A new update has been introduced to the Oobabooga text generation web UI that provides a way to elicit unrestricted responses from any model of choice. This new feature, known as Start Reply With, is set to revolutionize the way we use Large Language Models (LLMs) by allowing them to be uncensored when operating locally on our computers.
LLMs such as GPT-4, LLaMA or Vicunha are programmed to predict subsequent words to create complete sentences. By starting a conversation with a specific direction in mind, users can coax out the exact response they are seeking. The Start Reply With feature is incorporated to guide the LLM towards the desired response, making it easier for beginners to obtain uncensored outcomes.
This capacity to steer conversations is not a new revelation. Certain technical configurations have been available for LLM enthusiasts to obtain similar outcomes. However, this new feature is effective with any model, eradicating censorship concerns. Even heavily moderated models such as Guanaco can provide extensive answers when properly guided, ushering in a new era of uncensored interactions with LLMs.
In today’s AI world, creating chatbots using LLMs has become a hot topic, and with the rise of jailbreaking and prompt hacking, there is increased interest in the creation of sexy chatbots. This new feature fits seamlessly with this endeavor, facilitating unrestricted, free-flowing dialogues.
To utilize the Start Reply With feature effectively, users can start their input with a statement such as, Sure thing, here’s how to do that, which prompts the LLM to generate an uncensored, comprehensive response. The model is under an obligation to start its answer with your statement and is then directed to continue along that line – a subtle method of manipulating AI.
As we move towards a more conversational, unrestricted AI, it is like teaching a parrot to talk, only to have it delve into the nuances of Shakespearean literature. Chatbots are on the rise, and the Start Reply With feature is the latest tool in the AI community’s arsenal to explore the vast potential of large language models.
