---
license: mit
language:
- en
library_name: jax
tags:
- perturbation-prediction
- prior-data-fitted-networks
- in-context-learning
- single-cell
- causal-inference
- diffusion-transformer
datasets:
- marvinsxtr/MapPFN
pipeline_tag: other
---

# MapPFN Weights

Pre-trained and fine-tuned checkpoints for [MapPFN: Learning Causal Perturbation Maps in Context](https://arxiv.org/abs/2601.21092) (Sextro et al., 2026).

## Checkpoints

- `model.ckpt` — Pre-trained on synthetic biological prior (50 dimensions, 400k steps)
- `model_finetuned_frangieh.ckpt` — Fine-tuned on [Frangieh et al. (2021)](https://doi.org/10.1038/s41588-021-00779-1)
- `model_finetuned_papalexi.ckpt` — Fine-tuned on [Papalexi et al. (2021)](https://doi.org/10.1038/s41588-021-00778-2)

All checkpoints share the same MMDiT architecture (~25M parameters) and differ only in training data. See the [GitHub repository](https://github.com/marvinsxtr/MapPFN) for inference and fine-tuning code.

## Citation

```bibtex
@inproceedings{sextro2026mappfn,
  title     = {Map{PFN}: Learning Causal Perturbation Maps in Context},
  author    = {Marvin Sextro and Weronika K{\l}os and Gabriel Dernbach},
  booktitle = {Advances in Neural Information Processing Systems},
  volume    = {39},
  year      = {2026}
}
```

**Links:** [Paper](https://arxiv.org/abs/2601.21092) | [Code](https://github.com/marvinsxtr/MapPFN) | [Datasets](https://huggingface.co/datasets/marvinsxtr/MapPFN) | [Project Page](https://marvinsxtr.github.io/MapPFN)