge. This competition uniquely brings forth meticulously crafted high-quality human-human dialogues to rigorously assess current state-of-the-art dialogue systems.

The Tasks:

1. Commonsense Dialogue Response Generation (Task1):
Participants are tasked with submitting dialogue response generation systems, with evaluations based on the persona-grounded dialogue dataset. Embrace the challenge! Baseline model

2. Commonsense Persona Knowledge Linking (Task2):
This track challenges you to develop systems that link knowledge to a dialogue, checking the veracity of persona-grounded knowledge. Ready to step up? Baseline model

Timeline:

• Warm-up Round: November 1 - 30, 2023
• Round 1: December 1, 2023 - February 29, 2024
• Round 2: March 1 - 5, 2024
• Challenge Conclusion: March 15, 2024

Prizes:

An impressive total of $35,000 awaits the victors! Dive into the details below:

Task1 Prizes:

• $15,000
• $7,000
• $3,000

Task2 Prizes:

• $5,000
• $3,000
• $2,000

Full Challenge Rules and Open Sourcing Criteria

Join the Conversation:

Have feedback, questions, or are on the lookout for teammates? Visit the AIcrowd Community. Engage with peers on the Discord Channel for collaborations, discussions, and direct interactions with organizers.

All the best
Team CPDC 2023

View post 0

👥 Looking for teammates?

Competing is more fun with a team!

Introduce yourself here, and find others who are looking to team up!

Format:

• A short introduction about you and your background.
• What brings you to this challenge?
• Some ideas you wish to explore as a part of this challenge?

Cheers,
Team CPDC

View post 0

About the Commonsense Persona-Grounded Dialogue Chall-1f6f43 category

View post 0

💬 Feedback & Suggestions

We are constantly trying to improve this challenge for you and would appreciate any feedback you might have!

Please reply to this thread with your suggestions and feedback on making the challenge better for you!

• What have been your major pain points so far?
• What would you like to see improved?

Cheers,
Team CPDC

View post 0

👾 Welcome to NeurIPS 2023: The Neural MMO Challenge

Announcing the Neural MMO Challenge at NeurIPS 2023!

Join us for the largest RL competition at NeurIPS 2023! With NMMO 2.0, we are unveiling faster performance, an added emphasis on task completion, and a novel RL baseline. This event invites participants to engage deeply in the world of reinforcement learning, challenging them with tasks and opponents they’ve never encountered.

Dive into the vast universe of Neural MMO and test your skills against unseen challenges and competitors.

Your Challenge

The crux of this competition is to empower the next generation of AI to tackle unprecedented tasks against novel adversaries in unexplored terrains.

Competition Tracks

• Reinforcement Learning: Refine the RL algorithm, model, and reward dynamics.
• Curriculum Generation: Craft the task generator, sampler, and rewards using Python.
• No Holds Barred Track: Show us what you’ve got! No boundaries – utilize any method, but don’t try hacking our systems!
  • LLM Agents: Employ GPT or LLMs to create scripted agents. Evaluation support is underway; refer to Documentation/Discord for more details.

Timeline

• October: Kick-off - Warm-up rounds against baselines.
• November-December: The main competition heats up! Evaluate task completion against peers. Competition tasks intensify over rounds.
• TBA before NeurIPS: Submission deadline, final evaluation for top 16 in each track.
• NeurIPS: Champions declared!

Prizes
Boasting a total of $20K in rewards sponsored by Parametrix.ai. A detailed breakdown for each track and round will be shared soon. Standout performers in the Reinforcement Learning and Curriculum Generation tracks must open-source their submission code. Though optional for the No Holds Barred track, it’s highly encouraged.

Rules
To ensure fairness and integrity, we’ve established guidelines for each track. Please ensure you’re well-acquainted with the rules relevant to your track. Read the detailed rules here.

Voice your thoughts and feedback right here.
Challenges become memorable with allies! Find teammates over here.
Engage with fellow competitors on our Discord channel or WeChat.
Best of luck to all participants!

content_poster_updated533×800 310 KB

View post 0

💬 Feedback & Suggestions

We are constantly trying to improve this challenge for you and would appreciate any feedback you might have!

Please reply to this thread with your suggestions and feedback on making the challenge better for you!

• What have been your major pain points so far?
• What would you like to see improved?

Cheers,
Team NMMO 2023

View post 0

👥 Looking for teammates?

View post 0

👥 Looking for teammates?

Competing is more fun with a team!

