Hugging Face releases TRELLIS.2 fine-tuning scripts
Understand this story
Hugging Face has released fine-tuning scripts and a container for the TRELLIS.2 3D generation model, which can run locally or on Hugging Face Jobs to customize the model with users' own 3D assets. The approach uses LoRA from the PEFT library, fine-tuning only the sparse structure stream DiT and the shape stream DiT, while freezing the material generation stage. Officially, it was validated on 188 footwear objects from the GSO dataset, with 9 held out for evaluation and the remaining 179 used for training, and it is claimed that sole geometry accuracy improved. Before use, the Hugging Face account must already have access to the restricted model DINOv3; when using a custom dataset, meshes must first be uploaded to the Hub. Each run on Hugging Face Jobs uses one A100 80 GB.
AI overview based on source reports · Updated 2 days ago
Coverage timeline
Follow the coverage to explore different perspectives.
- Hugging Face BlogThe Definitive Fine-Tuning Guide for 3D Generation Models
This blog provides fine-tuning scripts and Docker containers for adapting TRELLIS.2 to custom 3D assets, runnable locally or on Hugging Face Jobs. Using LoRA adapters on only the sparse-structure and shape flow DiTs, trained on 188 GSO footwear objects, the authors report the mean F-score on nine held-out objects rising from 0.276 to 0.803, Chamfer distance falling from 0.174 to 0.033, and normal consistency rising from 0.530 to 0.804. The texture flow DiT is frozen because the dataset lacks the separate material maps that stage requires.
Activity for this story
- Current comparable activity
- 3
- Peak comparable activity
- 6Nov 13 00:00
- Change in the past 24 hours
- -49%
Trends compare the same fully observed participants, which may cover fewer participants than current activity. Move the pointer or tap the chart to inspect hourly activity; use the left and right arrow keys with a keyboard.