Update README.md
Browse files
README.md
CHANGED
|
@@ -20,7 +20,7 @@ base_model:
|
|
| 20 |
# Kaiko midnight
|
| 21 |
Midnight - Training State-of-the-Art Pathology Foundation Models with Orders of Magnitude Less Data
|
| 22 |
|
| 23 |
-
This repository contains the model checkpoints for the Midnight-12k model presented in our paper titled "Training state-of-the-art pathology foundation models with orders of magnitude less data." Our approach achieves competitive performance compared to leading pathology foundation models (FMs), despite being trained on significantly fewer whole slide images (WSIs).
|
| 24 |
|
| 25 |
## Overview
|
| 26 |
|
|
@@ -129,7 +129,7 @@ Our best model **Midnight-92k/392** consistently outperforms or matches leading
|
|
| 129 |
| [UNI-2](https://huggingface.co/MahmoodLab/UNI2-h) | **0.776** | **0.885** | **0.924** | **0.651** | **0.863** | **0.970** | 0.777 | 0.829 | **0.951** | **0.873** | **0.666** | 0.626 | 0.644 | **0.431** |
|
| 130 |
| **[Midnight-92k](#usage)** | **0.767** | **0.882** | 0.889 | 0.615 | 0.793 | **0.967** | **0.823** | 0.831 | 0.948 | **0.872** | 0.643 | 0.629 | 0.656 | **0.425** |
|
| 131 |
| [Virchow2](https://huggingface.co/paige-ai/Virchow2) | 0.766 | 0.835 | 0.890 | 0.633 | 0.818 | 0.966 | **0.791** | **0.865** | 0.938 | 0.860 | 0.646 | 0.640 | 0.674 | 0.403 |
|
| 132 |
-
| **
|
| 133 |
| [Kaiko-B8](https://github.com/kaiko-ai/towards_large_pathology_fms) | 0.757 | 0.799 | 0.876 | 0.641 | **0.842** | 0.960 | 0.761 | 0.830 | 0.920 | 0.836 | 0.650 | **0.644** | 0.686 | 0.391 |
|
| 134 |
| [H-Optimus-0](https://huggingface.co/bioptimus/H-optimus-0) | 0.755 | 0.831 | 0.752 | 0.620 | 0.813 | 0.962 | 0.769 | **0.850** | 0.943 | 0.847 | **0.672** | **0.644** | **0.687** | **0.425** |
|
| 135 |
| [Prov_GigaPath](https://github.com/prov-gigapath/prov-gigapath) | 0.752 | 0.853 | 0.794 | 0.626 | **0.846** | 0.959 | 0.727 | 0.831 | 0.944 | 0.812 | 0.657 | 0.628 | **0.688** | 0.405 |
|
|
|
|
| 20 |
# Kaiko midnight
|
| 21 |
Midnight - Training State-of-the-Art Pathology Foundation Models with Orders of Magnitude Less Data
|
| 22 |
|
| 23 |
+
This repository contains the model checkpoints for the **Midnight-12k** model presented in our paper titled "Training state-of-the-art pathology foundation models with orders of magnitude less data." Our approach achieves competitive performance compared to leading pathology foundation models (FMs), despite being trained on significantly fewer whole slide images (WSIs).
|
| 24 |
|
| 25 |
## Overview
|
| 26 |
|
|
|
|
| 129 |
| [UNI-2](https://huggingface.co/MahmoodLab/UNI2-h) | **0.776** | **0.885** | **0.924** | **0.651** | **0.863** | **0.970** | 0.777 | 0.829 | **0.951** | **0.873** | **0.666** | 0.626 | 0.644 | **0.431** |
|
| 130 |
| **[Midnight-92k](#usage)** | **0.767** | **0.882** | 0.889 | 0.615 | 0.793 | **0.967** | **0.823** | 0.831 | 0.948 | **0.872** | 0.643 | 0.629 | 0.656 | **0.425** |
|
| 131 |
| [Virchow2](https://huggingface.co/paige-ai/Virchow2) | 0.766 | 0.835 | 0.890 | 0.633 | 0.818 | 0.966 | **0.791** | **0.865** | 0.938 | 0.860 | 0.646 | 0.640 | 0.674 | 0.403 |
|
| 132 |
+
| [**Midnight-12k**](#usage) | 0.763 | 0.803 | **0.907** | 0.639 | 0.840 | **0.967** | 0.790 | 0.815 | 0.931 | **0.869** | 0.656 | 0.625 | 0.664 | 0.412 |
|
| 133 |
| [Kaiko-B8](https://github.com/kaiko-ai/towards_large_pathology_fms) | 0.757 | 0.799 | 0.876 | 0.641 | **0.842** | 0.960 | 0.761 | 0.830 | 0.920 | 0.836 | 0.650 | **0.644** | 0.686 | 0.391 |
|
| 134 |
| [H-Optimus-0](https://huggingface.co/bioptimus/H-optimus-0) | 0.755 | 0.831 | 0.752 | 0.620 | 0.813 | 0.962 | 0.769 | **0.850** | 0.943 | 0.847 | **0.672** | **0.644** | **0.687** | **0.425** |
|
| 135 |
| [Prov_GigaPath](https://github.com/prov-gigapath/prov-gigapath) | 0.752 | 0.853 | 0.794 | 0.626 | **0.846** | 0.959 | 0.727 | 0.831 | 0.944 | 0.812 | 0.657 | 0.628 | **0.688** | 0.405 |
|