nielsr HF Staff commited on
Commit
3dbf23a
·
verified ·
1 Parent(s): d750f93

Improve dataset card with metadata, paper, and code links

Browse files

Hi, I'm Niels from the Hugging Face community science team. I've updated the dataset card to include links to the paper, project page, and official GitHub repository. I also added the `any-to-any` task category to improve the discoverability of the dataset on the Hub and provided a brief description of the ViBES framework.

Files changed (1) hide show
  1. README.md +33 -3
README.md CHANGED
@@ -1,3 +1,33 @@
1
- ---
2
- license: cc-by-4.0
3
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: cc-by-4.0
3
+ task_categories:
4
+ - any-to-any
5
+ ---
6
+
7
+ # ViBES: A Conversational Agent with Behaviorally-Intelligent 3D Virtual Body
8
+
9
+ [**Paper**](https://huggingface.co/papers/2512.14234) | [**Project Page**](https://ai.stanford.edu/~juze/ViBES/) | [**GitHub**](https://github.com/Juzezhang/ViBES)
10
+
11
+ This repository contains the dataset for **ViBES** (Voice in Behavioral Expression and Synchrony), a conversational 3D agent that jointly plans language and movement and executes dialogue-conditioned body actions.
12
+
13
+ ## Overview
14
+ ViBES is a speech-language-behavior (SLB) model with a mixture-of-modality-experts (MoME) backbone: modality-partitioned transformer experts for speech, facial expression, and body motion. The model processes interleaved multimodal token streams with hard routing by modality, while sharing information through cross-expert attention.
15
+
16
+ The dataset release currently includes the facial expression part, with body motion data to follow. The data is processed from sources such as AMASS and BEAT2 to facilitate training and benchmarking of multi-turn conversational behavior.
17
+
18
+ ## Dataset Preprocessing
19
+ To process the data for tokenization and training, please refer to the following scripts in the [official repository](https://github.com/Juzezhang/ViBES):
20
+ - `preprocess/dataset_process_amass_genmo.py`: Generate GENMO motion vectors for AMASS.
21
+ - `preprocess/dataset_process_beat2_genmo.py`: Generate GENMO motion vectors for BEAT2.
22
+
23
+ ## Citation
24
+ If you find our work useful in your research, please consider citing:
25
+ ```bibtex
26
+ @inproceedings{
27
+ zhang2026vibes,
28
+ title={ViBES: A Conversational Agent with Behaviorally-Intelligent 3D Virtual Body},
29
+ author={Juze Zhang and Changan Chen and Xin Chen and Heng Yu and Tiange Xiang and Ali Sartaz Khan and Shrinidhi Kowshika Lakshmikanth and Ehsan Adeli},
30
+ booktitle={CVPR},
31
+ year={2026},
32
+ }
33
+ ```