In our initial ToyBlocks tests, ManiFlow, π0.5, & MolmoAct 2 generalized poorly to new spatial configurations.
Controlled experiments let us test changes to training & model design, then measure which techniques improve generalization.
Part of a thread · earlier post
In our initial ToyBlocks tests, ManiFlow, π0.5, & MolmoAct 2 generalized poorly to new spatial configurations.
Controlled experiments let us test changes to training & model design, then measure which techniques improve generalization.
Ai2 · x.comPublished · added here
Original title: In our initial ToyBlocks tests, ManiFlow, π0.5, & MolmoAct 2 generalized poorly to new spatial configurations.