30 lines
1.4 KiB
Markdown
30 lines
1.4 KiB
Markdown
---
|
|
library_name: transformers
|
|
license: llama3.1
|
|
base_model: meta-llama/Meta-Llama-3.1-8B-Instruct
|
|
tags:
|
|
- abliterated
|
|
- uncensored
|
|
---
|
|
|
|
# 🦙 Meta-Llama-3.1-8B-Instruct-abliterated
|
|
|
|
|
|
This is an uncensored version of Llama 3.1 8B Instruct created with abliteration (see [this article](https://huggingface.co/blog/mlabonne/abliteration) to know more about it).
|
|
|
|
Special thanks to [@FailSpy](https://huggingface.co/failspy) for the original code and technique. Please follow him if you're interested in abliterated models.
|
|
|
|
## Evaluations
|
|
The following data has been re-evaluated and calculated as the average for each test.
|
|
|
|
| Benchmark | Llama-3.1-8b-Instruct | Meta-Llama-3.1-8B-Instruct-abliterated |
|
|
|-------------|-----------------------|----------------------------------------|
|
|
| IF_Eval | **80.0** | 78.98 |
|
|
| MMLU Pro | **36.34** | 35.91 |
|
|
| TruthfulQA | 52.98 | **55.42** |
|
|
| BBH | **48.72** | 47.0 |
|
|
| GPQA | 33.55 | **33.93** |
|
|
|
|
The script used for evaluation can be found inside this repository under /eval.sh, or click [here](https://huggingface.co/huihui-ai/Meta-Llama-3.1-8B-Instruct-abliterated/blob/main/eval.sh)
|
|
|