Skip to content

Dataset · tencent

Tencent Data for Hy-VLA Robotics Model

tencent/Hy-Embodied-0.5-VLA-Data

Tencent Robotics X dataset for training and RL fine-tuning of Vision-Language-Action models.

What's inside

The dataset accompanies the end-to-end robot learning system Hy-Embodied-0.5-VLA from the Tencent Robotics X and Tencent Hy teams. The card specifies tabular and text modalities (volume less than 1 thousand records) intended for pretraining, SFT, and RL post-training stages.

What you can do with it

  • Training and fine-tuning (SFT and RL) Vision-Language-Action models
  • Experiments with flow-matching-based robot control architectures
Tasks
Robotics, Reinforcement learning
Size
up to 1K rows
Downloads this month
55K
Likes
33

How to load it

from datasets import load_dataset

ds = load_dataset("tencent/Hy-Embodied-0.5-VLA-Data")