Run the official Stable Diffusion releases in a Docker container with txt2img, img2img, depth2img, pix2pix, upscale4x, and inpaint.
License: GNU Affero General Public License v3.0
Maybe use another base docker instead of tensorflow(
stable-diffusion-docker/Dockerfile
Line 1 in 2e4a48a
Thank you for providing this, it is really exciting to have a simpler way to run SD using docker.
I tried this command and see this error. Is my GPU in a weird state perhaps? I'm providing the result of nvidia-smi beneath this error message.
$./build.sh run --W 256 --H 256 --half --attention-slicing --prompt 'abstract art'
load pipeline start: 2022-10-11T20:08:43.514057
/usr/local/lib/python3.8/dist-packages/torch/cuda/__init__.py:83: UserWarning: CUDA initialization: Unexpected error from cudaGetDeviceCount(). Did you run some cuda functions before calling NumCudaDevices() that might have already set an error? Error 34: CUDA driver is a stub library (Triggered internally at ../c10/cuda/CUDAFunctions.cpp:109.)
return torch._C._cuda_getDeviceCount() > 0
Fetching 16 files: 100%|██████████| 16/16 [00:00<00:00, 25448.94it/s]
ftfy or spacy is not installed using BERT BasicTokenizer instead of ftfy.
Traceback (most recent call last):
File "/usr/local/bin/docker-entrypoint.py", line 174, in <module>
main()
File "/usr/local/bin/docker-entrypoint.py", line 157, in main
stable_diffusion(
File "/usr/local/bin/docker-entrypoint.py", line 37, in stable_diffusion
pipe = StableDiffusionPipeline.from_pretrained(
File "/usr/local/lib/python3.8/dist-packages/diffusers/pipeline_utils.py", line 179, in to
module.to(torch_device)
File "/usr/local/lib/python3.8/dist-packages/torch/nn/modules/module.py", line 927, in to
return self._apply(convert)
File "/usr/local/lib/python3.8/dist-packages/torch/nn/modules/module.py", line 579, in _apply
module._apply(fn)
File "/usr/local/lib/python3.8/dist-packages/torch/nn/modules/module.py", line 579, in _apply
module._apply(fn)
File "/usr/local/lib/python3.8/dist-packages/torch/nn/modules/module.py", line 579, in _apply
module._apply(fn)
[Previous line repeated 1 more time]
File "/usr/local/lib/python3.8/dist-packages/torch/nn/modules/module.py", line 602, in _apply
param_applied = fn(param)
File "/usr/local/lib/python3.8/dist-packages/torch/nn/modules/module.py", line 925, in convert
return t.to(device, dtype if t.is_floating_point() or t.is_complex() else None, non_blocking)
File "/usr/local/lib/python3.8/dist-packages/torch/cuda/__init__.py", line 217, in _lazy_init
torch._C._cuda_init()
RuntimeError: Unexpected error from cudaGetDeviceCount(). Did you run some cuda functions before calling NumCudaDevices() that might have already set an error? Error 34: CUDA driver is a stub library
$ nvidia-smi
Tue Oct 11 20:10:52 2022
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 470.63.01 Driver Version: 470.63.01 CUDA Version: 11.4 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|===============================+======================+======================|
| 0 NVIDIA GeForce ... Off | 00000000:01:00.0 Off | N/A |
| 0% 37C P8 6W / 120W | 100MiB / 6075MiB | 0% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
| 0 N/A N/A 2125 G ...xorg-server-1.20.11/bin/X 66MiB |
| 0 N/A N/A 2171 G ...hell-40.1/bin/gnome-shell 31MiB |
+-----------------------------------------------------------------------------+
I am getting the below torch CUDA 11.7 install error when attempting to either call ./build.sh or docker build -f Dockerfile -t stable-diffusion:v2 --no-cache .. Has anyone encountered this / would have any pointers on a way around this? It seems the --extra-index-url is being ignored / is not being pulled from, but looking at different SO's / gists am not able to find a good fix for this. Any pointers super appreciated!
Docker build logs:
philippeheitzmann@Philippes-Laptop stable-diffusion-docker % docker build -f Dockerfile -t stable-diffusion-webui:v2 --no-cache .