Introduce yourself here, and find others who are looking to team up!

Format:

• A short introduction about you and your background.
• What brings you to this challenge?
• Some ideas you wish to explore as a part of this challenge?

Cheers,
Team NMMO 2023

View post 0

About the The Neural MMO Challenge 2023 category

View post 0

Update on Challenge Timeline and Important Environment Update

View post 0

Update on Challenge Timeline and Important Environment Update

Hello Participants,

We’re reaching out to inform you of a few crucial updates regarding the challenge timeline:

1. Phase 2 will now conclude on November 15th.
2. Phase 3 is set to commence immediately after, from November 16th to November 30th.

Please make a note of these changes and adjust your strategies accordingly. We have updated the timeline on the challenge page to reflect these changes.

Additionally, we’d like to highlight the importance of updating your environment. Kindly reference this post to ensure that you have the latest version, CityLearn 2.1b12, which fixes some crucial bugs.

We hope these updates help you in your journey. Feel free to reach out with any queries.

Warm regards,
Team CityLearn 2023

View post 0

Scoring Announcement: Public vs. Private

This response is specifically for the MosquitoAlert Challenge 2023.

1. The three submissions are designated per team, not per individual. Teams will be provided with a form to select submissions after the challenge concludes.
2. The scores for the entire private dataset are pre-calculated, so there’s no concern about potential failures.

View post 1

Scoring Announcement: Public vs. Private

Hello Participants,

We’d like to provide clarity regarding the scoring mechanism in the challenges.

While all submissions run on the public & private dataset, only the public scores are displayed while challenge is live. However, as the competition concludes, you will have the opportunity to designate your top three submissions. Only the private scores of these selected submissions will be considered in calculating your final leaderboard position.

Furthermore, all private scores for all your submissions will be disclosed at the conclusion of the challenge.

Thank you for your participation, and we look forward to your final standings!

All the best,
Team AIcrowd

View post 0

💬 Join Office Hours: 15 October 4:30 PM UTC

Hello everyone,

We are thrilled to announce an exclusive series of office hours with Challenge Organisers, offering an interactive session to discuss the aspects of the MeltingPot Challenge! Here’s your opportunity to dive deeper into the nuances of environments, baselines and evaluation protocols, and have your queries answered. Following are the details of the first office hours:

Dates: October 15th, 2023 (Sunday)

Time: 4:30 PM - 6:00 PM UTC (60 mins public discussions + 30 mins private queries)

Where: Office HourVoice Channel on Discord

What to Expect: Live Q&A space with discussions focussed on:

• Clarifications on queries about Melting Pot suite.
• Baselines, evaluation protocol and potential customisations.
• Clarifications on queries about potential solution strategies.

Mark your calendars and ensure you make the most out of these insightful sessions! Be sure to join the MeltingPot Discord Channel to stay updated with all upcoming activities and communications.

Stay curious,
Team MeltingPot

View post 0

🏛️ Join the Live Townhall Event for MeltingPot Challenge

Hello MeltingPot Community,

The electrifying MeltingPot Challenge 2023 is already off to an invigorating start, and to enrich this collaborative journey, we’re excited to bring to you the MeltingPot 2023 Townhall Event!

About the Challenge

Dive deep into multi-agent reinforcement learning and navigate the world of mixed-motive cooperation. Organised by renowned researchers from the Cooperative AI Foundation, MIT, and Google DeepMind, this contest beckons researchers to redefine the horizons of MARL and successful cooperative intelligence.

Townhall Insights

When: 7th October, 2023
Time: 12:30 PM UTC
Join: Townhall Zoom Link

Dive deeper into strategy, refinement, and innovation with our expert panel and engage in vibrant discussions, gaining novel insights and formulating your award-winning strategy.

What’s in Store

• Exclusive Insights: Engage and interact with our experienced panel, seeking answers to your most pressing queries directly from the organizers.
• Collaborative Discussions: Share, learn, and innovate with peers and industry leaders, driving your project toward unparalleled excellence.

Not able to join? Worry not! A recording will be available post-event, ensuring you don’t miss out on these valuable insights.

We encourage and welcome your pre-event questions! Please drop them below, and our panel will address them during the townhall.

Join the event this Saturday 12:30PM UTC onwards Townhall: MeltingPot Challenge 2023

View post 0

🚨 Important Updates for Round 2

Hi @tfriedel
Hope this post provides the clarification: Scoring Announcement: Public vs. Private
Let us know if you have any more questions!

View post 2