SageV2-8B 长上下文角色扮演模型发布
原标题：sinhal/SageV2-8B-LongContext-NSFW-RP
AI 摘要
Hugging Face 用户 sinhal 发布了 SageV2-8B-LongContext-NSFW-RP，一个基于 Meta-Llama-3.1-8B-Instruct-abliterated 微调的 8B 角色扮演模型，主打长对话中角色一致性与减少重复。作者用 40 轮固定脚本测试，称 6-gram 重叠从 35.0% 降至 9.3%，并给出 Q2_K 至 Q8_0 的 GGUF 量化版本及采样参数建议。该模型面向本地部署的角色扮演场景，属于社区开源发布。
正文节选
--- base_model: mlabonne/Meta-Llama-3.1-8B-Instruct-abliterated library_name: llama.cpp pipeline_tag: text-generation language: - en tags: - roleplay - llama-3.1 - gguf - unsloth - uncensored - abliterated - trl - sft - not-for-all-audiences license: llama3.1 model_name: SageV2-8B-LongContext-RP ---  # Sage V2 — 8B Long-Context Roleplay Most roleplay models are great for five turns. By turn thirty they're recycling the same sentence, every character sounds like t