search

droid-dataset / droid_policy_learning

DROID Policy Learning and Evaluation
MIT License
140 stars 13 forks source link

An end-to-end visual language diffusion policy #8

Closed COST-97 closed 6 months ago

COST-97 commented 6 months ago

Hello: Outstanding work! On the data set DROID and the six downstream task data, have you tried to merge these data together and only train an end-to-end visual language diffusion policy? Then examine the generalization ability of this VLA model? Thank you so much!

kpertsch commented 6 months ago

Thanks for your question! We have so far only trained single-task policies with the 6 fine-tuning datasets. We are working on training multi-task & multi-scene policies -- stay tuned :)