zerofata/L3.3-GeneticLemonade-Final-v2-70B
The zerofata/L3.3-GeneticLemonade-Final-v2-70B is a 70 billion parameter language model, fine-tuned from zerofata/L3.3-GeneticLemonade-Unleashed-70B, with a 32768 token context length. This model is specifically designed for creative, character-driven roleplay (RP) and erotic roleplay (ERP), aiming to produce longer, narrative-heavy responses with accurate and proactive character portrayal. It features reduced positivity bias and a distinct writing style compared to its predecessor, making it suitable for nuanced interactive storytelling.
Loading preview...
Model Overview
zerofata/L3.3-GeneticLemonade-Final-v2-70B is a 70 billion parameter model, fine-tuned from the zerofata/L3.3-GeneticLemonade-Unleashed-70B base model. It is specifically developed as an uncensored creative model, excelling in character-driven roleplay (RP) and erotic roleplay (ERP).
Key Capabilities
- Character-Driven Narrative: Designed to generate longer, narrative-heavy responses where characters are portrayed accurately and proactively.
- Reduced Positivity Bias: Offers a more balanced and less overtly positive tone compared to previous iterations, leading to a more nuanced writing style.
- Creative Writing: Intended for scenarios requiring imaginative and detailed storytelling.
Training Details
The model was trained on a dataset comprising approximately 4.3 million tokens, including 700 RP conversations, 2000 creative writing/instruct samples, and about 400 summaries. Notably, it did not incorporate the DPO dataset used in some prior models.
Usage Recommendations
For optimal performance, specific SillyTavern settings are suggested, including a temperature range of 0.9-1, MinP of 0.03-0.04, and TopP of 0.9-1.0. It utilizes the Llama-3-Instruct-Names chat template, requiring the "System same as user" option to be unchecked.
Top 3 parameter combinations used by Featherless users for this model. Click a tab to see each config.