cbtogu commited on
Commit
2ed1f15
·
verified ·
1 Parent(s): c6d9b14

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +112 -3
README.md CHANGED
@@ -1,3 +1,112 @@
1
- ---
2
- license: apache-2.0
3
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ ---
4
+ # GigaBrain Challenge 2026 – Task 6 VLA Policy Model
5
+
6
+ This repository provides the **Vision-Language-Action (VLA) policy model for Task 6** of the **GigaBrain Challenge 2026**.
7
+
8
+ Official Challenge Website:
9
+
10
+ https://gigaai-research.github.io/GigaBrain-Challenge-2026/
11
+
12
+ The GigaBrain Challenge aims to advance research in **embodied intelligence**, including:
13
+
14
+ - Vision-Language-Action (VLA) models
15
+ - world models
16
+ - robot learning systems
17
+ - closed-loop embodied agents
18
+
19
+ ---
20
+
21
+ # Task Overview
22
+
23
+ **Task 6** is one of the official tasks in the GigaBrain Challenge 2026 benchmark suite.
24
+
25
+ Each task evaluates an embodied agent’s ability to generate actions based on multimodal observations and task instructions.
26
+
27
+ For full task descriptions please refer to:
28
+
29
+ https://gigaai-research.github.io/GigaBrain-Challenge-2026/index.html#tracks
30
+
31
+ ---
32
+
33
+ # Model Description
34
+
35
+ This repository contains the **VLA policy model used for Task 6**.
36
+
37
+ The model is designed to:
38
+
39
+ - take **visual observations and language instructions** as input
40
+ - generate **robot actions**
41
+ - interact with a **world model** in a closed-loop evaluation environment
42
+
43
+ The VLA model serves as the **action generation component** in the challenge evaluation pipeline.
44
+
45
+ During evaluation:
46
+ Observation → VLA Model → Action → World Model → Next Observation
47
+
48
+ This loop continues until the task is completed or the episode terminates.
49
+
50
+ Participants may use this model as:
51
+
52
+ - a reference implementation
53
+ - a starting point for model development
54
+ - a component in their own embodied AI systems
55
+
56
+ ---
57
+
58
+ # Dataset
59
+
60
+ The dataset for all tasks is provided in the official challenge dataset repository:
61
+
62
+ https://huggingface.co/datasets/open-gigaai/CVPR-2026-WM-Track-Dataset
63
+
64
+ Access requires:
65
+
66
+ - Team registration
67
+ - Agreement to the challenge license
68
+
69
+ Please request access on the dataset page.
70
+
71
+ ---
72
+
73
+ # License and Usage Terms
74
+
75
+ This model is released as part of the GigaBrain Challenge 2026.
76
+
77
+ Use of this model and the associated datasets is governed by the GigaBrain Challenge 2026 Data & Model License Agreement.
78
+
79
+ Key terms include:
80
+
81
+ Intended for research and competition participation
82
+
83
+ Dataset redistribution is not permitted
84
+
85
+ Commercial use requires permission from the organizers
86
+
87
+ For full license terms please refer to the dataset repository:
88
+
89
+ https://huggingface.co/datasets/open-gigaai/CVPR-2026-WM-Track-Dataset
90
+
91
+ # Citation
92
+
93
+ If you use this model or dataset in academic research, please cite the challenge:
94
+ ```bibtex
95
+ @misc{gigabrain_challenge_2026,
96
+ title={GigaBrain Challenge 2026},
97
+ year={2026},
98
+ howpublished={https://gigaai-research.github.io/GigaBrain-Challenge-2026/}
99
+ }
100
+ ```
101
+
102
+ # Disclaimer
103
+
104
+ This model is provided as-is without warranty of any kind.
105
+
106
+ The organizers are not responsible for any issues arising from its use.
107
+
108
+ # Contact
109
+
110
+ For questions about the challenge, please visit the official website:
111
+
112
+ https://gigaai-research.github.io/GigaBrain-Challenge-2026/