This repository contains the Hugging Face checkpoint used by
Xiaomi-Robotics-1 for RoboCasa evaluation. It includes the model weights,
custom Transformers model and processor code, tokenizer files, and
RoboCasa action normalization statistics.
Requirements
The reference environment uses:
Python 3.11
PyTorch 2.8.0
Transformers 4.57.1
FlashAttention 2
The custom model and processor must be loaded with trust_remote_code=True.
Loading
import torch
from transformers import AutoModel, AutoProcessor
model_id = "XiaomiRobotics/Xiaomi-Robotics-1-RoboCasa"
processor = AutoProcessor.from_pretrained(
model_id,
trust_remote_code=True,
use_fast=False,
)
model = AutoModel.from_pretrained(
model_id,
trust_remote_code=True,
attn_implementation="flash_attention_2",
dtype=torch.bfloat16,
).cuda()
For reproducible multi-GPU evaluation, use the standard eight-server launcher in
the companion source repository.
RoboCasa Evaluation
Use the companion Xiaomi-Robotics-1 source repository and follow
eval_robocasa/README.md. The server launcher accepts either this Hugging
Face repository ID after downloading it locally or a local snapshot path.
Reference configuration:
tasks: 24
episodes per task: 100
crop ratio: 0.95
Reference result:
| Metric |
Value |
| Episodes |
2400 |
| Successes |
1781 |
| Episode success rate |
74.21% |
License
Apache License 2.0. See LICENSE.
Citation
@article{team2026xiaomi,
title={Xiaomi-Robotics-1: Scaling Vision-Language-Action Models with over 100K Hours of Real-World Trajectories},
author={Team, Xiaomi Robotics and Guo, Jun and Jin, Piaopiao and Li, Jason and Li, Peiyan and Li, Yingyan and Liu, Futeng and Peng, Wanli and Qin, Optimus and Su, Yifei and others},
journal={arXiv preprint arXiv:2607.15330},
year={2026}
}
