what is your field Are you a ai artist ?
what is your field Are you a ai artist ?
FI am a PhD computer engineer
Fmy main programming language is c#
FI have 2 web based games
FMonster MMORPG
Monster MMORPG: Free Online Best Multiplayer Game for Pokemon Go Players. Play to capture, battle, train, PVP with 2,300+ unique Monsters on more than 500 maps
FPokemon Pets
PokemonPets: Free Online Multiplayer Pokémon Game for Pokemon Go Players. Play to capture, battle, train, PVP with your favorite Pokémon on more than 500 maps
MWOW
MI am 3d Artist and my brother is React,html.css,java developer
Mmy portfolio
SIs the art workflow based on stable diffusion?
F@Discord Admin I am having this problem, asked you some days ago in youtube comments, can you help me here please? 
FI was using Stable Difffusion normally like you demonstrated in the video, but today I got an error.
OutOfMemoryError Traceback (most recent call last)
<ipython-input-9-9817a4f7d2d0> in <module>
10
11 with autocast("cuda"), torch.inference_mode():
---> 12 images = pipe(
13 prompt,
14 height=height,
13 frames
/usr/local/lib/python3.8/dist-packages/diffusers/models/cross_attention.py in get_attention_scores(self, query, key, attention_mask)
221
222 attention_probs = attention_scores.softmax(dim=-1)
--> 223 attention_probs = attention_probs.to(dtype)
224
225 return attention_probs
OutOfMemoryError: CUDA out of memory. Tried to allocate 2.50 GiB (GPU 0; 14.76 GiB total capacity; 12.21 GiB already allocated; 1.22 GiB free; 12.26 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
Now I'm only able to generate images with 256 x 256 px and lower.
Any idea how to fix this and generate images of 512 x 512 again?
Thank you!
OutOfMemoryError Traceback (most recent call last)
<ipython-input-9-9817a4f7d2d0> in <module>
10
11 with autocast("cuda"), torch.inference_mode():
---> 12 images = pipe(
13 prompt,
14 height=height,
13 frames
/usr/local/lib/python3.8/dist-packages/diffusers/models/cross_attention.py in get_attention_scores(self, query, key, attention_mask)
221
222 attention_probs = attention_scores.softmax(dim=-1)
--> 223 attention_probs = attention_probs.to(dtype)
224
225 return attention_probs
OutOfMemoryError: CUDA out of memory. Tried to allocate 2.50 GiB (GPU 0; 14.76 GiB total capacity; 12.21 GiB already allocated; 1.22 GiB free; 12.26 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
Now I'm only able to generate images with 256 x 256 px and lower.
Any idea how to fix this and generate images of 512 x 512 again?
Thank you!
Fnot at all atm
Fthis is in google colab?
Fyes@
Fif so what is your batch size
Fyes!
Fshow me settings please
Fok
F!accelerate launch train_dreambooth.py \
--pretrained_model_name_or_path=$MODEL_NAME \
--pretrained_vae_name_or_path="stabilityai/sd-vae-ft-mse" \
--output_dir=$OUTPUT_DIR \
--revision="fp16" \
--with_prior_preservation --prior_loss_weight=1.0 \
--seed=1337 \
--resolution=512 \
--train_batch_size=1 \
--train_text_encoder \
--mixed_precision="fp16" \
--use_8bit_adam \
--gradient_accumulation_steps=1 \
--learning_rate=1e-6 \
--lr_scheduler="constant" \
--lr_warmup_steps=496 \
--num_class_images=744 \
--sample_batch_size=4 \
--max_train_steps=4960 \
--save_interval=10000 \
--save_sample_prompt="felipe" \
--concepts_list="concepts_list.json"
Reduce the
--pretrained_model_name_or_path=$MODEL_NAME \
--pretrained_vae_name_or_path="stabilityai/sd-vae-ft-mse" \
--output_dir=$OUTPUT_DIR \
--revision="fp16" \
--with_prior_preservation --prior_loss_weight=1.0 \
--seed=1337 \
--resolution=512 \
--train_batch_size=1 \
--train_text_encoder \
--mixed_precision="fp16" \
--use_8bit_adam \
--gradient_accumulation_steps=1 \
--learning_rate=1e-6 \
--lr_scheduler="constant" \
--lr_warmup_steps=496 \
--num_class_images=744 \
--sample_batch_size=4 \
--max_train_steps=4960 \
--save_interval=10000 \
--save_sample_prompt="felipe" \
--concepts_list="concepts_list.json"
Reduce the --save_interval
to lower than --max_train_steps
to save weights from intermediate steps.
--save_interval--max_train_steps--save_sample_prompt
can be same as --instance_prompt
to generate intermediate samples (saved along with weights in samples directory).
--save_sample_prompt--instance_promptFok this is training settings
Fyou get error during training?
Fno, I was using normally and suddenly one day it stopped working
Fworking only with 256 x 256 images or lower
Fok
Fshow me
Fwhere you get error
Fwhen you click generate images?
Fok, I'll make a video from scratch
Fjust 1 sec
Fok
Fby the way
Ffirst try
Fnum_samples = 1
Fah, ok!
MFelipe are u brazilian?
Fyep!
MShow
FTá funcionando pra vc ai direitinho?
Fworked with one sample
MEu não cheguei a usar esse colab ainda, utilizo geralmente o do automatic1111 mesmo
AGoogle Colab
Fbut why this time with only one sample?
Fand now it's not working even with 1 sample
GHey all! Trying to setup controlnet with Automatic as per the latest video
Gbut I'm hitting an error I'm not sure how to fix
Gonce I hit generate I get this in CMD
G