chore: update embedding run IDs and centralize them per model

README.md

@@ -1,3 +1,42 @@
 # Tissue Classification
-* repository for classification of various tissue types using WSIs
-* detailed description can be found on [Youtrack](https://youtrack.rationai.cloud.e-infra.cz/articles/TC)
+
+This repository contains a pipeline for tissue classification in whole-slide
+images (WSIs). It includes preprocessing utilities, dataset splitting tools,
+machine learning training and evaluation code, Hydra configurations, and
+submission scripts for running individual pipeline stages.
+
+## Repository Structure
+
+- `preprocessing/` contains source code for WSI preprocessing, tiling, mask
+  generation, embedding extraction, quality control, and related statistics.
+- `ml/` contains machine learning code for data loading, model training,
+  evaluation, callbacks, and prediction outputs.
+- `split/` contains utilities for train/test and k-fold dataset splitting.
+- `configs/` contains Hydra configuration files for preprocessing, splitting,
+  logging, datasets, and machine learning experiments.
+- `scripts/` contains executable submission scripts for running pipeline stages.
+- `pyproject.toml` and `uv.lock` define the Python project metadata and locked
+  dependencies.
+- `LICENSE` contains the MIT License for this repository.
+
+## Setup
+
+The project uses Python 3.12 and `uv` for dependency management.
+
+```bash
+uv sync
+```
+
+## Usage
+
+Pipeline stages can be run through the corresponding scripts in `scripts/`.
+For example:
+
+```bash
+uv run python scripts/submit_tiling.py
+uv run python scripts/submit_embeddings.py
+uv run python scripts/submit_train_linear_probe.py
+```
+
+Configuration is managed with Hydra. Base configurations are stored in
+`configs/`, with task-specific configurations grouped by pipeline stage.

configs/data/dataset.yaml

@@ -16,7 +16,9 @@ dataset:
     filter_tiles_run_id: "4e8f5d3c82124ea5a8f871a42d3ed9ba"
     stratified_kfold_run_id: "850c81506684450b9af92296acfd045a"
     stratified_group_kfold_run_id: "382b41d2fa894514908e8067949c4326"
-    embedding_run_id: "c325e3a5033b4077b6febb0e3e6b0bd6"
+    embedding_run_ids:
+      virchow2: "262d680f52374614baaf19f1e4d05940"
+      provgigapath: "5099ce57e886411dababfc1234a88b23"
     tissue_masks_run_id: "52bc0924f8624b259819c480c7cf213f"
     tissue_stats_run_id: "16ae2d003d88471b924e5f332415232a"
 

configs/experiment/ml/final_linear_provgigapath_adamw.yaml

@@ -6,7 +6,7 @@ defaults:
 
 embedding_model_name: ProvGigaPath
 embedding_dim: 1536
-embedding_run_id: 410c8672471348ceb4c58817f70fa097
+embedding_run_id: ${dataset.mlflow_artifacts.embedding_run_ids.provgigapath}
 kfold_strategy: stratified_group
 kfold_run_id: ${dataset.mlflow_artifacts.stratified_group_kfold_run_id}
 mlflow_artifact_path: linear_classifier_final_provgigapath

configs/experiment/ml/final_linear_provgigapath_lbfgs.yaml

@@ -6,7 +6,7 @@ defaults:
 
 embedding_model_name: ProvGigaPath
 embedding_dim: 1536
-embedding_run_id: 410c8672471348ceb4c58817f70fa097
+embedding_run_id: ${dataset.mlflow_artifacts.embedding_run_ids.provgigapath}
 kfold_strategy: stratified_group
 kfold_run_id: ${dataset.mlflow_artifacts.stratified_group_kfold_run_id}
 mlflow_artifact_path: linear_classifier_final_provgigapath

configs/experiment/ml/train_linear_provgigapath_adamw_group_kfold.yaml

@@ -6,7 +6,7 @@ defaults:
 
 embedding_model_name: ProvGigaPath
 embedding_dim: 1536
-embedding_run_id: 410c8672471348ceb4c58817f70fa097
+embedding_run_id: ${dataset.mlflow_artifacts.embedding_run_ids.provgigapath}
 mlflow_artifact_path: linear_classifier_provgigapath
 
 metadata:

configs/ml/task/final_linear_classifier.yaml

@@ -12,7 +12,7 @@ mode: fit
 
 embedding_model_name: Virchow2
 embedding_dim: 2560
-embedding_run_id: ${dataset.mlflow_artifacts.embedding_run_id}
+embedding_run_id: ${dataset.mlflow_artifacts.embedding_run_ids.virchow2}
 kfold_strategy: stratified
 kfold_run_id: ${dataset.mlflow_artifacts.stratified_kfold_run_id}
 filter_tiles_run_id: ${dataset.mlflow_artifacts.filter_tiles_run_id}

configs/ml/task/kfold_linear_classifier.yaml

@@ -12,7 +12,7 @@ mode: fit
 
 embedding_model_name: Virchow2
 embedding_dim: 2560
-embedding_run_id: ${dataset.mlflow_artifacts.embedding_run_id}
+embedding_run_id: ${dataset.mlflow_artifacts.embedding_run_ids.virchow2}
 kfold_strategy: stratified
 kfold_run_id: ${dataset.mlflow_artifacts.stratified_kfold_run_id}
 filter_tiles_run_id: ${dataset.mlflow_artifacts.filter_tiles_run_id}