Upload final_model weights and essential eval artifacts for both variants
Browse files
README.md
CHANGED
|
@@ -187,12 +187,14 @@ bench = ds["benchmark"] # 60 images
|
|
| 187 |
## Repo layout
|
| 188 |
|
| 189 |
```
|
| 190 |
-
without_preprocess/
|
| 191 |
-
final_model.pt
|
| 192 |
-
|
|
|
|
| 193 |
benchmark_eval_results.json
|
| 194 |
-
with_preprocess/
|
| 195 |
final_model.pt
|
|
|
|
| 196 |
results.json
|
| 197 |
benchmark_eval_results.json
|
| 198 |
```
|
|
|
|
| 187 |
## Repo layout
|
| 188 |
|
| 189 |
```
|
| 190 |
+
without_preprocess/ ← recommended (full-page test & benchmark)
|
| 191 |
+
final_model.pt ← production weights
|
| 192 |
+
model_card.json
|
| 193 |
+
results.json ← test metrics + classification report
|
| 194 |
benchmark_eval_results.json
|
| 195 |
+
with_preprocess/ ← center-crop train/val only; not for full-page deploy
|
| 196 |
final_model.pt
|
| 197 |
+
model_card.json
|
| 198 |
results.json
|
| 199 |
benchmark_eval_results.json
|
| 200 |
```
|
with_preprocess/benchmark_confusion_matrix.png
ADDED
|
with_preprocess/confusion_matrix.png
ADDED
|
without_preprocess/benchmark_confusion_matrix.png
ADDED
|
without_preprocess/confusion_matrix.png
ADDED
|