Given the impressive capabilities demonstrated by pre-trained foundation models, we must now grapple with how to harness these capabilities towards useful tasks. Since many such tasks are hard to specify programmatically, researchers have turned towards a different paradigm: fine-tuning from human feedback. The MineRL BASALT competition aims to spur research on this important class of techniques, in the domain of the popular video game Minecraft.The competition consists of a suite of four tasks with hard-to-specify reward functions.We define these tasks by a paragraph of natural language: for example, "create a waterfall and take a scenic picture of it", with additional clarifying details. Participants train a separate agent for each task, using any method they want; we expect participants will choose to fine-tune the provided pre-trained models. Agents are then evaluated by humans who have read the task description. To help participants get started, we provide a dataset of human demonstrations of the four tasks, as well as an imitation learning baseline that leverages these demonstrations.We believe this competition will improve our ability to build AI systems that do what their designers intend them to do, even when intent cannot be easily formalized. This achievement will allow AI to solve more tasks, enable more effective regulation of AI systems, and make progress on the AI alignment problem.
Tue 3:00 a.m. - 3:20 a.m.
|
Competition results and highlights
(
Presentation
)
A quick reminder on what BASALT competition was about, followed by the preliminary competition results and highlights from the evaluation so far (e.g., surprising decision human evaluators take). |
🔗 |
Tue 3:20 a.m. - 4:45 a.m.
|
Solutions for solving BASALT's fuzzy tasks by participants
(
Presentation
)
Join to learn what methods were successful, effective and/or simple from the submission developers themselves. The invited participants will be presenting their solutions in 5-10min slots. |
🔗 |
Tue 4:45 a.m. - 5:00 a.m.
|
Break
|
🔗 |
Tue 5:00 a.m. - 5:30 a.m.
|
Retrospective reminiscing with organizers and participants
(
Panel discussion
)
With results and methods gone over, organizers and participants join together to discuss what went well, what went less well, what was unexpected and freeform from participants on what it was like to compete. |
🔗 |
Tue 5:30 a.m. - 6:00 a.m.
|
Prospective envisioning with advisors
(
Panel discussion
)
Instead of dwelling on the past, our advisors and invited speakers will join together to ponder the impact of the BASALT competition, as well as what might lay ahead of us in the future editions. |
🔗 |
-
|
Fifteen-minute Competition Overview Video
(
Overview
)
SlidesLive Video » |
Byron Galbraith · Anssi Kanervisto · Steven Wang · Stephanie Milani · Sharada Mohanty · Rohin Shah · Karolis Ramanauskas · Brandon Houghton 🔗 |