---
license: mit
library_name: pytorch
tags:
- hdtree
- pytorch
- mnist
- single-cell
- clustering
---

# HDTree ICML Checkpoints

This repository hosts pretrained checkpoints for the public HDTree ICML release.

Paper:
- ICML 2026 poster page: https://icml.cc/virtual/2026/poster/61839
- arXiv abstract: https://arxiv.org/abs/2506.23287
- arXiv PDF: https://arxiv.org/pdf/2506.23287

Code: https://github.com/zangzelin/code_HDTree_icml

## Files

| File | Dataset | Configuration | Notes |
|---|---|---|---|
| `checkpoints/mnist/hdtree_mnist_best_epoch59_acc0.97570.pth` | MNIST | `configs/mnist.yaml` | Best MNIST checkpoint from the full run by checkpoint validation accuracy. |
| `checkpoints/limb/hdtree_limb_i10_epoch199_acc0.53921.pth` | Limb | `configs/limb.yaml` default | Limb sweep i10/default checkpoint. |

## Reported Metrics

MNIST full run summary:

| ACC | DP | LP | NMI |
|---:|---:|---:|---:|
| 0.97310 | 0.93262 | 0.97310 | 0.92999 |

Limb i10 run summary (`batch_size=1000`, `K=10`, `exaggeration_lat=0.5`, `nu_lat=0.3`):

| ACC | DP | LP | NMI |
|---:|---:|---:|---:|
| 0.52860 | 0.41029 | 0.58370 | 0.49042 |

The included `logs/` files contain the original run outputs used to record these metrics.

## Download

```bash
pip install huggingface_hub
huggingface-cli download zangzelin/HDTree-ICML-checkpoints --local-dir .
```

Then use the public code repository's validation script, for example:

```bash
bash scripts/validate_checkpoint.sh checkpoints/limb/hdtree_limb_i10_epoch199_acc0.53921.pth
```

## Checksums

See `SHA256SUMS`.