| <!doctype html> |
| <html> |
| <head> |
| <meta charset="utf-8" /> |
| <meta name="viewport" content="width=device-width, initial-scale=1" /> |
| <title>SpaceRaven-Finetune</title> |
| <link rel="stylesheet" href="style.css" /> |
| </head> |
| <body> |
| <div class="stars"></div> |
| <div class="card"> |
| <div class="header-row"> |
| <h1>SpaceRaven-Finetune</h1> |
| <span class="badge">SFT + ORPO</span> |
| </div> |
| <p class="description"> |
| Production-grade SFT + ORPO fine-tuning scripts for small reasoning & instruction models. Built with Unsloth + TRL, stabilized for long thinking tokens on a single RTX 4090. |
| </p> |
| |
| <div class="stats-grid"> |
| <div class="stat-box"> |
| <span class="stat-label">Target Hardware</span> |
| <span class="stat-value">1x RTX 4090</span> |
| </div> |
| <div class="stat-box"> |
| <span class="stat-label">Stack</span> |
| <span class="stat-value">Unsloth + TRL</span> |
| </div> |
| </div> |
|
|
| <div class="info-section"> |
| <h3>Finetune Collection</h3> |
| <p> |
| Ministral-3 · Qwen3 · Qwen3.5 · Llama-3.2 · Phi-2<br> |
| More models will follow. |
| </p> |
| </div> |
|
|
| <div class="footer-links"> |
| <a href="https://github.com/RealRaven/SpaceRaven-Finetune" target="_blank" class="btn">View Github Repository</a> |
| </div> |
| </div> |
| </body> |
| </html> |
