Improve model card and add metadata
Browse filesHi! I'm Niels, part of the community science team at Hugging Face. I'm opening this PR to improve the documentation for this model.
This PR adds:
- The `text-generation` pipeline tag to the metadata for better discoverability.
- Links to the paper ([Solve the Loop: Attractor Models for Language and Reasoning](https://huggingface.co/papers/2605.12466)), project page, and official GitHub repository.
- A sample usage snippet based on the GitHub documentation.
- The BibTeX citation from the paper.
These changes help users understand how to use the model and provide proper context regarding the research behind it.
README.md
CHANGED
|
@@ -1,3 +1,37 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: mit
|
| 3 |
-
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: mit
|
| 3 |
+
pipeline_tag: text-generation
|
| 4 |
+
---
|
| 5 |
+
|
| 6 |
+
# Solve the Loop: Attractor Models for Language and Reasoning
|
| 7 |
+
|
| 8 |
+
This repository contains the 770M parameter Attractor Model introduced in [Solve the Loop: Attractor Models for Language and Reasoning](https://huggingface.co/papers/2605.12466).
|
| 9 |
+
|
| 10 |
+
Attractor Models offer a promising alternative to purely feed-forward computation by iteratively refining latent representations. In this architecture, a backbone module first proposes output embeddings, then an attractor module refines them by solving for the fixed point using implicit differentiation. This model delivers a Pareto improvement over standard Transformers in language modeling and demonstrates strong performance on reasoning tasks.
|
| 11 |
+
|
| 12 |
+
- **Paper:** [Solve the Loop: Attractor Models for Language and Reasoning](https://huggingface.co/papers/2605.12466)
|
| 13 |
+
- **Project Page:** [attractor-models.github.io](https://attractor-models.github.io/)
|
| 14 |
+
- **Repository:** [jacobfa/Attractor](https://github.com/jacobfa/Attractor)
|
| 15 |
+
|
| 16 |
+
## Usage
|
| 17 |
+
|
| 18 |
+
To use this model, you need to install the official package from the [GitHub repository](https://github.com/jacobfa/Attractor).
|
| 19 |
+
|
| 20 |
+
```python
|
| 21 |
+
from attractor.models.attractor import Attractor, AttractorConfig
|
| 22 |
+
|
| 23 |
+
# Construct the configuration for the 770m model
|
| 24 |
+
config = AttractorConfig.from_name("attractor-large-770m")
|
| 25 |
+
model = config.construct_model()
|
| 26 |
+
```
|
| 27 |
+
|
| 28 |
+
## Citation
|
| 29 |
+
|
| 30 |
+
```bibtex
|
| 31 |
+
@article{feinashley2026attractor,
|
| 32 |
+
title={Solve the Loop: Attractor Models for Language and Reasoning},
|
| 33 |
+
author={Fein-Ashley, Jacob and Rashidinejad, Paria},
|
| 34 |
+
year={2026},
|
| 35 |
+
url={https://arxiv.org/abs/2605.12466}
|
| 36 |
+
}
|
| 37 |
+
```
|