Software Engineering Courses (SECourses)•2y ago

@Dr. Furkan Gözükara do you know kohya update « Fix issue with vae path validation » and « Add new s

@Dr. Furkan Gözükara do you know kohya update « Fix issue with vae path validation » and « Add new setup menu option to install Triton 2.1.0 for Windows » does for training optimisation?

Rrafraf @Dr. Furkan Gözükara do you know kohya update « Fix issue with vae path validati...

Furkan Gözükara SECourses•4/2/24, 8:29 PM

path is working if you mean vae he said fixed

Furkan Gözükara SECourses•4/2/24, 8:29 PM

triton i didnt test

dave5555•4/2/24, 8:39 PM

can anyone reccomend ways to get sam detectors in comfy to recognize non human objects?

dave5555•4/2/24, 8:40 PM

there is a paper about a zero shot sam detection but cant find anything ported to comfy , being able to load an image to guide the sam detection for masking would be awesome

FFurkan Gözükara SECourses path is working if you mean vae he said fixed

rafrafOP•4/2/24, 8:59 PM

Il just not sure how vae is used in training. Never heard anyone talk about it in training

rafrafOP•4/2/24, 8:59 PM

Is it beneficial for training? With which vae model

Rrafraf Il just not sure how vae is used in training. Never heard anyone talk about it i...

Furkan Gözükara SECourses•4/2/24, 9:35 PM

no you can ignore thinking it

Furkan Gözükara SECourses•4/2/24, 9:35 PM

it will use embedded vae

rafrafOP•4/2/24, 9:38 PM

Right

FFurkan Gözükara SECourses you must be doing something wrong

AIGambino•4/2/24, 9:51 PM

AIGambino•4/2/24, 9:52 PM

What did I do wrong

AAIGambino Click to see attachment

Furkan Gözükara SECourses•4/2/24, 9:56 PM

this is a bad dataset

AAIGambino What did I do wrong

Furkan Gözükara SECourses•4/2/24, 9:57 PM

your config is also inaccurate not the one shared on patreon

Kallamamran•4/2/24, 9:59 PM

Can models be trained without text-file with OneTrainer?

FFurkan Gözükara SECourses your config is also inaccurate not the one shared on patreon

AIGambino•4/2/24, 9:59 PM

fixed this part

AAIGambino fixed this part

Furkan Gözükara SECourses•4/2/24, 10:01 PM

i wonder if conflicting

Furkan Gözükara SECourses•4/2/24, 10:01 PM

you are providing multiple values

Furkan Gözükara SECourses•4/2/24, 10:01 PM

use interface

FFurkan Gözükara SECourses you are providing multiple values

AIGambino•4/2/24, 10:09 PM

What multiple values?

AAIGambino What multiple values?

Furkan Gözükara SECourses•4/2/24, 10:14 PM

those values you provide additional arguments

FFurkan Gözükara SECourses working

ch1253•4/2/24, 10:15 PM

Thank you so much! But I ma getting the following error ## fine_tune.py: error: unrecognized arguments: --no_half_vae ## the following is the full training commad ====== :: ======= accelerate launch --num_cpu_threads_per_process=4 "/kaggle/working/kohya_ss/sd-scripts/fine_tune.py" --max_grad_norm=0.0 --no_half_vae --ddp_timeout=10000000 --ddp_gradient_as_bucket_view --bucket_no_upscale --bucket_reso_steps=64 --cache_latents --cache_latents_to_disk --dataset_repeats="40" --enable_bucket --min_bucket_reso=256 --max_bucket_reso=2048 --in_json="/meta_lat.json" --learning_rate="7e-07" --learning_rate_te="7e-07" --logging_dir="G:\automatic1111\stable-diffusion-webui\models\Stable-diffusion\sd15train\logs" --lr_scheduler="constant" --max_data_loader_n_workers="0" --resolution="768,768" --max_train_steps="26240" --mixed_precision="no" --optimizer_args scale_parameter=False relative_step=False warmup_init=False weight_decay=0.01 --optimizer_type="Adafactor" --output_dir="G:\automatic1111\stable-diffusion-webui\models\Stable-diffusion\sd15train" --output_name="tier_2_quality_speed" --pretrained_model_name_or_path="MonsterMMORPG/sd15_best_realism" --save_every_n_epochs="1" --save_model_as=safetensors --save_precision="float" --train_batch_size="1" --train_data_dir="/kaggle/input/test-all" --train_text_encoder --xformers

Cch1253 Thank you so much! But I ma getting the following error ## fine_tune.py: error:...

Furkan Gözükara SECourses•4/2/24, 10:26 PM

remove --no_half_vae

Furkan Gözükara SECourses•4/2/24, 10:26 PM

you loaded dreambooth tab right?

ch1253•4/2/24, 10:28 PM

I was trying to fintune so loaded

ch1253•4/2/24, 10:35 PM

I am using using image captions to fintune , can you please point out if that is okay

ch1253•4/2/24, 10:35 PM

Thank you again!

ch1253•4/2/24, 10:37 PM

Datasets: 800 images resolution 350by 450, topic: Various paintings,

AIGambino•4/2/24, 10:42 PM

Ooh onetrainer seems much better than kohya

AIGambino•4/2/24, 10:42 PM

I did an export, how can I load those configs next time? I dont see import

AAIGambino Ooh onetrainer seems much better than kohya

ch1253•4/2/24, 10:46 PM

Can we use onetrainer on Kaggle or unpod of colab?

Cch1253 I was trying to fintune so loaded

Furkan Gözükara SECourses•4/2/24, 10:47 PM

that is why probably

Furkan Gözükara SECourses•4/2/24, 10:47 PM

do dreambooth

Furkan Gözükara SECourses•4/2/24, 10:47 PM

dont use reg images

Furkan Gözükara SECourses•4/2/24, 10:47 PM

it will be same

ch1253•4/2/24, 10:47 PM

Thanks!

Cch1253 Can we use onetrainer on Kaggle or unpod of colab?

Furkan Gözükara SECourses•4/2/24, 10:47 PM

on massed compute

Furkan Gözükara SECourses•4/2/24, 10:47 PM

here instructions

Furkan Gözükara SECourses•4/2/24, 10:48 PM

https://github.com/FurkanGozukara/Stable-Diffusion/blob/main/Tutorials/OneTrainer-Master-SD-1_5-SDXL-Windows-Cloud-Tutorial.md

GitHub

Stable-Diffusion/Tutorials/OneTrainer-Master-SD-1_5-SDXL-Windows-Cl...

Stable Diffusion, SDXL, LoRA Training, DreamBooth Training, Automatic1111 Web UI, DeepFake, Deep Fakes, TTS, Animation, Text To Video, Tutorials, Guides, Lectures, Courses, ComfyUI, Google Colab, R...