Update README.md
Browse files
README.md
CHANGED
|
@@ -1,5 +1,12 @@
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
| 3 |
---
|
|
|
|
|
|
|
| 4 |
Experiment, can DUF can be taken one or more steps further?
|
| 5 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
| 3 |
---
|
| 4 |
+
### TeeZee/GALAXY-XB-v.01 ###
|
| 5 |
+
|
| 6 |
Experiment, can DUF can be taken one or more steps further?
|
| 7 |
+
|
| 8 |
+
### Technical notes:
|
| 9 |
+
- 8 layers removed from both models, as per original paper.
|
| 10 |
+
- base version of upstage/SOLAR-10.7B-v1.0 used for merge
|
| 11 |
+
- no finetuning done yet, this is just a merge, first step in DUF paper
|
| 12 |
+
- next step, if evaluation proves that its at least as 'smart' as base model, should be finetuning to 'recover' after merge
|