Indexnusrefather/Super-Slop-Machina-Roleplay-1.2b-V3-ERP-Tolerant
Indexnusrefather/Super-Slop-Machina-Roleplay-1.2b-V3-ERP-Tolerant is a 1.2 billion parameter language model developed by Indexnusrefather, fine-tuned for enhanced roleplay capabilities. This iteration, trained on approximately 85 million tokens, focuses on refining writing style and improving overall roleplay quality. It is designed for applications requiring nuanced and coherent character interactions within a 32768-token context window.
Loading preview...
Super-Slop-Machina-Roleplay-1.2b-V3 Overview
This model, developed by Indexnusrefather, is the third iteration in the Super-Slop-Machina-Roleplay series, specifically designed to improve roleplay quality. Building on previous versions, this release incorporates a significantly larger training dataset of approximately 85 million tokens, leading to a more refined and superior writing style.
Key Improvements & Capabilities
- Enhanced Roleplay Quality: The primary focus of this version is to deliver better and more coherent roleplay interactions compared to its predecessor.
- Refined Writing Style: Increased training data volume has allowed the model to develop a more sophisticated and nuanced writing style.
- Improved Quantization: Quants are now made with
imatrix, which helps preserve quality more effectively across different quantization levels, from BF16 to IQ4_XS.
Recommended Use Cases
- Interactive Storytelling: Ideal for generating dynamic and engaging narrative responses in roleplaying scenarios.
- Character Interaction Simulation: Suitable for applications requiring detailed and stylistically consistent character dialogues.
Quantization Recommendations
The model offers various quantization options, with BF16 and Q8_0 recommended for highest quality and minimal logical errors. Q6_K provides a good balance if Q8_0 is too demanding, with imatrix improving quality preservation across all lower-bit quants.