mws-physical-ai/cosmos3-g1brainco-datasets

Cosmos3 G1+BrainCo Datasets Synthetic video+action datasets produced while fine-tuning nvidia/Cosmos3-Nano (15.8B MoT, Qwen3-VL-8B backbone) as a joint World-Action-Model (WAM) for the Unitree G1 humanoid + BrainCo dexterous hand, and while self-distilling that policy with Direct Preference Optimization (DPO). Project / code / paper: MWS-Physical-AI/Cosmos_3_tuning ("Cosmos3-Nano → Unitree G1 + BrainCo: перенос навыков манипуляции" — transferring manipulation skills learned by… See the full description on the dataset page: https://huggingface.co/datasets/mws-physical-ai/cosmos3-g1brainco-datas

<iframe src="https://silytics.si/embed/item/89c40bcc-8d4a-4e94-a0ea-c9a9b73fc930" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 18 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.