Update README.md
Browse files
README.md
CHANGED
|
@@ -3,6 +3,14 @@ license: cc-by-nc-sa-4.0
|
|
| 3 |
---
|
| 4 |
See Model Card from MermaidMistral or MermaidLLama or MermaidStable3B for more information, this is the same dataset applied to the Solar 10.7B Base Model,
|
| 5 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6 |
Interesting findings is that the hand curated 800 examples I made of as diverse edge cases I could come up with seem to still improve the models ability,
|
| 7 |
Which shows both in the training loss and eval loss trends as they continue to get closer to 0 together, with the eval loss at roughly the same as training loss values.
|
| 8 |
|
|
|
|
| 3 |
---
|
| 4 |
See Model Card from MermaidMistral or MermaidLLama or MermaidStable3B for more information, this is the same dataset applied to the Solar 10.7B Base Model,
|
| 5 |
|
| 6 |
+
How to use my model, please ensure the response includes
|
| 7 |
+
|
| 8 |
+
### Response:
|
| 9 |
+
```mermaid\n
|
| 10 |
+
graph TB
|
| 11 |
+
|
| 12 |
+
|
| 13 |
+
It is critical because I have it trained to work in my pipeline that anytime I put "```mermaid" it knows to create a mermaid block of anything in its context.
|
| 14 |
Interesting findings is that the hand curated 800 examples I made of as diverse edge cases I could come up with seem to still improve the models ability,
|
| 15 |
Which shows both in the training loss and eval loss trends as they continue to get closer to 0 together, with the eval loss at roughly the same as training loss values.
|
| 16 |
|