---
library_name: cellpose
pipeline_tag: image-segmentation
base_model: mouseland/cellpose-sam
base_model_relation: quantized
license: bsd-3-clause
arXiv: 2609.21038
tags:
- cellpose
- cellpose-sam
- microscopy
- quantization
- image-segmentation
---

<div align="center">
  <img src="https://cdn-uploads.huggingface.co/production/uploads/68ed488e386d9944bc2ea3a5/z6lWF8anMTDnO-E3zrFTU.png" alt="Capicú Technologies" />
</div>

# Cellpose-SAM WQuant — W2A16-G64

Ready-to-load `W2A16-G64` quantized variant of
[Cellpose-SAM cpsam_v2](https://huggingface.co/mouseland/cellpose-sam).

## Load

```python
import runpy
from huggingface_hub import hf_hub_download

loader = runpy.run_path(
    hf_hub_download("capicu-ai/cellpose-sam-wquant-w2a16-g64", "runtime_loader.py")
)
model = loader["load_model"]()
```

Pass `device="cuda"` to `load_model` for CUDA.

## Result

- Weight size: 96.3 MiB
- Compression versus FP32: 12.08x
- Evaluated paired samples: 20
- Complete bounded-panel criterion met: No
- Reload verification normalized RMSE: 0.0

`results.json` and `model_registry.json` contain the compact machine-readable
record.

## Citation

If you use this model or code, please cite our paper:

```bibtex
@article{cruzromero2026retention,
  title={Retention-Constrained Post-Training Quantization of Cellpose-SAM for Stem Cell Microscopy},
  author={Cruz Romero, Sebasti{\'a}n A.},
  journal={arXiv preprint arXiv:2609.21038},
  note={Accepted at NeurIPS 2026 LXAI Workshop},
  year={2026}
}
```