Update README.md
Browse files
README.md
CHANGED
@@ -1,17 +1,18 @@
|
|
1 |
-
|
2 |
-
|
3 |
-
gap.png](https://cdn-uploads.huggingface.co/production/uploads/6593502ca2607099284523db/2Uh-yodcWOcKfoHwL8Ze4.png)
|
4 |
-
|
5 |
-
license: odc-by
|
6 |
language:
|
7 |
- en
|
8 |
tags:
|
9 |
- TTT*,
|
10 |
-
-
|
11 |
-
|
12 |
-
|
|
|
|
|
|
|
13 |
---
|
14 |
-
|
15 |
---
|
16 |
-
|
|
|
17 |
---
|
|
|
1 |
+
---
|
2 |
+
license: apache-2.0
|
|
|
|
|
|
|
3 |
language:
|
4 |
- en
|
5 |
tags:
|
6 |
- TTT*,
|
7 |
+
- QAT,
|
8 |
+
---
|
9 |
+
|
10 |
+
![Bridging the information
|
11 |
+
gap.png](https://cdn-uploads.huggingface.co/production/uploads/6593502ca2607099284523db/2Uh-yodcWOcKfoHwL8Ze4.png)
|
12 |
+
|
13 |
---
|
14 |
+
## "The Key" is a small and unique Spatial awareness/inference temperature dataset with "problem solving in diverse scenarios"( giving measured and appropriate inference in models that are too specific or obliterated) the inferred reflection in the form of a Haiku --This is experimental for QAT TTT* implementation--
|
15 |
---
|
16 |
+
## For public Use please give credit to Intelligent Estate and Enjoy, please respond with results and if you achieve a "State Change" in your model. Seems to work very well for QAT methods bringing coding/obliterated models back to coherence.
|
17 |
+
If training with Unsloth use the "Writing content training method but works best with Lamacpp" at Quantization time
|
18 |
---
|