Update README.md
Browse files
README.md
CHANGED
|
@@ -19,7 +19,7 @@ The oddity of geometric structure being fully encoded BY the transformer does no
|
|
| 19 |
represents the necessary statistic data that can be consumed by the geometric head. The format changes a little, but not by much, since the geometric observer
|
| 20 |
system in it's current format is meant to be an observer to provide influence, and not a direct controller YET.
|
| 21 |
|
| 22 |
-
# Anchors performed perfectly to specification
|
| 23 |
|
| 24 |
|
| 25 |
|
|
@@ -27,7 +27,7 @@ The entire spectrum showed the anchors began roughly rounded and blobby, slowly
|
|
| 27 |
|
| 28 |
Absolutely perfect. The BCE worked perfectly. I'm floored at to hell well that element worked, even if it's not perfect yet. The math is aligned.
|
| 29 |
|
| 30 |
-
# SVD 3x3 kernel optimization
|
| 31 |
The 3x3 SVD kernel has shown more than potential for use.
|
| 32 |
|
| 33 |
It's not as stable as egen but it's most definitely a powerful potential. I'll be working to stabilize the variants using fp64 and rounding spectrum over time.
|
|
@@ -43,7 +43,7 @@ so I'll be adjusting the entire notebook spectrum.
|
|
| 43 |
The results will be better sorted for the next sweep, as the current sweep doesn't have easy dropdown/hide settings in the GUI. It should be easy to see and use.
|
| 44 |
|
| 45 |
|
| 46 |
-
# Actionable Utility
|
| 47 |
Almost all tested shapes have a potential to teach the system for tasks.
|
| 48 |
|
| 49 |
The large array of math will require a streamlined series of sweeps to run in a very optimal environment.
|
|
|
|
| 19 |
represents the necessary statistic data that can be consumed by the geometric head. The format changes a little, but not by much, since the geometric observer
|
| 20 |
system in it's current format is meant to be an observer to provide influence, and not a direct controller YET.
|
| 21 |
|
| 22 |
+
## Anchors performed perfectly to specification
|
| 23 |
|
| 24 |
|
| 25 |
|
|
|
|
| 27 |
|
| 28 |
Absolutely perfect. The BCE worked perfectly. I'm floored at to hell well that element worked, even if it's not perfect yet. The math is aligned.
|
| 29 |
|
| 30 |
+
## SVD 3x3 kernel optimization
|
| 31 |
The 3x3 SVD kernel has shown more than potential for use.
|
| 32 |
|
| 33 |
It's not as stable as egen but it's most definitely a powerful potential. I'll be working to stabilize the variants using fp64 and rounding spectrum over time.
|
|
|
|
| 43 |
The results will be better sorted for the next sweep, as the current sweep doesn't have easy dropdown/hide settings in the GUI. It should be easy to see and use.
|
| 44 |
|
| 45 |
|
| 46 |
+
## Actionable Utility
|
| 47 |
Almost all tested shapes have a potential to teach the system for tasks.
|
| 48 |
|
| 49 |
The large array of math will require a streamlined series of sweeps to run in a very optimal environment.
|