millicentli commited on
Commit
fb2385c
·
verified ·
1 Parent(s): acade10

Create README.md

Browse files
Files changed (1) hide show
  1. README.md +8 -0
README.md ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: meta-llama/Llama-3.1-8B-Instruct
3
+ library_name: peft
4
+ ---
5
+
6
+ This is the model used to invert Llama-3-8B activations using a T5-Base model (single activation inversion).
7
+
8
+ The model is used in [Do Activation Verbalization Methods Convey Privileged Information?](https://arxiv.org/abs/2509.13316) and the base repository used to train this model is from [vec2text](https://github.com/vec2text/vec2text). Read more of our paper + the underlying vec2text repo for training information.