libr6 commited on
Commit
3521906
·
verified ·
1 Parent(s): dcd0590

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +134 -1
README.md CHANGED
@@ -1,3 +1,136 @@
1
  ---
2
- license: apache-2.0
 
 
 
 
 
 
 
 
 
 
 
 
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
+ language:
3
+ - en
4
+ tags:
5
+ - gguf
6
+ - robotics
7
+ - embodied-ai
8
+ - vision-language-action
9
+ - world-action-model
10
+ - cpp
11
+ - inference
12
+ library_name: gguf
13
+ pipeline_tag: robotics
14
+ license: other
15
  ---
16
+
17
+ # Embodied.cpp GGUF Models
18
+
19
+ This repository provides **GGUF model weights converted for use with [Embodied.cpp](https://arxiv.org/abs/2607.02501)**, a portable C++ inference runtime for embodied AI models on heterogeneous robots.
20
+
21
+ These files are **deployment-oriented conversions** of upstream checkpoints. They are intended for **inference with Embodied.cpp**, rather than for training or fine-tuning in the original Python stacks.
22
+
23
+ ## About Embodied.cpp
24
+
25
+ **Embodied.cpp** is a portable inference runtime for embodied AI models, including Vision-Language-Action (VLA) models and World-Action Models (WAMs). It focuses on low-latency closed-loop deployment across heterogeneous hardware and robot systems.
26
+
27
+ Paper:
28
+ - **Embodied.cpp: A Portable Inference Runtime of Embodied AI Models on Heterogeneous Robots**
29
+ - arXiv: https://arxiv.org/abs/2607.02501
30
+
31
+ ## What This Repository Contains
32
+
33
+ This repository hosts **GGUF-converted model artifacts** used by `embodied.cpp`.
34
+
35
+ The goal is to make upstream embodied model checkpoints easier to deploy from C++ with a unified runtime and serving interface.
36
+
37
+ ## Conversion Provenance
38
+
39
+ The GGUF files in this repository are converted from upstream checkpoints using the conversion scripts in the `embodied.cpp` project.
40
+
41
+ ### 1. pi0.5
42
+
43
+ Upstream checkpoint:
44
+ - https://huggingface.co/lerobot/pi05_libero_finetuned_v044
45
+
46
+ Conversion scripts:
47
+ - `scripts/convert_pi05_to_gguf.py`
48
+ - `scripts/convert_pi05_mmproj_to_gguf.py`
49
+
50
+ Notes:
51
+ - The converted artifacts are intended for `embodied.cpp` pi0.5 inference.
52
+ - pi0.5 uses separate GGUF exports for the main policy weights and the multimodal projector / vision-side component.
53
+
54
+ ### 2. Hunyuan-VLA-0.5
55
+
56
+ Upstream checkpoint:
57
+ - https://huggingface.co/tencent/Hy-Embodied-0.5-VLA-RoboTwin
58
+
59
+ Conversion code:
60
+ - Conversion scripts are included in the `embodied.cpp` repository.
61
+
62
+ Notes:
63
+ - These GGUF files target `embodied.cpp` deployment of HY-VLA on RoboTwin-style evaluation and related runtime paths.
64
+
65
+ ### 3. LingBot-VA
66
+
67
+ Upstream source:
68
+ - https://huggingface.co/datasets/robbyant/libero-long-lerobot
69
+
70
+ Conversion code:
71
+ - Conversion scripts are included in the `embodied.cpp` repository.
72
+
73
+ Notes:
74
+ - These GGUF artifacts are prepared for the LingBot-VA path in `embodied.cpp`.
75
+
76
+ ## Intended Use
77
+
78
+ These GGUF files are intended for:
79
+
80
+ - Running embodied AI models with `embodied.cpp`
81
+ - C++ inference and server deployment
82
+ - Simulation and robotics evaluation pipelines supported by `embodied.cpp`
83
+
84
+ These files are **not** the original training checkpoints and may not be directly compatible with upstream Python inference code without the `embodied.cpp` runtime.
85
+
86
+ ## Usage
87
+
88
+ Please use these weights with the corresponding `embodied.cpp` model server or evaluation pipeline.
89
+
90
+ Typical usage includes:
91
+
92
+ - `vla-pi05-server`
93
+ - `vla-hy-vla-server`
94
+ - `wam-lingbot-server`
95
+
96
+ Please refer to the `embodied.cpp` project repository and documentation for build instructions, model-specific runtime requirements, and evaluation examples.
97
+
98
+ ## Roadmap
99
+
100
+ This repository will continue to be updated with GGUF releases for:
101
+
102
+ - **pi0.5**
103
+ - **Hunyuan-VLA-0.5**
104
+ - **LingBot-VA**
105
+
106
+ Additional `embodied.cpp`-compatible model exports may be added in the future.
107
+
108
+ ## Attribution
109
+
110
+ All model credit belongs to the original upstream authors and maintainers.
111
+
112
+ Upstream sources referenced here:
113
+
114
+ - pi0.5 (LeRobot checkpoint): https://huggingface.co/lerobot/pi05_libero_finetuned_v044
115
+ - HY-VLA RoboTwin checkpoint: https://huggingface.co/tencent/Hy-Embodied-0.5-VLA-RoboTwin
116
+ - LingBot-VA related source dataset: https://huggingface.co/datasets/robbyant/libero-long-lerobot
117
+
118
+ ## License and Restrictions
119
+
120
+ This repository contains converted deployment artifacts derived from upstream resources.
121
+
122
+ Please review and comply with the **original upstream licenses, model terms, and dataset restrictions** before use or redistribution.
123
+
124
+ Because different upstream sources may have different licenses or usage constraints, users are responsible for checking the original source pages.
125
+
126
+ ## Citation
127
+
128
+ If you find these GGUF conversions or `embodied.cpp` useful, please cite:
129
+
130
+ ```bibtex
131
+ @article{xu2026embodiedcpp,
132
+ title={Embodied.cpp: A Portable Inference Runtime of Embodied AI Models on Heterogeneous Robots},
133
+ author={Xu, Ling and Han, Chuyu and Li, Borui and Wu, Hao and Jiang, Shiqi and Cao, Ting and Li, Chuanyou and Zhong, Sheng and Wang, Shuai},
134
+ journal={arXiv preprint arXiv:2607.02501},
135
+ year={2026}
136
+ }