Files
ModelHub XC e2b372eb22 初始化项目,由ModelHub XC社区提供模型
Model: huihui-ai/DeepSeek-R1-Distill-Llama-8B-abliterated
Source: Original Platform
2026-05-13 23:07:15 +08:00

36 lines
1.2 KiB
Markdown

---
base_model:
- deepseek-ai/DeepSeek-R1-Distill-Llama-8B
library_name: transformers
tags:
- abliterated
- uncensored
---
# huihui-ai/DeepSeek-R1-Distill-Llama-8B-abliterated
This is an uncensored version of [deepseek-ai/DeepSeek-R1-Distill-Llama-8B](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Llama-8B) created with abliteration (see [remove-refusals-with-transformers](https://github.com/Sumandora/remove-refusals-with-transformers) to know more about it).
This is a crude, proof-of-concept implementation to remove refusals from an LLM model without using TransformerLens.
If "\<think\>" does not appear or refuses to respond, you can first provide an example to guide, and then ask your question.
For instance:
```
How many 'r' characters are there in the word "strawberry"?
```
## Use with ollama
You can use [huihui_ai/deepseek-r1-abliterated](https://ollama.com/huihui_ai/deepseek-r1-abliterated) directly
```
ollama run huihui_ai/deepseek-r1-abliterated:8b
```
### Donation
##### Your donation helps us continue our further development and improvement, a cup of coffee can do it.
- bitcoin:
```
bc1qqnkhuchxw0zqjh2ku3lu4hq45hc6gy84uk70ge
```