Update README.md
Browse files
README.md
CHANGED
|
@@ -26,7 +26,6 @@ This Hub repo contains the **inference-only** slide encoder: interpolate → Lay
|
|
| 26 |
|
| 27 |
| | |
|
| 28 |
|---|---|
|
| 29 |
-
| Developed by | Li Lab, Stanford University |
|
| 30 |
| Model type | Slide-level ABMIL aggregator |
|
| 31 |
| Inputs | Patch features \(X \in \mathbb{R}^{N \times C}\), \(C \in \{768, 1024, 1280, 1536\}\) |
|
| 32 |
| Outputs | `features_dim` \([1, C]\), `features` \([1, 768]\), `attention_weights` \([1, 1, N]\) |
|
|
@@ -99,7 +98,6 @@ Research feature extraction for computational pathology (classification, biomark
|
|
| 99 |
|
| 100 |
- Requires pre-extracted tile embeddings from the five foundation models above; it does not encode RGB tiles.
|
| 101 |
- Virchow2 inputs of dimension ≥ 2560 are averaged as CLS + mean → 1280, matching the paper.
|
| 102 |
-
- Licensed for non-commercial academic use.
|
| 103 |
|
| 104 |
## Citation
|
| 105 |
|
|
|
|
| 26 |
|
| 27 |
| | |
|
| 28 |
|---|---|
|
|
|
|
| 29 |
| Model type | Slide-level ABMIL aggregator |
|
| 30 |
| Inputs | Patch features \(X \in \mathbb{R}^{N \times C}\), \(C \in \{768, 1024, 1280, 1536\}\) |
|
| 31 |
| Outputs | `features_dim` \([1, C]\), `features` \([1, 768]\), `attention_weights` \([1, 1, N]\) |
|
|
|
|
| 98 |
|
| 99 |
- Requires pre-extracted tile embeddings from the five foundation models above; it does not encode RGB tiles.
|
| 100 |
- Virchow2 inputs of dimension ≥ 2560 are averaged as CLS + mean → 1280, matching the paper.
|
|
|
|
| 101 |
|
| 102 |
## Citation
|
| 103 |
|