Update README.md

README.md

@@ -31,7 +31,7 @@ It has the following specifications:
 
 * Model: ViT-B/32 
 * Input: RBG image of 640x640 pixels 
-* Output/embedding layer size: 512 
+* Output/embedding layer size: 256 
 * Training loss: TripletSemiHardLoss (see [TensorFlow.org](https://www.tensorflow.org/addons/tutorials/losses_triplet)) with batch size 10 (2 anchors, 2 positives, 2 negatives) 
 * Fixed learning rate of 0.000015 with Adam optimizer 
 * Epochs: 100 
@@ -94,11 +94,11 @@ Overall, we find that the model performs very well for classes that are present
 
 | Metric       | Worst-case (without text filter) | Best-case (without text filter) | Worst-case (with text filter) | Best-case (with text filter) | 
 |--------------|------------|-----------|------------|-----------| 
-| Accuracy @ 1 | 0.21       |  0.99     | 0.64       |  0.99     | 
-| Accuracy @ 3 | 0.35       |  1.00     | 0.73       |  1.00     | 
-| Accuracy @ 5 | 0.39       |  1.00     | 0.78       |  1.00     | 
-| Accuracy @ 10| 0.48       |  1.00     | 0.84       |  1.00     | 
-| Accuracy @ 25| 0.75       |  1.00     | 0.90       |  1.00     | 
+| Accuracy @ 1 | 0.22       |  0.99     | 0.63       |  1.00     | 
+| Accuracy @ 3 | 0.31       |  1.00     | 0.80       |  1.00     | 
+| Accuracy @ 5 | 0.63       |  1.00     | 0.91       |  1.00     | 
+| Accuracy @ 10| 0.76       |  1.00     | 0.92       |  1.00     | 
+| Accuracy @ 25| 0.86       |  1.00     | 0.96       |  1.00     | 
 
 ### _Mock snippets_
 
@@ -107,7 +107,7 @@ Overall, we find that the model performs very well for classes that are present
 | Accuracy @ 1 |  0.99                  |  0.99                  | 
 | Accuracy @ 3 |  0.99                  |  1.00                  | 
 | Accuracy @ 5 |  0.99                  |  1.00                  | 
-| Accuracy @ 10|  0.99                  |  1.00                  | 
+| Accuracy @ 10|  1.00                  |  1.00                  | 
 | Accuracy @ 25|  1.00                  |  1.00                  | 
 
 Note that the final model is trained on all data, so we expect performance to increase somewhat as compared to these metrics.