[+] Building 7.1s (9/14)
=> [internal] load build definition from Dockerfile 0.0s
=> => transferring dockerfile: 599B 0.0s
=> [internal] load .dockerignore 0.0s
=> => transferring context: 34B 0.0s
=> [internal] load metadata for docker.io/library/python:3.10-slim-bullseye 0.8s
=> [auth] library/python:pull token for registry-1.docker.io 0.0s
=> CACHED [1/9] FROM docker.io/library/python:3.10-slim-bullseye@sha256:55b99191de1e61f0f95464ec44e656939abd150a0a4d9aa497bebbfa4e22b554 0.0s
=> [internal] load build context 0.0s
=> => transferring context: 105B 0.0s
=> [2/9] RUN rm -rf /usr/local/cuda/lib64/stubs 0.2s
=> [3/9] COPY requirements.txt / 0.0s
=> ERROR [4/9] RUN pip install -r requirements.txt --extra-index-url https://download.pytorch.org/whl/cu117 6.0s
------
> [4/9] RUN pip install -r requirements.txt --extra-index-url https://download.pytorch.org/whl/cu117:
#0 1.127 Looking in indexes: https://pypi.org/simple, https://download.pytorch.org/whl/cu117
#0 1.507 Collecting diffusers[torch]==0.16.1
#0 1.645 Downloading diffusers-0.16.1-py3-none-any.whl (934 kB)
#0 2.323 ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 934.9/934.9 kB 1.4 MB/s eta 0:00:00
#0 2.601 Collecting onnxruntime==1.15.0
#0 2.614 Downloading onnxruntime-1.15.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl (5.4 MB)
#0 5.013 ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 5.4/5.4 MB 2.2 MB/s eta 0:00:00
#0 5.178 Collecting safetensors==0.3.1
#0 5.196 Downloading safetensors-0.3.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl (1.2 MB)
#0 5.614 ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1.2/1.2 MB 3.0 MB/s eta 0:00:00
#0 5.763 ERROR: Could not find a version that satisfies the requirement torch==2.0.1+cu117 (from versions: 1.10.2, 1.11.0, 1.12.0, 1.12.1, 1.13.0, 1.13.1, 2.0.0, 2.0.1)
#0 5.763 ERROR: No matching distribution found for torch==2.0.1+cu117
#0 5.949
#0 5.949 [notice] A new release of pip is available: 23.0.1 -> 23.1.2
#0 5.949 [notice] To update, run: pip install --upgrade pip
------
ERROR: failed to solve: executor failed running [/bin/sh -c pip install -r requirements.txt --extra-index-url https://download.pytorch.org/whl/cu117]: exit code: 1
Dockerfile:
FROM python:3.10-slim-bullseye
RUN rm -rf /usr/local/cuda/lib64/stubs
COPY requirements.txt /
# RUN apt-get install python3-pip
RUN pip install -r requirements.txt --extra-index-url https://download.pytorch.org/whl/cu117
RUN useradd -m huggingface
USER huggingface
WORKDIR /home/huggingface
ENV USE_TORCH=1
RUN mkdir -p /home/huggingface/.cache/huggingface \
&& mkdir -p /home/huggingface/input \
&& mkdir -p /home/huggingface/output
COPY docker-entrypoint.py /usr/local/bin
COPY token.txt /home/huggingface
ENTRYPOINT [ "docker-entrypoint.py" ]
Environment:
M1 Macbook
Docker version 23.0.2, build 569dd73db1
Hi
first of all, great work. I love it.
I would like to change the default download-location to another drive, because I'm running out of storage space.
So far I was unable to change the location. Any help would be great
I'm using Ubuntu 20 and a rx 6900 XT.
I tried setting --extra-index-url in Dockerfile to https://download.pytorch.org/whl/rocm5.1.1 which I believe is correct for AMD systems.
I then had to modify the run script to be
docker run --rm \
--group-add $(getent group render | cut -d':' -f 3) \
--device=/dev/kfd --device=/dev/dri --security-opt seccomp=unconfined \
-v huggingface:/home/huggingface/.cache/huggingface \
-v "$PWD"/output:/home/huggingface/output \
"$CWD" "$@"`Which was suggested by https://github.com/RadeonOpenCompute/ROCm-docker
Then it works.
