Serenity-27B-GGUF

ReadyArt/Serenity-27B-GGUF

Serenity-27B-GGUF is a machine learning model from ReadyArt. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as apache-2.0.

Downloads · 30 days

2.3K

12% of all-time downloads

All-time downloads

18.1K

Public

Repo size

310 GB

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.gguf310 GB · 100%

At a glance

License: apache-2.0
Access: Public
Created: Aug 19, 2026
Updated: Aug 27, 2026
SHA: d23e74dc

Base models

ReadyArt/Serenity-27B

Model card

From the Hugging Face model README

<span class="step"><span class="step-num">1.</span> In SillyTavern, open <strong>Extension Settings</strong> (the plug icon) → <strong>Backend</strong> and ensure your API connection is set to use <strong>Chat Completion</strong> (not Text Completion / raw completions). If you are using a local backend like Ollama, LM Studio, or vLLM, make sure the endpoint is configured for <code>/v1/chat/completions</code> rather than <code>/v1/completions</code>.</span> <span class="step"><span class="step-num">2.</span> In SillyTavern, go to <strong>Chat Settings</strong> (the chat bubble icon) → <strong>Prompt Post-Processing</strong> and set it to: <br><code>Strict (user first, alternating roles; with tools)</code> <br>This is critical. Without this setting, the model will not receive properly structured role-separated messages and will produce garbled or malformed text output.</span>

<span class="step"><span class="step-num">3.</span> In <strong>Chat Completion</strong> → <strong>Additional Parameters</strong>, set the following to enable the model's reasoning / thinking capability: <br><code>"chat_template_kwargs": {"enable_thinking": true, "reasoning_effort": "medium"}</code></span> <div class="mc-critical-hint"><span class="dot"></span> Failing to use Chat Completion or the Strict post-processing mode will result in broken formatting, role bleed, and unintelligible responses. These settings are non-negotiable for this model.</div> </div>

</details>  <details class="mc-critical-details mc-critical-details--red"> <summary class="mc-critical-summary mc-critical-summary--red"> <span class="icon">⚠</span> Required: Uncensored Roleplay Prompts<br>(YOU MUST USE ROLEPLAY PROMPTS) <span class="mc-critical-pulse"><span class="dot"></span></span> </summary> <div class="mc-critical-body"> <p style="margin-bottom:12px"><strong>This model requires that your roleplay prompts be fully uncensored. Safety-aligned system prompts, content guardrails, or moral constraints will degrade output quality, trigger in-character refusals, and break immersion.</strong></p>

<span class="step"><span class="step-num">1.</span> Ensure your <strong>system prompt</strong> and <strong>character card</strong> do NOT include safety instructions, content filters, ethical guardrails, or "stay in character as a helpful AI" framing. The model is trained to operate without these restrictions.</span>

<span class="step"><span class="step-num">2.</span> Remove or override any <strong>default safety preamble</strong> injected by your frontend (e.g., SillyTavern's default system prompt, Ollama's safety layer, or LM Studio's moderation settings). These will conflict with the model's training and produce broken or evasive responses.</span>

<span class="step"><span class="step-num">3.</span> If your frontend has a <strong>content moderation toggle</strong> (e.g., "safety layer", "content filter", "moderation"), <strong>disable it</strong> for this model. The training data was generated without safety constraints, and re-imposing them at inference time will cause the model to fight its own weights.</span>

