25 lines
813 B
Markdown
25 lines
813 B
Markdown
---
|
|
license: llama3.1
|
|
language:
|
|
- en
|
|
tags:
|
|
- abliterated
|
|
- uncensored
|
|
base_model:
|
|
- cognitivecomputations/Dolphin3.0-Llama3.1-8B
|
|
library_name: transformers
|
|
---
|
|
|
|
# huihui-ai/Dolphin3.0-Llama3.1-8B-abliterated
|
|
|
|
|
|
This is an uncensored version of [cognitivecomputations/Dolphin3.0-Llama3.1-8B](https://huggingface.co/cognitivecomputations/Dolphin3.0-Llama3.1-8B) 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.
|
|
|
|
## Use with ollama
|
|
|
|
You can use [huihui_ai/dolphin3-abliterated](https://ollama.com/huihui_ai/dolphin3-abliterated) directly
|
|
```
|
|
ollama run huihui_ai/dolphin3-abliterated
|
|
```
|