I'm not sure if this is expected, but it takes just under 2 min to generate an image. I can hear gpu affect my speakers (the same way games running at a high fps can do) the last 20 sec which could be a hint at some gpu usage at the end. It happens when the progress bar iteration thing shows up.
The original txt2img.py script is rapidly becoming outdated and some of the options from txt2img.py are hard to remember. Since then, stable-diffusion-docker has continued to add newer pipelines from diffusers like onnx, upscale4x and inpaint, add newer features like negative prompts and attention slicing, support newer models like stable diffusion 1.5, 2.0, and 2.1, and simplify and standardize options where possible.
In the future, some of the original txt2img options will be renamed for easy-of-use and compatibility with other tools and pipelines:
| txt2img | stable-diffusion-docker |
|---|---|
--H |
--height |
--W |
--width |
--n_iter |
--iters |
--n_samples |
--samples |
--ddim_steps |
--steps |
GPU is a 1660, I checked and drivers are all up to date, torch seems to recognize GPU correctly. I'm running Windows 11, Docker runs under WSL2.
I tried with different parameters, tried onxx, doesn't work any of them (with onxx it doesn't start the iteration process, just ends and gives black image).
Here is what I get in terminal:
./build.sh run --height 256 --width 256 --attention-slicing --xformers-memory-efficient-attention --half --skip "abstract art"
load pipeline start: 2023-04-24T13:04:55.748155
vae/diffusion_pytorch_model.safetensors not found
`text_config_dict` is provided which will be used to initialize `CLIPTextConfig`. The value `text_config["id2label"]` will be overriden.
loaded models after: 2023-04-24T13:05:12.496854
100%|██████████| 50/50 [00:36<00:00, 1.38it/s]
/usr/local/lib/python3.10/site-packages/diffusers/pipelines/pipeline_utils.py:1363: RuntimeWarning: invalid value encountered in cast
images = (images * 255).round().astype("uint8")
completed pipeline: 2023-04-24T13:05:52.171115
Thank you very much in advance!
$ ./build.sh run --model 'timbrooks/instruct-pix2pix' --image image.png image.png 'Change the dog to be a cat'
loaded image from image.png: 2023-10-01T12:22:36.120824
load pipeline start: 2023-10-01T12:22:36.342914
Loading pipeline components...: 86%|████████▌ | 6/7 [00:00<00:00, 6.20it/s]`text_config_dict` is provided which will be used to initialize `CLIPTextConfig`. The value `text_config["id2label"]` will be overriden.
`text_config_dict` is provided which will be used to initialize `CLIPTextConfig`. The value `text_config["bos_token_id"]` will be overriden.
`text_config_dict` is provided which will be used to initialize `CLIPTextConfig`. The value `text_config["eos_token_id"]` will be overriden.
Loading pipeline components...: 100%|██████████| 7/7 [00:01<00:00, 5.22it/s]
loaded models after: 2023-10-01T12:22:42.096667
Traceback (most recent call last):
File "/usr/local/bin/docker-entrypoint.py", line 297, in <module>
main()
File "/usr/local/bin/docker-entrypoint.py", line 293, in main
stable_diffusion_inference(pipeline)
File "/usr/local/bin/docker-entrypoint.py", line 143, in stable_diffusion_inference
result = p.pipeline(**remove_unused_args(p))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.11/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context
return func(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.11/site-packages/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_instruct_pix2pix.py", line 265, in __call__
do_classifier_free_guidance = guidance_scale > 1.0 and image_guidance_scale >= 1.0
^^^^^^^^^^^^^^^^^^^^^^^^^^^
TypeError: '>=' not supported between instances of 'NoneType' and 'float'
First I want to express my appreciation for your contribution to the community, Thank you.
Now for my question.
I forked the repository to add changes to meet my requirements, unfortunately I failed to run the commands provided by the 'build.sh', except './build.sh'.
The error was 'nvidia-container-cli: initialization error: load library failed: libnvidia-ml.so.1: cannot open shared object file: no such file or directory: unknown.'
