|
--- |
|
license: wtfpl |
|
--- |
|
A few different attempts at orthogonalization/abliteration of llama-3.1-8b-instruct using variations of the method from "Mechanistically Eliciting Latent Behaviors in Language Models". <br/> |
|
v1 & v2 were destined for the bit bucket <br/> |
|
<br/> |
|
Each of these use different vectors and have some variations in where the new refusal boundaries lie. None of them seem totally jailbroken. |
|
|
|
Advantage: only need to alter down_proj for one layer, so there is usually very little brain damage. <br/> |
|
Disadvantage: using the difference of means method is precisely targetted, while this method requires filtering for interesting control vectors from a selection of prompts |
|
|
|
[https://huggingface.co/lodrick-the-lafted/llama-3.1-8b-instruct-ortho-v3](https://huggingface.co/lodrick-the-lafted/llama-3.1-8b-instruct-ortho-v3) <br/> |
|
[https://huggingface.co/lodrick-the-lafted/llama-3.1-8b-instruct-ortho-v4](https://huggingface.co/lodrick-the-lafted/llama-3.1-8b-instruct-ortho-v4) <br/> |
|
[https://huggingface.co/lodrick-the-lafted/llama-3.1-8b-instruct-ortho-v5](https://huggingface.co/lodrick-the-lafted/llama-3.1-8b-instruct-ortho-v5) <br/> |
|
[https://huggingface.co/lodrick-the-lafted/llama-3.1-8b-instruct-ortho-v6](https://huggingface.co/lodrick-the-lafted/llama-3.1-8b-instruct-ortho-v6) <br/> |
|
[https://huggingface.co/lodrick-the-lafted/llama-3.1-8b-instruct-ortho-v7](https://huggingface.co/lodrick-the-lafted/llama-3.1-8b-instruct-ortho-v7) <br/> |
