LynnReal-Hand published
The story so far
LynnReal-Hand has released an interactive video world model based on hand control, enabling continuous interaction through hand movements. The model is trained with 100K interaction segments and hand perception supervision to enhance fine motor and object interaction capabilities. The full sequence model FVD reaches 181.67, and MANO Dice is 93.02. The three-step causal model MANO Dice is 92.49. Single-card H200 540p streaming inference achieves 18.03 FPS with a latency of 741ms. LynnReal-Hand will open-source the dataset, model weights, and code to support research on hand-controlled video world models.
AI-written from the reports · Updated 13 hours ago
Coverage timeline
Follow the coverage to see every side of the story.
- Hugging Face Daily PapersLynnReal-Hand: Unlocking World Interaction with Hand Control
LynnReal-Hand has achieved a video world model for continuous interaction through hand gestures, trained with 100K interaction segments and hand perception supervision to enhance fine motor and object interaction capabilities. Its full sequence model FVD reaches 181.67, MANO Dice 93.02, and the three-step causal model MANO Dice 92.49. Single-card H200 540p streaming inference achieves 18.03 FPS with a delay of 741ms. It will open-source the dataset, model weights, and code.
Heat over time
- Now
- 3
- Peak
- 418 Nov, 17:00
- 24h change
- –
Hover or tap the chart to see each hour; use the arrow keys on a keyboard.