CivArchive
    miraszalma - v1.0
    NSFW
    Preview 140290617

    LoRA was trained for only 1170 steps, so it may be slightly underfit. Strongly recommend to set the LoRA weight between 1.5–2, with 34 steps.

    该 LoRA 仅训练了 1170 步,因此可能存在轻微的欠拟合。强烈建议将 LoRA 权重设置为 1.5 到 2 之间,并使用 34 步进行生成。

    Description

    version1

    LORA
    Flux.1 D

    Details

    Downloads
    24
    Platform
    CivitAI
    Platform Status
    Available
    Created
    8/20/2026
    Updated
    8/25/2026
    Deleted
    -

    Files

    flux-lora-000012.safetensors

    Mirrors

    CivitAI (1 mirrors)