Link complete DEG-s model release
Browse files
README.md
CHANGED
|
@@ -89,6 +89,8 @@ compatibility, not behavior-classification accuracy.
|
|
| 89 |
- The DEG-s folder has `flow/checkpoint.pt` but no top-level checkpoint or spatial checkpoint.
|
| 90 |
DeepEthogram's current model discovery skips this incomplete feature extractor. This release
|
| 91 |
provides three discoverable flow generators and the DEG-m/DEG-f feature extractors.
|
|
|
|
|
|
|
| 92 |
- DEG-f has no `classification_metrics.h5`; the legacy folders use `hyperparameters.yaml`,
|
| 93 |
rather than `config.yaml`. Some inference helpers require newer run files. Use this bundle
|
| 94 |
for pretraining initialization and use the new run produced by fine-tuning for inference.
|
|
|
|
| 89 |
- The DEG-s folder has `flow/checkpoint.pt` but no top-level checkpoint or spatial checkpoint.
|
| 90 |
DeepEthogram's current model discovery skips this incomplete feature extractor. This release
|
| 91 |
provides three discoverable flow generators and the DEG-m/DEG-f feature extractors.
|
| 92 |
+
The [individual DEG-s release](https://huggingface.co/jbohnslav/deepethogram-deg-s)
|
| 93 |
+
provides a complete checkpoint.
|
| 94 |
- DEG-f has no `classification_metrics.h5`; the legacy folders use `hyperparameters.yaml`,
|
| 95 |
rather than `config.yaml`. Some inference helpers require newer run files. Use this bundle
|
| 96 |
for pretraining initialization and use the new run produced by fine-tuning for inference.
|