This is the post-RL model.
We first fine-tuned OpenVLA-OFT on Libero10 using one trajectory per task and then applied SimpleVLA-RL for online reinforcement learning, resulting in the final model.
This is the post-RL model.
We first fine-tuned OpenVLA-OFT on Libero10 using one trajectory per task and then applied SimpleVLA-RL for online reinforcement learning, resulting in the final model.