A new method called Role-Conditioned Sub-Token Routing (RoleSub) compresses vision-language-action (VLA) model representations, outperforming token-only pruning in 33 of 36 settings on the LIBERO suit…
Bagel Labs released WorldDiT, a world model for robotics that jointly learns action prediction and scene evolution using shared parameters, achieving strong benchmark scores on LIBERO while staying be…
Researchers introduced WorldDiT, a unified diffusion transformer architecture that couples action generation with visual world modeling and achieves strong robot control performance without a large pr…
A reproducible 100-step LoRA fine-tuning run for the OpenVLA 7-billion-parameter robotics model on a Colab A100 GPU has been documented, providing a verifiable integration test that confirms the train…
Researchers at MIT introduced the Temporal Ratio (TR), an attention-based metric that predicts how well video-action models generalize to new tasks after finetuning on robotic data. They found that TR…
Researchers introduced MIRTH, an enhanced vision-language-action (VLA) framework for robotic control that incorporates dual-scale temporal memory hubs, latent reasoning tokens, and parallel action dec…
Researchers introduced ActQuant, a sub-4-bit quantization framework for Vision-Language-Action (VLA) models that reduces computational demands for edge deployment. The method achieved 95% performance …