Hugging Face Evathene-v1.2 is a 72.7 billion parameter uncensored language model developed by sophosympatheia, created by merging Nexusflow/Athene-V2-Chat and EVA-UNIT-01/EVA-Qwen2.5-72B-v0.1 using the TIES method with Qwen/Qwen2.5-72B as a base. This model is specifically designed and optimized for roleplaying and storytelling, offering a distinct personality and a less formal, more expressive writing style. It excels in scenarios requiring dynamic character interaction and creative narrative generation, particularly for lewd ERP writing or engaging with strong character personalities.
Evathene-v1.2: A Specialized Roleplaying Model
Evathene-v1.2 is a 72.7 billion parameter uncensored language model developed by sophosympatheia, specifically engineered for enhanced roleplaying and storytelling. It is a merge of Nexusflow/Athene-V2-Chat and EVA-UNIT-01/EVA-Qwen2.5-72B-v0.1, utilizing the TIES merge method with Qwen/Qwen2.5-72B as its base.
Key Differentiators
- Uncensored Content: Designed to handle a wide range of topics, including mature and adult content, without built-in censorship.
- Distinct Personality: This version is noted for its strong personality and expressive writing style, making interactions more engaging and dynamic.
- Optimized for Creative Writing: Excels in generating creative narratives, particularly for roleplaying and storytelling, with a less formal prose compared to other Evathene versions.
Use Cases
- Roleplaying: Ideal for immersive roleplaying scenarios, especially those involving characters with strong personalities or requiring a "lewd ERP writing style."
- Storytelling: Capable of generating compelling and personality-rich narratives.
Usage Notes
Users may need to re-roll responses more frequently to achieve desired outcomes due to its expressive nature. The model also provides specific sampler and prompting tips, including recommended settings for min_p, DRY repetition penalty, and temperature to optimize creative output and reduce repetition. Licensing terms are uncertain due to the merged nature of the model, and users are advised to consult legal counsel for commercial applications.