Decensored using a custom training script guided by activations, similar to ablation/"abliteration" scripts but not exactly the same approach.
The training script is released under the MIT license: https://github.com/nkpz/DeLMAT
Very happy with how this turned out. Guardrails yeeted straight into the void.
- Downloads last month
- 8
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support