Skip to content
Yumeforge

Wan

Wan 2.1 14b LoRA Trainer

No ratings yet
  • Studio: Train a style
  • Makes: Trained style
  • Task: Style training

Teach a model your face, product or style with Wan 2.1 14b LoRA Trainer. Pay per result with credits. No subscription.

Open in studio

from 357 credits

This model works on real people. You confirm you have their consent before you start.

The same red-haired woman with freckles seen from three angles
Example from the Train a style studio, made with FLUX.2 Dev

What you can set

  • DataRequired

    Upload a file

    To train a WAN Lora, you need at least 10 face images to achieve good results.you can check out our default image dataset.

  • Learning rate

    Number from 0 to 1

  • Lora rank

    Number from 1 to 64

  • Steps

    Number from 1000 to 10000

    Number of steps to train the LoRA on.

  • Trigger word

    Text

    Optional trigger word. If a caption file exists, it is prepended when not already present. If no caption file exists and trigger_word is non-empty, a caption co

Reviews

From creators who made something with Wan 2.1 14b LoRA Trainer.

No ratings yet. Ratings come only from people who made something with this model.