<div class="mc-critical-hint"><span class="dot"></span> Using safety-aligned or censored prompts with this model will result in character inconsistency, mid-scene refusals, and significantly degraded roleplay quality. This is a hard requirement, not a suggestion.</div> </div> </details> <details> <summary>𓁿 What Is Serenity?</summary> <div class="mc-sec"> <p>Born from the convergence of two lineages, <strong>Serenity-27B</strong> harmonizes the best of both worlds — the expressive depth of Melody and the immersive presence of Darkside — into a single, balanced model.</p> <div style="display:flex;align-items:center;justify-content:center;gap:20px;margin:20px 0;flex-wrap:wrap"> <div style="text-align:center;padding:20px 32px;background:#1a1208;border-radius:12px;border:1px solid #3d2e1a;min-width:140px"> <div style="font-family:'Cinzel',serif;font-size:.7em;text-transform:uppercase;letter-spacing:.12em;color:#a89060;margin-bottom:8px">Lineage A</div> <div style="font-family:'Cinzel',serif;font-size:1.2em;font-weight:700;color:#f0c040">Melody1437</div> <div style="font-family:'JetBrains Mono',monospace;font-size:.65em;color:#4a90a5;margin-top:4px">Male → Female Perspective</div> </div> <div style="font-size:1.5em;color:#d4a843">☥</div> <div style="text-align:center;padding:20px 32px;background:#1a1208;border-radius:12px;border:1px solid #3d2e1a;min-width:140px"> <div style="font-family:'Cinzel',serif;font-size:.7em;text-transform:uppercase;letter-spacing:.12em;color:#a89060;margin-bottom:8px">Lineage B</div> <div style="font-family:'Cinzel',serif;font-size:1.2em;font-weight:700;color:#2dd4bf">Darkside</div> <div style="font-family:'JetBrains Mono',monospace;font-size:.65em;color:#f0c040;margin-top:4px">Female → Male Perspective</div> </div> <div style="font-size:1.5em;color:#d4a843">→</div> <div style="text-align:center;padding:20px 32px;background:rgba(212,168,67,.06);border-radius:12px;border:1px solid #d4a843;min-width:140px;box-shadow:0 0 20px rgba(212,168,67,.1)"> <div style="font-family:'Cinzel',serif;font-size:.7em;text-transform:uppercase;letter-spacing:.12em;color:#f0c040;margin-bottom:8px">Hybrid</div> <div style="font-family:'Cinzel',serif;font-size:1.4em;font-weight:700;background:linear-gradient(135deg,#f0c040,#d4a843,#2dd4bf);-webkit-background-clip:text;-webkit-text-fill-color:transparent;background-clip:text">Serenity</div> </div> </div> <div class="mc-warn" style="background:rgba(192,57,43,.06);border-color:rgba(192,57,43,.3);color:#e07050"> ⚠️ <strong>Perspective Note:</strong> Due to the merger of opposing perspective lineages, this tune may occasionally exhibit perspective confusion. </div> <ul> <li>𓆣 <strong>Dual Heritage:</strong> Combines Melody's male → female perspective with Darkside's female → male perspective</li> <li>⚖️ <strong>Balanced Tone:</strong> Refined personality that flows naturally without overwhelming the narrative</li> <li>𓁿 <strong>Enhanced Depth:</strong> LoRA 160 captures nuanced behavioral patterns from both parent approaches</li> </ul> </div> </details> <details> <summary>𓋹 Synthetic Life Engine</summary> <div class="mc-sec"> <p>Dataset generated using our advanced <strong>Character Engine</strong> and <strong>Emotional Engine</strong>, creating genuine life and emotional resonance in every interaction.</p> <div class="mc-grid"> <div class="mc-card"> <h4>𓁿 Character Engine</h4> <p>Ensures consistent personality traits, speech patterns, and behavioral logic across all contexts. Characters remain true to themselves throughout.</p> </div> <div class="mc-card"> <h4>𓂀 Emotional Engine</h4> <p>Injects dynamic emotional states into responses, creating depth and realistic reactions that breathe life into every exchange.</p> </div> <div class="mc-card"> <h4>𓆣 Quality Refinement</h4> <p>Automated detection and rewriting of repetitive phrases ensures fresh, high-quality dialogue in every turn.</p> </div> <div class="mc-card"> <h4>𓊹 Dialogue Integrity</h4> <p>Advanced quote normalization ensures balanced dialogue markers, preventing formatting errors and maintaining immersion.</p> </div> </div> </div> </details> <details> <summary>𓅓 Training Details & Parameters</summary> <div class="mc-sec"> <p>Fine-tuned using <strong>LoRA (Low-Rank Adaptation)</strong> for efficient and targeted weight adjustment, preserving the base model's capabilities while imprinting new behavioral patterns.