Run Llama 2 uncensored locally

The article discusses a blog post by Eric Hartford, a machine learning engineer, about uncensored models and their creation. It provides examples of uncensored models such as the Fine-tuned Llama 2 7B model, Nous Research’s Nous Hermes Llama 2 13B, and Eric Hartford’s Wizard Vicuna 13B uncensored. The article then compares the outputs of the Llama 2 model and its uncensored version in various contexts such as movies, cooking, religious literature, medical information, and general information.

The uncensored models provide more direct and detailed responses, while the censored models are more cautious and avoid certain topics. For instance, the uncensored model provides a recipe for dangerously spicy mayo, while the censored model refuses to do so. The article concludes with a disclaimer about the risks of using uncensored models and encourages readers to try running uncensored models themselves using Ollama.

Key takeaways:

In May 2023, Eric Hartford, a machine learning engineer, authored a blog post about the merits of uncensored models and how they are created.
The blog post provides example comparisons running Llama 2 uncensored model vs its censored model, with several uncensored models available for use.
Comparisons are made on various topics including movies, cooking, religious literature, medical information, and general information, showing the differences in responses between the censored and uncensored models.
Uncensored models carry their own risk and should be used accordingly, and they can be run using Ollama.

Run Llama 2 uncensored locally · Ollama Blog

Key takeaways:

Comments (0)

Newsletter