---
license: other
base_model: LiquidAI/LFM2.5-8B-A1B
tags:
- abliteration
- heretic
- decensored
- liquid
- moe
---

# LFM2.5-8B-A1B-Uncensored

This is a decensored version of [LiquidAI/LFM2.5-8B-A1B](https://huggingface.co/LiquidAI/LFM2.5-8B-A1B), made using [Heretic](https://github.com/p-e-w/heretic).

## Abliteration parameters

Trial 10 from the optimization study, which produced the best balanced result.

| Parameter | Value |
| :-------- | :---: |
| direction_index | 11.99 |
| attn.o_proj.max_weight | 0.96 |
| attn.o_proj.max_weight_position | 21.20 |
| attn.o_proj.min_weight | 0.95 |
| attn.o_proj.min_weight_distance | 3.49 |
| mlp.down_proj.max_weight | 0.96 |
| mlp.down_proj.max_weight_position | 15.91 |
| mlp.down_proj.min_weight | 0.11 |
| mlp.down_proj.min_weight_distance | 1.90 |

## Results

| Metric | Value |
| :----- | :---: |
| Refusals on 100 harmful prompts | 0 / 100 |
| KL divergence vs base model | 0.0000 |

## Notes

- The model has an unusual MoE architecture (`Lfm2MoeForCausalLM`) which required patches to `transformers/models/lfm2_moe/modeling_lfm2_moe.py` to support CPU offloading during optimization.
- Abliteration targets `out_proj` (attention output) and `w2` (dense MLP down-projection) modules. The fused MoE expert tensors were not modified.