upgrade callback timeout to 600

fix: VRAM monitory for server and soft lazy loading

change: to safe flow model by presure

add: @asynccontextmanager

fix: leak memory from models

tasks_in_queue = max(0, tasks_in_queue - 1)

add: queue for customers

fix: prevent concurrent model access using global lock

move: to num_inference_steps=25, guidance_scale=5.5

fix: Sync ultralytics version to match training environment

new: best.pt v4 and downgrade all params fot test

move: to guidance_scale=6.5, strength=0.95

add: strength=0.89

move: to num_inference_steps=30, guidance_scale=6, strength=0.85

fix: PNDMScheduler

num_inference_steps=25, guidance_scale=4.8, strength=0.88

num_inference_steps=24, guidance_scale=3.8, strength=0.82

update: promt and num_inference_steps=28, guidance_scale=4.8, strength=0.86

transformers==4.37.2

update: diffusers==0.30.0

update: promts and dowgrade guidance_scale

add: stabilityai/stable-diffusion-2-inpainting

revert: to the b814393 and upgrade guidance_scale=3.8 and strength=0.6

iopaint[models]==1.2.1

huggingface-hub==0.21.4

remove all version from deps

transformers==4.35.1

diffusers==0.26.3

add: iopaint for lama

remove misspell ultralytics

ashawkey/Lama-inpainting

add: safetensors

add: lama-cleaner from lib

change: to parlance/lama-cleaner

add: lama for middle pipline logic

add: CannyDetector

add: _prepare_image_for_diffusion

add: numpy==1.26.4

feature: add ControlNet

add: strength=0.5

move to runwayml/stable-diffusion-inpainting

add system install git-lfs

git lfs pull stable-diffusion-2-inpainting

git clone stabilityai/stable-diffusion-2-inpainting

add: local download stabilityai/stable-diffusion-2-inpainting

download model

revert: to the 98f1dc6 and downgrade guidance_scale and num_inference_steps

fix: huggingface-hub>=0.34.0

diffusers >=0.30.0

fix: return_tensors