Official

arielreplicate / scalable_diffusion_with_transformers

Latent diffusion models, replacing the commonly-used U-Net backbone with a transformer that operates on latent patches

  • Public
  • 655 runs
  • T4
  • GitHub
  • Paper
  • License
  • Prediction

    arielreplicate/scalable_diffusion_with_transformers:089c3506e8fb6cdc7f7b0165cac893f1ae03e044fbff88ef7164d9f0b0dead85
    ID
    g7a6k4brknhrpmb2evkdfknw2m
    Status
    Succeeded
    Source
    Web
    Hardware
    Total duration
    Created
    by @arielreplicate

    Input

    cfg_scale
    4
    class_name
    centipede
    VAE_Decoder
    sd-vae-ft-mse
    num_outputs
    4
    DiT_resolution
    256x256
    num_sampling_steps
    250

    Output

    outputoutputoutputoutput
    Generated in
  • Prediction

    arielreplicate/scalable_diffusion_with_transformers:089c3506e8fb6cdc7f7b0165cac893f1ae03e044fbff88ef7164d9f0b0dead85
    ID
    a2zgmpyjwzawnayk63qxhu5zty
    Status
    Succeeded
    Source
    Web
    Hardware
    Total duration
    Created

    Input

    cfg_scale
    4
    class_name
    water ouzel, dipper
    VAE_Decoder
    sd-vae-ft-mse
    num_outputs
    2
    DiT_resolution
    512x512
    num_sampling_steps
    250

    Output

    outputoutput
    Generated in
  • Prediction

    arielreplicate/scalable_diffusion_with_transformers:089c3506e8fb6cdc7f7b0165cac893f1ae03e044fbff88ef7164d9f0b0dead85
    ID
    q735uuvsv5bepbupx4nmo554ke
    Status
    Succeeded
    Source
    Web
    Hardware
    Total duration
    Created

    Input

    cfg_scale
    "4"
    class_name
    tench, Tinca tinca
    VAE_Decoder
    sd-vae-ft-ema
    num_outputs
    9
    DiT_resolution
    256x256
    num_sampling_steps
    250

    Output

    outputoutputoutputoutputoutputoutputoutputoutputoutput
    Generated in
  • Prediction

    arielreplicate/scalable_diffusion_with_transformers:a11f7289b60685a4e48a0b09e41d091e27c2beb447877bc2117bfd6f9781c193
    ID
    staemf3zg5cmbprtlycrkjpoym
    Status
    Succeeded
    Source
    Web
    Hardware
    Total duration
    Created

    Input

    cfg_scale
    4
    class_name
    boxer
    VAE_Decoder
    sd-vae-ft-ema
    num_outputs
    1
    DiT_resolution
    512x512
    num_sampling_steps
    250

    Output

    output
    Generated in

Want to make some of these yourself?

Run this model