Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,19 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: mit
|
| 3 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: mit
|
| 3 |
+
---
|
| 4 |
+
|
| 5 |
+
# Experimentation Head
|
| 6 |
+
Everything led to the RoPE enhancement. You will find the proof in a trainer attached to this with basic instructions on how to activate it in Colab.
|
| 7 |
+
|
| 8 |
+
This is proof for the baseline of this concept. The foundational math works.
|
| 9 |
+
Everything relational to this behavior is to be expanded directly into a full wide bert with the relational complexity of a standard bert.
|
| 10 |
+
|
| 11 |
+
# The experimental expansion
|
| 12 |
+
This model will be trained wide and shallow. The VRAM will likely fill the entirey A100 80 gig for the first sets and then shrink abruptly from there, but not necessarily required.
|
| 13 |
+
|
| 14 |
+
# Cantor Fractal mathematics
|
| 15 |
+
The cantor fractal routing has been proven, the rope has now been proven to allow skipping, and the system is entirely relationally compatible with truthful training.
|
| 16 |
+
|
| 17 |
+
I will be drafting a paper tomorrow specific to this ruleset and the importance of the necessity for certain guidelines to be upheld for full cohesion with certain elements.
|
| 18 |
+
|
| 19 |
+
If you're reading this, thank you for following me and baring with this endurance test of irrationality mixed with rationality.
|