Home - News ( United States | United Kingdom | Italy | Germany ) - Football scores

Showing content from https://diffengine.readthedocs.io/en/latest/run_guides/run_t2i_adapter.html below:

Website Navigation

• diffengine

T2I Adapter Training — diffengine 1.0.0 documentation

You can also check configs/t2i_adapter/README.md file.

All configuration files are placed under the configs/t2i_adapter folder.

Following is the example config fixed from the stable_diffusion_xl_t2i_adapter_fill50k config file in configs/t2i_adapter/stable_diffusion_xl_t2i_adapter_fill50k.py:

from mmengine.config import read_base

with read_base():
    from .._base_.datasets.fill50k_t2i_adapter_xl import *
    from .._base_.default_runtime import *
    from .._base_.models.stable_diffusion_xl_t2i_adapter import *
    from .._base_.schedules.stable_diffusion_3e import *


optim_wrapper.update(
    optimizer=dict(lr=1e-5),
    accumulative_counts=2,
)

Run train

# single gpu
$ diffengine train ${CONFIG_FILE}
# multi gpus
$ NPROC_PER_NODE=${GPU_NUM} diffengine train ${CONFIG_FILE}

# Example
$ diffengine train stable_diffusion_xl_t2i_adapter_fill50k

Once you have trained a model, specify the path to the saved model and utilize it for inference using the diffusers.pipeline module.

import torch
from diffusers import StableDiffusionXLAdapterPipeline, T2IAdapter, AutoencoderKL
from diffusers.utils import load_image

checkpoint = 'work_dirs/stable_diffusion_xl_t2i_adapter_fill50k/step75000'
prompt = 'cyan circle with brown floral background'
condition_image = load_image(
    'https://github.com/okotaku/diffengine/assets/24734142/1af9dbb0-b056-435c-bc4b-62a823889191'
).resize((1024, 1024))

adapter = T2IAdapter.from_pretrained(
        checkpoint, subfolder='adapter', torch_dtype=torch.float16)

vae = AutoencoderKL.from_pretrained(
    'madebyollin/sdxl-vae-fp16-fix',
    torch_dtype=torch.float16,
)
pipe = StableDiffusionXLAdapterPipeline.from_pretrained(
    'stabilityai/stable-diffusion-xl-base-1.0', adapter=adapter, vae=vae, torch_dtype=torch.float16)
pipe.to('cuda')

image = pipe(
    prompt,
    image=condition_image,
    num_inference_steps=50,
).images[0]
image.save('demo.png')

You can check configs/t2i_adapter/README.md for more details.

RetroSearch is an open source project built by @garambo | Open a GitHub Issue

Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo

HTML: 3.2 | Encoding: UTF-8 | Version: 0.7.4