Lambent/arsenic-12B-custom-heretic-1
Lambent/arsenic-12B-custom-heretic-1 is a 12 billion parameter language model, a decensored version of Lambent/arsenic-nemo-unleashed-12B. Created using the Heretic v1.0.1 tool, this model is specifically modified to reduce refusals compared to its original counterpart. It maintains a 32768 token context length and is intended for use cases requiring less restrictive content generation.
Loading preview...
Model Overview
Lambent/arsenic-12B-custom-heretic-1 is a 12 billion parameter language model derived from Lambent/arsenic-nemo-unleashed-12B. This version has been specifically modified using the Heretic v1.0.1 tool to achieve a significant reduction in content refusals.
Key Differentiators
- Decensored Output: Engineered to provide less restrictive responses compared to its base model.
- Reduced Refusals: Demonstrates a notable decrease in refusal rates, with 4 out of 55 refusals compared to 13 out of 55 for the original model.
- Abliteration Parameters: Specific parameters were adjusted during the decensoring process, including
direction_index(16.67),attn.o_proj.max_weight(1.06), andmlp.down_proj.max_weight(0.98).
Use Cases
This model is suitable for applications where a less constrained language model is desired, particularly for generating content that might otherwise be filtered or refused by more heavily moderated models. Its primary advantage lies in its ability to produce a wider range of responses by design.