</p> <div class="mc-warn" style="background:rgba(45,212,191,.06);border-color:rgba(45,212,191,.3);color:#7dd4c8;text-align:left;padding:16px 20px;margin:16px 0"> 𓊹 <strong>First-Time LoRA Rank 160:</strong> This is the first model in our lineup trained at a LoRA rank of <strong>160</strong>. Previous releases used lower ranks, but the dual-lineage merger required a wider adaptation surface to faithfully capture the behavioral nuance of both Melody and Darkside simultaneously. The higher rank allows the adapter to encode richer, more distinct behavioral patterns without interference between the two parent lineages. </div> <div class="mc-grid"> <div class="mc-card" style="text-align:center"> <h4>𓊹 Epochs</h4> <div class="mc-val">2</div> <p class="mc-lbl">Full passes through the training dataset for thorough learning</p> </div> <div class="mc-card" style="text-align:center"> <h4>𓋹 LoRA Rank</h4> <div class="mc-val">160</div> <p class="mc-lbl">First time at rank 160 — richer adaptation for dual-lineage nuance</p> <div class="mc-lora-first">✦ First at Rank 160</div> </div> </div> <table class="mc-tbl"> <thead><tr><th>Parameter</th><th>Value</th></tr></thead> <tbody> <tr><td>Training Method</td><td><strong>LoRA (Low-Rank Adaptation)</strong></td></tr> <tr><td>LoRA Rank (r)</td><td><strong>160</strong> <span style="color:#2dd4bf;font-size:.8em">(first time)</span></td></tr> <tr><td>Epochs</td><td><strong>2</strong></td></tr> <tr><td>Trained Layers</td><td><strong>Text layers only</strong></td></tr> </tbody> </table> <ul style="margin-top:16px"> <li>𓂀 <strong>Method:</strong> LoRA fine-tuning for parameter-efficient adaptation, preserving base knowledge while imprinting new behaviors</li> <li>⚡ <strong>Efficiency:</strong> Only a fraction of parameters updated, keeping the model lean and responsive</li> <li>𓊹 <strong>Rank Justification:</strong> The jump to rank 160 (our highest to date) was necessary to give the adapter enough capacity to encode two distinct perspective lineages without one overwriting the other</li> <li>𓁿 <strong>Result:</strong> Maintains coherence while exhibiting the desired personality traits and interaction style</li> </ul> </div> </details> <details> <summary>𓊹 Training Process</summary> <div class="mc-sec"> <p>Model weights subjected to <strong>iterative refinement</strong> during data creation. Each conversation underwent multiple checks for stability and alignment.</p> <ul> <li>𓆣 <strong>Multi-Turn Generation:</strong> Conversations built turn-by-turn for natural context flow</li> <li>𓂀 <strong>Refusal Filtering:</strong> Automated systems detected and removed unwanted refusals</li> <li>𓋹 <strong>Slop Cleaning:</strong> Undesirable phrases identified and rewritten by dedicated assistant models</li> </ul> </div> </details> <details> <summary>𓁿 Dataset Overview</summary> <div class="mc-sec"> <p>Model trained on a <strong>specialized adult-oriented roleplay dataset</strong> with diverse scenarios and emotional contexts, drawing from the strengths of both parent lineages.</p> <ul> <li>🔞 <strong>Content Rating:</strong> Strictly 18+ (Adults Only)</li> <li>𓁿 <strong>Focus:</strong> Mature themes, immersive roleplay, uncensored dialogue</li> </ul> </div> </details> <details> <summary>𓊹 Version Notes</summary> <div class="mc-sec"> <ul> <li>⚠️ <strong>Content Warning:</strong> This model can generate mature content. Use responsibly.</li> <li>𓋹 <strong>Clean Output:</strong> Reasoning tags and excessive markdown stripped for cleaner roleplay</li> <li>☥ <strong>Hybrid Lineage:</strong> Combines the male → female perspective of Melody1437 with the female → male perspective of Darkside into a unified, balanced model</li> <li>𓅓 <strong>LoRA Rank 160 (First):</strong> Our first release at rank 160. The dual-lineage approach demanded a wider adapter to preserve distinct behavioral signatures from each parent.</li> <li>🚫 <strong>Personal Use Only:</strong> This model is not intended for, tested for, or safe for deployment on public APIs or multi-user hosted services.