base on Steer LLM outputs towards a certain topic/subject and enhance response capabilities using activation engineering by adding steering vectors # LLM Steer A Python module to steer LLM responses towards a certain topic/subject and to enhance capabilities (e.g., making it provide correct responses to tricky logical puzzles more often). A practical tool for using activation engineering by adding steer vectors to different layers of a Large Language Model (LLM). It should be used along with the transformers library. ## Demo - llm_steer v1: https://colab.research.google.com/github/Mihaiii/llm_steer/blob/main/demo/llm_steer_demo.ipynb - llm_steer v2: https://mihaiii-llm-steer.hf.space/ ## Basic usage Install it: `pip install llm_steer` Then use: ```python from llm_steer import Steer steered_model = Steer(model, tokenizer) ``` Add a steering vector on a particular layer of the model with a given coefficient and text. The coefficient can also be negative. ```python steered_model.add(layer_idx=20, coeff=0.4, text="logical") ``` Get all the applied steering vectors: ```python steered_model.get_all() ``` Remove all steering vectors to revert to initial model: ```python steered_model.reset_all() ``` ## Q / A Q: What's the difference between llm_steer and mentioning what you want in the system prompt? A: I see llm_steer as an enhancer. It can be used together with the system prompt. <br/> Q: How to determine the best parameters to be used? A: I don't have a method; it's all trial and error. I recommend starting middle layers and with a small coefficient and then slowly increase it. <br/> Q: What models are supported? A: I tested it on multiple architectures, including LLaMa, Mistral, Phi, StableLM. Keep in mind that llm_steer is meant to be used together with HuggingFace's transformers library, so it won't work on GGUF, for example. <br/> Q: I applied steering vectors, but the LLM outputs gibberish. What should I do? A: Try a lower coeff value or another layer. <br/> Q: Can I add multiple steering vectors on the same layer? Can I add the same steering vector on multiple layers? Can I add steering vectors with negative coefficients? A: Yes, and please do. llm_steer is built for experimenting. See the Colab for examples: https://colab.research.google.com/github/Mihaiii/llm_steer/blob/main/demo/llm_steer_demo.ipynb <br/> Q: Can I use steer vectors to enhance role-play characteristics (e.g., personas that are more funny or cocky)? A: Yes. <br/> Q: Can I use negative steering vectors to force it not to say "As an AI language model"? A: Yes. ## Credits / Thanks - [DL Explorers](https://www.youtube.com/@DLExplorers-lg7dt) for his video on [activation engineer](https://www.youtube.com/watch?v=J2Gx6FFEaRY&t=29s) which goes over [an article](https://www.greaterwrong.com/posts/5spBue2z2tw4JuDCx/steering-gpt-2-xl-by-adding-an-activation-vector) and [a colab he made](https://colab.research.google.com/github/githubpradeep/notebooks/blob/main/activation_engineering.ipynb). The resources mentioned in that video were the starting point of llm_steer. - Gary Bernhardt for his excellent [Python for programmers](https://www.executeprogram.com/courses/python-for-programmers) course. I needed a course that could help me go through the basics of Python without treating me like a dev noob (like most basic level tutorials treat their audience). - Andrej Karpathy for his [State of GPT](https://www.youtube.com/watch?v=bZQun8Y4L2A) video. I always wanted to make an open-source project, but there already was a repo for every idea I had. Not when it comes to tools for LLMs, though! ", Assign "at most 3 tags" to the expected json: {"id":"7286","tags":[]} "only from the tags list I provide: []" returns me the "expected json"