Sage V2 8B Roleplay
A Llama 3.1-derived roleplay fine-tune distributed in GGUF formats, with author recommendations for character cards and repetition-control settings.
What happened lately
Sage V2 8B Roleplay timeline
Sage V2 is sinhal’s roleplay fine-tune of mlabonne/Meta-Llama-3.1-8B-Instruct-abliterated. Its GGUF files were uploaded October 6, 2026; the model card was rewritten October 27 and updated October 28. The author documents local use with llama.cpp, SillyTavern and Ollama, including character-card prompts and sampler settings.
The model uses the Meta Llama 3.1 Community License. The author recommends Q6_K on an 8 GB GPU based on a specific RTX 4060 Laptop test; memory needs depend on context and runtime configuration. Repetition, speed and persona results are author-reported, not independently reproduced by AITrending. The card describes single-source synthetic training data, verbose responses and limited reasoning depth, and restricts its intended use to fictional roleplay excluding real people and minors.