zylim0702 / sdxl-lora-customize-training

Introducing a Lora Instant Training model for crafting stunning 1024x1024 visuals. Train your own Lora Model via zip photos for instant outputs. Try Lora Model using this link: https://replicate.com/zylim0702/sdxl-lora-customize-model.

  • Public
  • 21.7K runs
  • L40S
  • Prediction

    zylim0702/sdxl-lora-customize-training:a4eefb9bbc944eaefc56d67acba982ab73d85ec449de6606cf48b13f1483a495
    ID
    5jgdk6tbddz5rcgpphbiinilea
    Status
    Succeeded
    Source
    Web
    Hardware
    A40 (Large)
    Total duration
    Created
    by @zylim0702

    Input

    ti_lr
    0.0003
    is_lora
    lora_lr
    0.0001
    verbose
    lora_rank
    32
    resolution
    768
    input_images
    barbie.zip
    lr_scheduler
    constant
    token_string
    TOK
    caption_prefix
    a photo of TOK,
    lr_warmup_steps
    100
    max_train_steps
    1000
    num_train_epochs
    4000
    train_batch_size
    4
    unet_learning_rate
    0.000001
    checkpointing_steps
    999999
    clipseg_temperature
    1
    input_images_filetype
    infer
    crop_based_on_salience
    use_face_detection_instead

    Output

    Generated in

Want to make some of these yourself?

Run this model