bio.rodeo
ModelsOrganizationsProvidersLeaderboardAboutSign in
bio.rodeo

The authoritative source for evaluating biological foundation models. No hype, just honest analysis.

Categories
  • DNA & Gene
  • RNA
  • Protein
  • Small molecule
  • Single-cell
  • Spatial omics
  • Pathology
  • Imaging
  • Metabolomics
  • Biosignals
  • Language model
bio.rodeoModelsOrganizationsProvidersLeaderboardAboutFAQSubmit a modelTermsPrivacyContact
© 2026 Pulsatance. All rights reserved. ~
Built by Pulsatance
models / pathology / h0-mini
PathologyImaging
OwkinBioptimusReleased January 2025

H0-mini

Histopathology feature extractor distilled from a 1.1B-parameter teacher into a ViT-Base, with leading robustness to staining and scanner variation.

The short version

  • —Extracts tile features for biomarker, subtype, and gene expression prediction
  • —Holds feature similarity across scanners and staining protocols on matched tissue
  • —Distills a ViT-giant teacher through paired DINO and iBOT objectives
  • —At 86M parameters, embedding a whole slide cohort stays cheap enough to iterate on
50Openness

Where to run it

No providers recorded yet. Browse all providers

H0-mini is a compact histopathology feature extractor produced by distilling a large pathology foundation model into a Vision Transformer roughly thirteen times smaller. It was built by Owkin and Bioptimus, and published at MICCAI 2025.

Pathology foundation models had been scaling along two axes at once, with checkpoints passing a billion parameters and pretraining corpora reaching millions of whole-slide images. That produced real gains and two side effects: embedding a large slide cohort became expensive, and robustness to the mundane variation of clinical practice — different scanners, different staining protocols — went largely unexamined even though it governs whether a model survives deployment. The teacher here is H-optimus-0, a 1.1 billion parameter ViT-giant; the student is a ViT-Base with 86 million parameters.

The notable result is not that distillation preserves accuracy but that it improves robustness. Across the PLISM benchmark of matched tissue imaged under varying stains and scanners, the distilled student retains feature correspondence far better than its own teacher and better than every other pathology foundation model evaluated. H0-mini has since served as the encoder backbone for downstream cell-level models including HistoPLUS.

#Key Features

  • Distillation rather than pretraining from scratch: The student is supervised by the teacher's class and patch tokens through paired DINO and iBOT objectives, giving it representations a same-size model trained from scratch on identical data does not reach.
  • Robustness to acquisition variation: On matched tissue tiles imaged with different scanners, the model's top-10 retrieval accuracy is the highest of all extractors benchmarked, and the advantage holds when staining varies as well.
  • Inference cost proportional to its size: Embedding a fixed set of tiles takes about a tenth of the wall time the 1.1B-parameter teacher requires, making cohort-scale feature extraction practical on modest hardware.
  • Downstream robustness, not just feature robustness: On private breast cancer cohorts spanning multiple scanners and stains, slide-level predictions of ER, PR, HER2, and germline BRCA status agree closely across acquisition conditions.
  • Released with its benchmark: The processed PLISM dataset and the plismbench evaluation harness are published alongside the weights, so robustness claims are independently checkable.

#Technical Details

Distillation follows the DINOv2 recipe with the teacher frozen. For two augmented views of each image, cross-entropy between teacher and student class-token prototype scores forms the DINO loss and the equivalent over patch tokens forms the iBOT loss, applied to all patches without masking. Stochastic depth and KoLeo regularization are removed, and an exponential moving average of the student becomes the final feature extractor. Training ran for 105,000 iterations at batch size 2,048 on 128 NVIDIA V100 32GB GPUs, totaling 4,350 GPU hours. The corpus is deliberately small: 43 million tiles at 224x224 and 20x magnification from 6,093 TCGA slides covering 16 cancer sites, matching Phikon's pretraining data exactly so the comparison isolates the effect of distillation.

On the HEST gene expression benchmark, H0-mini reaches a mean Pearson correlation of 0.404 across nine indications, third of the eleven extractors tabulated behind UNI2-h (0.429) and the teacher (0.422). On EVA it averages 0.782 across eight tasks, within 0.012 of the top entry but seventh in the table. A Wilcoxon signed-rank test over the 17 HEST and EVA tasks combined places it fourth overall. The original preprint's abstract advertised third place on HEST and fifth on EVA; the peer-reviewed version replaces those claims with a statement of comparable performance, and the tables are the authoritative figures. On PLISM — 46 tissue types under 13 staining conditions across 7 scanners, registered into 91 slides of 16,278 tiles each — it reports median cosine similarity of 0.92 cross-scanner and 0.79 across both axes, with top-10 accuracies of 0.86 and 0.18, the best in both columns.

#Applications

H0-mini is a drop-in tile encoder for computational pathology pipelines: extract patch embeddings, aggregate them with attention-based multiple instance learning or mean pooling, and train a light head for biomarker prediction, subtype classification, survival modeling, or spatial gene expression. Its size suits cohorts running to thousands of slides or inference confined to one commodity GPU, and its robustness profile suits slides arriving from multiple sites with heterogeneous scanners and stains.

#Impact

The work reframes distillation in digital pathology as more than a compression step, showing it confers an invariance property the teacher lacks, and the accompanying release of a processed PLISM benchmark made robustness a reported axis in subsequent pathology model releases. Two caveats attach. Distillation is not cheap — roughly 1.7x the compute of training an equivalent ViT-Base from scratch, since the teacher must be evaluated throughout — and the method requires access to the teacher's projection head, which is not always published. The weights carry a CC-BY-NC-ND-4.0 license behind a gated HuggingFace request, restricting use to non-commercial academic research, markedly more restrictive than the Apache-2.0 terms of the teacher.

At a glance

Parameters
86 Million
Released
January 2025
Category
Pathology
Organizations
Owkin / Bioptimus

Links

GitHub RepositoryResearch PaperResearch PaperHuggingFace ModelDataset

Tags

feature_extractionhistologyknowledge_distillationself_supervisedvision_transformer

Something wrong?

Much of this page is generated or calculated automatically. Flag anything that looks off and we will re-run it.