MotionGPT and MotionGPT3
MotionGPT3 is the successor framework that evolves the original MotionGPT's LLM-based motion-language approach by shifting to a Mixture-of-Transformers (MoT) architecture for improved motion understanding and generation capabilities.
About MotionGPT
OpenMotionLab/MotionGPT
[NeurIPS 2023] MotionGPT: Human Motion as a Foreign Language, a unified motion-language generation model using LLMs
Leverages discrete vector quantization to convert 3D motion into motion tokens—treating human movement as a "foreign language" vocabulary—enabling unified language modeling across motion and text modalities. Supports four key tasks: text-to-motion generation, motion captioning, motion prediction, and motion in-betweening through a three-stage pipeline (tokenizer pre-training, motion-language pre-training, and prompt-based instruction tuning). Integrates with HuggingFace for model distribution and builds on SMPL/HumanML3D datasets, with a PyTorch 2.0 implementation and web UI for interactive inference.
About MotionGPT3
OpenMotionLab/MotionGPT3
MotionGPT3: Human Motion as a Second Modality, a MoT-based framework for unified motion understanding and generation
Scores updated daily from GitHub, PyPI, and npm data. How scores work