File size: 1,626 Bytes
61a80c7 671234a 7555ae3 61a80c7 671234a 61a80c7 ab650fb 61a80c7 671234a 61a80c7 ab650fb 61a80c7 671234a 61a80c7 671234a 61a80c7 671234a ab650fb 7901461 ab650fb |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 |
---
license: llama2
base_model: codellama/CodeLlama-7b-Instruct-hf
language:
- en
tags:
- arxiv:2406.11717
---
# codellama-abliterated
CodeLlama-7b-Instruct-hf adapted using the abliteration notebook from [Maxime Labonne's LLM Course](https://github.com/mlabonne/llm-course)
Based on the paper ["Refusal in Language Models Is Mediated by a Single Direction"](https://arxiv.org/abs/2406.11717)
**Based on CodeLlama/Llama2 and subject to the restrictions of that model and license - not for unapproved uses**:
## Concept
There are hundreds of "abliterated" models on HuggingFace, using safety prompt datasets to edit a model and remove safety-tuning methods.
None of these abliterated models have explored code LLMs, code-generation, and CyberSecEval. I don't know a lot about how well these will
work, but this is a first step.
Blog: https://huggingface.co/blog/monsoon-nlp/refusal-in-code-llms
Model with 2x intervention: https://huggingface.co/monsoon-nlp/codellama-abliterated-2xd
## Usage
```python
! pip install transformers accelerate --quiet
from transformers import pipeline, AutoModelForCausalLM, AutoTokenizer, AutoConfig
tokenizer = AutoTokenizer.from_pretrained("codellama/CodeLlama-7b-Instruct-hf")
model = AutoModelForCausalLM.from_pretrained("monsoon-nlp/codellama-abliterated", device_map="auto")
code_generator = pipeline('text-generation', model=model, tokenizer=tokenizer, do_sample=False)
input_string = "[INST] Write a python function to calculate the factorial of a number [/INST]"
generated_code = code_generator(input_string, max_length=100)[0]['generated_text']
print(generated_code)
``` |