AgiBot releases GE-Act 2.0 world-action model
AgiBot Unveils GE-Act 2.0 Native World-Action Model With 100x Data Scaling
Why we picked it
The 100x data scaling demonstrates significant progress in embodied AI with practical applications in robotics.
AgiBot released GE-Act 2.0, a native world-action model pretrained from random initialization on embodied data. The model scaled from 300 to 30,000 hours of training data. This scaling unlocked zero-shot skills such as towel folding across two robot embodiments.
Original · pandaily
AgiBot Unveils GE-Act 2.0 Native World-Action Model With 100x Data Scaling
AgiBot released GE-Act 2.0, a native world-action model pretrained from random initialization on embodied data. Scaling from 300 to 30,000 hours unlocked zero-shot skills such as towel folding across two robot embodiments.
