|
|
--- |
|
|
datasets: |
|
|
- behavior-1k/2025-challenge-demos |
|
|
--- |
|
|
|
|
|
This is the model weights of Team Comet for the [2025 BEHAVIOR Challenge](https://behavior.stanford.edu/index.html). Our [submission](https://behavior.stanford.edu/challenge/leaderboard.html#privileged-information-track) achieved a Q-score of 0.2514, securing 2nd place overall and finishing just behind the winning team by a narrow margin—highlighting both the strong competitiveness of our approach and the effectiveness of our end-to-end VLA training strategy. |
|
|
|
|
|
Please refer to [this repository](https://github.com/mli0603/openpi-comet) for detailed usage. It provides a unified framework for pre-training, post-training, data generation and evaluation of π0.5 (Pi05) models on BEHAVIOR-1K. |
|
|
|
|
|
The [tech report](https://arxiv.org/abs/2512.10071) is available now. |
|
|
|
