Update README.md
Browse files
README.md
CHANGED
@@ -8,3 +8,30 @@ pinned: false
|
|
8 |
---
|
9 |
|
10 |
Edit this `README.md` markdown file to author your organization card.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
8 |
---
|
9 |
|
10 |
Edit this `README.md` markdown file to author your organization card.
|
11 |
+
|
12 |
+
```mermaid
|
13 |
+
graph TD;
|
14 |
+
A[Start Input Batch - N Images] -->|Augment Twice| B[Two Views - 2N Images]
|
15 |
+
B -->|Pass through Online Encoder| C[Online Network - f_theta]
|
16 |
+
B -->|Pass through Target Encoder| D[Target Network - f_xi]
|
17 |
+
|
18 |
+
C -->|Extract Representation| E[Online Representation - y_theta]
|
19 |
+
D -->|Extract Representation| F[Target Representation - y_xi]
|
20 |
+
|
21 |
+
E -->|Project| G[Online Projection - z_theta]
|
22 |
+
F -->|Project| H[Target Projection - z_xi]
|
23 |
+
|
24 |
+
G -->|Predict| I[Prediction Head - q_theta]
|
25 |
+
|
26 |
+
I -->|Compute Loss| J[Loss Calculation - Mean Squared Error]
|
27 |
+
H -->|Stop Gradient| J
|
28 |
+
|
29 |
+
J -->|Backpropagation| K[Update Online Network - f_theta]
|
30 |
+
K -->|Momentum Update| L[Update Target Network - f_xi, Slow Update]
|
31 |
+
|
32 |
+
L -->|Repeat for Next Batch| A
|
33 |
+
|
34 |
+
K -->|Trained Encoder Used for Downstream Tasks| M[Linear Classifier or Fine Tuning]
|
35 |
+
```
|
36 |
+
|
37 |
+
|