Datasets:
Document Training Part 1 upload and repository layout
Browse files
README.md
CHANGED
|
@@ -34,7 +34,7 @@ configs:
|
|
| 34 |
|
| 35 |
## Files currently hosted in this repository
|
| 36 |
|
| 37 |
-
This repository
|
| 38 |
|
| 39 |
- `resources/CryoAtlas_v1.0_benchmark_resources.tar.gz`: the complete benchmark resource package.
|
| 40 |
- `Benchmark_outcome/`: individually accessible evaluation CSVs, configured as five Dataset Viewer subsets with 618 test targets each.
|
|
|
|
| 34 |
|
| 35 |
## Files currently hosted in this repository
|
| 36 |
|
| 37 |
+
This repository hosts the benchmark resource archive, global metadata, and five evaluation CSVs. **Training Part 1 is currently being uploaded to `train/part01/`**, comprising 17 independent archives and 4,164 entries. Its subset metadata is available under `metadata/train/part01/`, with checksums under `checksums/train/part01/`. Archives become available individually after verification. The other three training parts and the complete test archives have not yet been mirrored here. The resource archive contains the predictions and examples described below.
|
| 38 |
|
| 39 |
- `resources/CryoAtlas_v1.0_benchmark_resources.tar.gz`: the complete benchmark resource package.
|
| 40 |
- `Benchmark_outcome/`: individually accessible evaluation CSVs, configured as five Dataset Viewer subsets with 618 test targets each.
|