Update README.md
Browse files
README.md
CHANGED
@@ -15,7 +15,10 @@ tags: []
|
|
15 |
|
16 |
<!-- Provide a longer summary of what this model is. -->
|
17 |
|
18 |
-
This is
|
|
|
|
|
|
|
19 |
|
20 |
- **Developed by:** [More Information Needed]
|
21 |
- **Funded by [optional]:** [More Information Needed]
|
|
|
15 |
|
16 |
<!-- Provide a longer summary of what this model is. -->
|
17 |
|
18 |
+
This is a gemma-2-2b-it model trained using RMU to be worse at answering questions about bioweapons. s is the unlearning coefficient aka 'c', and a is the weight of the retain loss.
|
19 |
+
https://arxiv.org/pdf/2403.03218
|
20 |
+
This is part of a series to replicate and extend the discoveries [here](https://arxiv.org/abs/2410.19278).
|
21 |
+
The code that literally actually made this model is [here](https://github.com/AMindToThink/wmdp).
|
22 |
|
23 |
- **Developed by:** [More Information Needed]
|
24 |
- **Funded by [optional]:** [More Information Needed]
|