Update README.md

README.md

@@ -1,3 +1,72 @@
----
-license: apache-2.0
----
+---
+license: apache-2.0
+base_model:
+- Qwen/Qwen2.5-VL-3B-Instruct
+---
+
+# SpatialEvo: Self-Evolving Spatial Intelligence via Deterministic Geometric Environments
+
+<a href="<PAPER_LINK>" target="_blank">
+    <img alt="Paper" src="https://img.shields.io/badge/arXiv-SpatialEvo-red?logo=arxiv" height="20" />
+</a>
+<a href="https://github.com/ZJU-REAL/SpatialEvo" target="_blank">
+    <img alt="Code" src="https://img.shields.io/badge/Code-SpatialEvo-white?logo=github" height="20" />
+</a>
+<a href="https://huggingface.co/lidingm/SpatialEvo-7B" target="_blank">
+    <img alt="Model" src="https://img.shields.io/badge/%F0%9F%A4%97%20_Model-SpatialEvo_7B-ffc107?color=ffc107&logoColor=white" height="20" />
+</a>
+<a href="https://huggingface.co/datasets/lidingm/SpatialEvo-160K" target="_blank">
+    <img alt="Data" src="https://img.shields.io/badge/%F0%9F%A4%97%20_Data-SpatialEvo_160k-ffc107?color=ffc107&logoColor=white" height="20" />
+</a>
+
+## SpatialEvo-3B
+
+
+This repository contains **SpatialEvo-3B**, introduced in [SpatialEvo: Self-Evolving Spatial Intelligence via Deterministic Geometric Environments](<PAPER_LINK>).
+
+## Model Description
+
+SpatialEvo-3B is fine-tuned from **Qwen2.5-VL-3B-Instruct** using the SpatialEvo self-evolving framework. Instead of relying on manually annotated datasets or model voting to construct pseudo-labels, SpatialEvo leverages a **Deterministic Geometric Environment (DGE)** that programmatically computes exact ground truth from 3D point clouds and camera poses, enabling zero-noise online reinforcement learning across 16 spatial reasoning task categories.
+
+A single shared-parameter policy co-evolves as both a **Questioner** and a **Solver** under GRPO optimization, while a lightweight **Task Scheduler** drives adaptive curriculum learning based on historical accuracy — without any manual stage design or human annotation.
+
+## Performance
+
+| Benchmark | Baseline | SpatialLadder | SpaceR | SpatialSSRL | **SpatialEvo (Ours)** |
+|-----------|----------|---------------|--------|-------------|----------------------|
+| VSI-Bench | 28.1 | **45.7** | 36.0 | 28.0 | 39.2 |
+| RealWorldQA | 63.4 | 57.1 | 61.4 | 65.4 | **66.5** |
+| EmbSpatial | 55.9 | 57.6 | 55.6 | 59.8 | **61.2** |
+| SpatialViz | 24.2 | 28.6 | **31.9** | 25.9 | 25.4 |
+| STARE | 33.1 | 26.4 | 36.8 | 36.8 | **36.9** |
+| CoreCognition | 56.8 | **58.3** | 29.1 | 57.6 | 57.4 |
+| ViewSpatial | 36.2 | 43.0 | 35.9 | 38.4 | **42.3** |
+| V-STAR | 74.9 | 36.7 | 75.4 | **77.0** | 75.4 |
+| MMStar | 54.6 | 45.8 | 44.9 | **56.5** | 55.2 |
+| **AVG** | 47.5 | 44.4 | 45.2 | 49.5 | **51.1** |
+
+All baselines are evaluated on Qwen2.5-VL-3B. **Bold** denotes the best result per benchmark.
+
+## Usage
+
+```python
+from transformers import AutoProcessor, Qwen2_5_VLForConditionalGeneration
+
+model = Qwen2_5_VLForConditionalGeneration.from_pretrained("lidingm/SpatialEvo-3B")
+processor = AutoProcessor.from_pretrained("lidingm/SpatialEvo-3B")
+```
+
+## Citation
+
+If you find SpatialEvo useful, please consider citing our work:
+
+```bibtex
+
+```
+
+## Related Resources
+
+- 📄 [Paper](<PAPER_LINK>)
+- 💻 [GitHub Repository](https://github.com/ZJU-REAL/SpatialEvo)
+- 🤗 [SpatialEvo-7B](https://huggingface.co/lidingm/SpatialEvo-7B)
+- 🤗 [SpatialEvo-160K Dataset](https://huggingface.co/datasets/lidingm/SpatialEvo-160K)