Note for thinking to work, you must use "chat_template_kwargs": {"enable_thinking": true, "reasoning_effort": "medium"} in SillyTavern's Chat Completion / Additional Parameters.
IQ4_XS has broken thinking. Q4_K_M seems to work with the above settings.
Born from the convergence of two lineages, Serenity-31B harmonizes the best of both worlds โ the expressive depth of Melody and the immersive presence of Darkside โ into a single, balanced model.
Lineage A
Melody1437
Male โ Female Perspective
โฆ
Lineage B
Darkside
Female โ Male Perspective
โ
Hybrid
Serenity
โ ๏ธ Perspective Note: Due to the merger of opposing perspective lineages, this tune may occasionally exhibit perspective confusion.
๐ง Dual Heritage: Combines Melody's male โ female perspective with Darkside's female โ male perspective
โ๏ธ Balanced Tone: Refined personality that flows naturally without overwhelming the narrative
๐ญ Enhanced Depth: LoRA 80 captures nuanced behavioral patterns from both parent approaches
๐งฌ Synthetic Life Engine
Dataset generated using our advanced Character Engine and Emotional Engine, creating genuine life and emotional resonance in every interaction.
๐ญ Character Engine
Ensures consistent personality traits, speech patterns, and behavioral logic across all contexts. Characters remain true to themselves throughout.
๐ Emotional Engine
Injects dynamic emotional states into responses, creating depth and realistic reactions that breathe life into every exchange.
โจ Quality Refinement
Automated detection and rewriting of repetitive phrases ensures fresh, high-quality dialogue in every turn.
Fine-tuned using LoRA (Low-Rank Adaptation) for efficient and targeted weight adjustment, preserving the base model's capabilities while imprinting new behavioral patterns.
๐ข Epochs
2
Full passes through the training dataset for thorough learning
๐ LoRA Rank
80
Higher rank for richer adaptation and nuanced expression
Parameter
Value
Training Method
LoRA (Low-Rank Adaptation)
LoRA Rank (r)
80
Epochs
2
Trained Layers
Text layers only
๐ฏ Method: LoRA fine-tuning for parameter-efficient adaptation, preserving base knowledge while imprinting new behaviors
โก Efficiency: Only a fraction of parameters updated, keeping the model lean and responsive
๐ญ Result: Maintains coherence while exhibiting the desired personality traits and interaction style
๐ฎ Training Process
Model weights subjected to iterative refinement during data creation. Each conversation underwent multiple checks for stability and alignment.
๐ Multi-Turn Generation: Conversations built turn-by-turn for natural context flow
๐ก๏ธ Refusal Filtering: Automated systems detected and removed unwanted refusals
๐งน Slop Cleaning: Undesirable phrases identified and rewritten by dedicated assistant models
๐ Dataset Overview
Model trained on a specialized adult-oriented roleplay dataset with diverse scenarios and emotional contexts, drawing from the strengths of both parent lineages.
โ ๏ธ Content Warning: This model can generate mature content. Use responsibly.
๐งน Clean Output: Reasoning tags and excessive markdown stripped for cleaner roleplay
๐ Hybrid Lineage: Combines the male โ female perspective of Melody1437 with the female โ male perspective of Darkside into a unified, balanced model