Lambent/Iris-12B-v1.1
Lambent/Iris-12B-v1.1 is a 12 billion parameter language model with a 32768 token context length, developed by Lambent. It is a fine-tuned version of the Iris model, specifically optimized to reduce generic "corporate AI disclaimers" and hedging in its responses. This model aims to maintain substantive skepticism where appropriate while delivering more direct and less evasive outputs, making it suitable for applications requiring straightforward communication.
Loading preview...
Overview
Lambent/Iris-12B-v1.1 is a 12 billion parameter language model, building upon the original Iris model. It features a substantial 32768 token context length, allowing for processing and generating longer, more complex texts. This version is a supervised fine-tune (SFT) specifically guided by the "Iris herself" persona, focusing on refining its conversational style.
Key Capabilities
- Reduced Corporate Hedging: The primary goal of this tune is to minimize generic "corporate AI disclaimers" and evasive language, promoting more direct and confident responses.
- Preserved Skepticism: While reducing hedging, the model is designed to retain substantive skepticism when it is logically warranted, aiming for a balanced and critical output.
- Creative Writing: The model's creative writing abilities are noted to be similar to the original Iris model, suggesting continued proficiency in generating imaginative content.
Good For
- Applications requiring direct and unambiguous communication from an AI.
- Use cases where avoiding overly cautious or generic AI responses is critical.
- Scenarios where a model needs to express skepticism based on reasoning rather than default disclaimers.
- Creative text generation tasks that benefit from the Iris model's established capabilities.