Xiaomi-Robotics-1
Xiaomithe DiT generates the action chunk via flow-matching [49]
1 more from this model
train the diffusion transformer (DiT) from scratch on the entire robot trajectory data with a flow-matching loss
general family · not yet filed
Flow matching is a training objective that represents an action distribution by learning a vector field that denoises actions.1 At each step, action tokens take the partially denoised actions from the previous step as input and predict the flow matching vector field toward the clean actions.2,3
Trade-off It gives a more expressive representation of continuous action chunks than discrete tokenization.4
Summarised by glm-5.3-flash from 2 sources; every sentence is checked against the quote it cites.
represents the action distribution via flow matching— π0.5\pi_{0.5}: a Vision-Language-Action Model with Open-World Generalization
the tokens corresponding to actions receive the partially denoised actions from the previous step of flow matching as input, and output the flow matching vector field— π0.5\pi_{0.5}: a Vision-Language-Action Model with Open-World Generalization
is trained to output the vector field from the VLM features and the noisy actions— SmolVLA: A vision-language-action model for affordable and efficient robotics
providing a more expressive representation over continuous-valued action chunks— π0.5\pi_{0.5}: a Vision-Language-Action Model with Open-World Generalization
Also called Flow matching training objective, flow matching objective, flow-matching loss, Flow-matching action generation, Flow-matching trajectory decoding, Flow matching for action generation.
10 quotes from 8 sources, one card per model, each opening with its strongest.
the DiT generates the action chunk via flow-matching [49]
train the diffusion transformer (DiT) from scratch on the entire robot trajectory data with a flow-matching loss
This stage jointly trains with next-token prediction, to preserve text prediction capabilities, and flow matching for the action expert
Training objective: flow matching
is trained with the flow matching objective
flow matching (our default), and a standard regression L1 loss
Training objective: flow matching
We fine-tune our VLM together with a flow-matching ... action generation model
GR00T N1 uses flow-matching (Lipman et al., ) to learn action generation.
During inference, we set , i.e., 5 steps, in flow matching to reduce latency with negligible performance degradation.