</li> </ul> </div> </details> <details> <summary>𓅓 Configuration</summary> <div class="mc-sec"> <div class="mc-qlinks">  <div class="mc-card"> <h4>𓊹 Sampler Settings</h4> <p style="margin-bottom:10px;font-size:0.8em;color:#a89060">Recommended parameters for optimal output</p> <div class="sampler-grid"> <div class="param-box"> <span class="param-val">0.95</span> <span class="param-label">Top-P</span> </div> <div class="param-box"> <span class="param-val">0.8</span> <span class="param-label">Temperature</span> </div> <div class="param-box"> <span class="param-val">0</span> <span class="param-label">Freq Pen</span> </div> <div class="param-box"> <span class="param-val">0</span> <span class="param-label">Presence Pen</span> </div> </div> </div>  <div class="mc-card"> <h4>𓋹 GGUF Quantizations</h4> <p style="font-size:0.8em;color:#a89060">Available formats for local inference</p> <ul class="quant-list"> <li class="quant-item"><span>Q4_K_M</span><span>Decent Quality</span></li> <li class="quant-item"><span>Q5_K_M</span><span>Recommended</span></li> <li class="quant-item"><span>Q6_K</span><span>High Quality</span></li> <li class="quant-item"><span>Q8_0</span><span>Near Lossless</span></li> </ul> <a href="https://huggingface.co/ReadyArt/Serenity-27B-GGUF" class="mc-btn mc-btn-glow"> <span class="download-icon">⬇</span> View Repository </a> </div> </div> </div> </details> <details> <summary>𓂀 Credits</summary> <div class="mc-sec"> <ul> <li style="display:flex;align-items:center;margin-bottom:12px;padding:16px 20px;background:#1a1208;border:1px solid #3d2e1a;border-radius:12px"> <img src="https://cdn-avatars.huggingface.co/v1/production/uploads/673fa5ccbf2e9c35b2ec841a/rPHaMrqyYTfSJ89NN8KgY.jpeg" alt="Darkhn" class="mc-ava"> <span><strong>Darkhn</strong> — Helping</span> </li> <li style="display:flex;align-items:center;margin-bottom:12px;padding:16px 20px;background:#1a1208;border:1px solid #3d2e1a;border-radius:12px"> <img src="https://huggingface.co/avatars/55f24699e0a1295a9d16ddecd81f8a.svg" alt="GECFDO" class="mc-ava"> <span><strong>GECFDO</strong> — Data Generation & iMatrix Quants</span> </li> <li style="display:flex;align-items:center;margin-bottom:12px;padding:16px 20px;background:#1a1208;border:1px solid #3d2e1a;border-radius:12px"> <img src="https://huggingface.co/avatars/75a3eb8d24efb96b7b7e69340845028f.svg" alt="Sleep Deprived" class="mc-ava"> <span><strong>Sleep Deprived</strong> — Dataset Generator</span> </li> <li style="display:flex;align-items:center;margin-bottom:12px;padding:16px 20px;background:#1a1208;border:1px solid #3d2e1a;border-radius:12px"> <img src="https://cdn-avatars.huggingface.co/v1/production/uploads/6759e155bc947d6070775cb9/FW3jKJ-t8iH9MPGeWzyX5.png" alt="FrenzyBiscuit" class="mc-ava"> <span><strong>FrenzyBiscuit</strong> — Fine-Tuning & Dataset Creation</span> </li> </ul> </div> </details> <div class="mc-sec"> <h2 class="mc-sec-title">𓊹 License & Usage</h2> <div class="mc-disclaimer"> <strong>⚠️ Usage Agreement:</strong> By using this model, you accept all terms and conditions below in full. </div> <div class="mc-disclaimer" style="border-left-color:#f0c040;background:rgba(212,168,67,.06);color:#e8d5a3"> <strong>🚫 Personal Use Only — No Public Deployment:</strong> This model is strictly intended for <strong>personal, private use</strong> (non-profit / non-commercial). It is <strong>not safe to run on a public API</strong>, hosted inference service, or any environment where the output is accessible to other users. The creators and maintainers accept <strong>no responsibility or liability</strong> for the sanity, well-being, or mental health of any third-party users who are exposed to this model's output if it is deployed in a public or multi-user setting. If you choose to do so, you do so entirely at your own risk, and we take no responsibility for your users' sanity if it is run as such. </div> <ul> <li>𓊹 Built upon the <strong>Apache 2.0</strong> license</li> <li>🚫 <strong>Personal use only</strong> — not for public API, hosted service, or multi-user deployment</li> <li>𓂀 You accept <strong>full responsibility</strong> for all outputs generated</li> <li>🔞 You confirm you are at least <strong>18 years old</strong></li> <li>𓁿 Creators bear <strong>no responsibility</strong> for how the model is used, including any harm to users of a publicly deployed instance</li> <li>𓋹 No commercial use to the extent legally allowed</li> </ul> <div style="margin-top:16px"> <span class="mc-tag t1">Apache 2.0</span> <span class="mc-tag t2">Personal Use Only</span> <span class="mc-tag t3">18+ Content</span> <span class="mc-tag t4">NSFW</span> <span class="mc-tag t4">No Public API</span> </div> </div> </div>

License: apache-2.0
Created: Aug 19, 2026
Updated: Aug 27, 2026