Document HemEviFuse checkpoint archive
Browse files
README.md
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: pytorch
|
| 3 |
+
tags:
|
| 4 |
+
- HemEviFuse
|
| 5 |
+
- graph-neural-network
|
| 6 |
+
- regression
|
| 7 |
+
- classification
|
| 8 |
+
---
|
| 9 |
+
# HemEviFuse experimental checkpoints
|
| 10 |
+
|
| 11 |
+
Archive of experimental PyTorch checkpoints, grouped by original experiment date and configuration. Includes joint regression/classification and individual task experiments, with multiple architectures and cross-validation folds.
|
| 12 |
+
|
| 13 |
+
Directory names preserve the archived experiment identifiers. These are historical experiments with differing loss functions and settings; they should not be treated as equivalent runs or automatically as final model releases. Each Fold file corresponds to one cross-validation fold. Loading requires the matching HemEviFuse model definitions and configuration.
|
| 14 |
+
|
| 15 |
+
`weights_sha256.jsonl` records relative paths, sizes, and SHA256 checksums once the upload is complete. No training datasets or account credentials are included. Only the archived `.pth` checkpoint files are included in this snapshot; active local training files are outside its scope.
|