---
license: cc-by-nc-sa-4.0
base_model: Qwen/Qwen2.5-1.5B-Instruct
language:
- en
pipeline_tag: text-generation
tags:
- 3d-scenes
- indoor-scenes
- furniture
- fine-tuned
- qwen2.5
- respace
- sg-llm
- spatial-reasoning
- text-to-3d
- scene-synthesis
- computer-graphics
---

# respace-sg-llm-1.5b

Fine-tuned version of [Qwen/Qwen2.5-1.5B-Instruct](https://huggingface.co/Qwen/Qwen2.5-1.5B-Instruct) for 3D indoor scene synthesis coined SG-LLM.

Mor information about ReSpace: http://respace.mnbucher.com

For detailed usage instructions, training details, and examples, see the associated repository: https://github.com/GradientSpaces/respace

## Raw Usage
It is not recommended to use SG-LLM separately without the scaffolding for addition/removal that is provided in the ReSpace repository. However, if you want to play around with model capabilities and limitations, you can use it via:

```python
from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("gradient-spaces/respace-sg-llm-1.5b")
tokenizer = AutoTokenizer.from_pretrained("gradient-spaces/respace-sg-llm-1.5b")
```

## Citation
If you use SG-LLM, the ReSpace framework, or if you found our work useful, please cite us as follows:

```bibtex
@article{bucher2025respace,
  title={ReSpace: Text-Driven 3D Scene Synthesis and Editing with Preference Alignment},
  author={Bucher, Martin JJ and Armeni, Iro},
  journal={arXiv preprint arXiv:2506.02459},
  year={2025}
}
```