Update README.md
Browse files
README.md
CHANGED
|
@@ -14,7 +14,7 @@ I was courious, what if a token look at other tokens, but without QKV? Instead,
|
|
| 14 |
## Result:
|
| 15 |
- Training ~14.5 eppch with scheduler's T_MAX=20, train accuracy: Add soon hopefully
|
| 16 |
- Test accuracy: 49.72%
|
| 17 |
-
## Architectural change compared to V7.
|
| 18 |
- Reducing layer in stream A and B
|
| 19 |
- Earlier stride in stream B
|
| 20 |
- Add Transformer-like stack
|
|
|
|
| 14 |
## Result:
|
| 15 |
- Training ~14.5 eppch with scheduler's T_MAX=20, train accuracy: Add soon hopefully
|
| 16 |
- Test accuracy: 49.72%
|
| 17 |
+
## Architectural change compared to V7.6, including:
|
| 18 |
- Reducing layer in stream A and B
|
| 19 |
- Earlier stride in stream B
|
| 20 |
- Add Transformer-like stack
|