So,my last try was to clone your original repo and follow the instructions exactly, and without any changes (not even a bit).
I got the same error.
What am I missing or doing wrong? Do I need to install any driver,sdk, or other action needed on the host running the container?
The host is MacBook pro 2019 intel base 16 GB ram 15’ update with the latest os, and using docker desktop latest version.
I will appreciate a lot any suggestion.
Thanks again,
Alon
In direct linux machine I get these error:
OSError: Unable to load weights from pytorch checkpoint file for '/home/huggingface/.cache/huggingface/diffusers/models--CompVis--stable-diffusion-v1-4/snapshots/48a1bd8a2d86483de1ac27e95e188569b57a494e/unet/diffusion_pytorch_model.bin' at '/home/huggingface/.cache/huggingface/diffusers/models--CompVis--stable-diffusion-v1-4/snapshots/48a1bd8a2d86483de1ac27e95e188569b57a494e/unet/diffusion_pytorch_model.bin'. If you tried to load a PyTorch model from a TF 2.0 checkpoint, please set from_tf=True.
Looks like this path is broken when building it /home/huggingface/.cache/huggingface/diffusers/models--CompVis--stable-diffusion-v1- but I can pind place where to fix it.
Hi,
is it possible to use the embeddings with the setup?
Hello, I have an other error.
./build.sh run 'a'
load pipeline start: 2022-10-03T17:08:45.114285
Fetching 19 files: 100%|██████████| 19/19 [00:00<00:00, 49528.76it/s]
ftfy or spacy is not installed using BERT BasicTokenizer instead of ftfy.
{'trained_betas'} was not found in config. Values will be initialized to default values.
loaded models after: 2022-10-03T17:08:59.515898
0%| | 0/51 [00:01<?, ?it/s]
Traceback (most recent call last):
File "/usr/local/bin/docker-entrypoint.py", line 174, in <module>
main()
File "/usr/local/bin/docker-entrypoint.py", line 157, in main
stable_diffusion(
File "/usr/local/bin/docker-entrypoint.py", line 54, in stable_diffusion
images = pipe(
File "/usr/local/lib/python3.8/dist-packages/torch/autograd/grad_mode.py", line 27, in decorate_context
return func(*args, **kwargs)
File "/usr/local/lib/python3.8/dist-packages/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py", line 249, in __call__
noise_pred = self.unet(latent_model_input, t, encoder_hidden_states=text_embeddings).sample
File "/usr/local/lib/python3.8/dist-packages/torch/nn/modules/module.py", line 1130, in _call_impl
return forward_call(*input, **kwargs)
File "/usr/local/lib/python3.8/dist-packages/diffusers/models/unet_2d_condition.py", line 234, in forward
sample, res_samples = downsample_block(
File "/usr/local/lib/python3.8/dist-packages/torch/nn/modules/module.py", line 1130, in _call_impl
return forward_call(*input, **kwargs)
File "/usr/local/lib/python3.8/dist-packages/diffusers/models/unet_blocks.py", line 537, in forward
hidden_states = attn(hidden_states, context=encoder_hidden_states)
File "/usr/local/lib/python3.8/dist-packages/torch/nn/modules/module.py", line 1130, in _call_impl
return forward_call(*input, **kwargs)
File "/usr/local/lib/python3.8/dist-packages/diffusers/models/attention.py", line 148, in forward
x = block(x, context=context)
File "/usr/local/lib/python3.8/dist-packages/torch/nn/modules/module.py", line 1130, in _call_impl
return forward_call(*input, **kwargs)
File "/usr/local/lib/python3.8/dist-packages/diffusers/models/attention.py", line 197, in forward
x = self.attn1(self.norm1(x)) + x
File "/usr/local/lib/python3.8/dist-packages/torch/nn/modules/module.py", line 1130, in _call_impl
return forward_call(*input, **kwargs)
File "/usr/local/lib/python3.8/dist-packages/diffusers/models/attention.py", line 265, in forward
hidden_states = self._attention(q, k, v, sequence_length, dim)
File "/usr/local/lib/python3.8/dist-packages/diffusers/models/attention.py", line 281, in _attention
attn_slice = attn_slice.softmax(dim=-1)
RuntimeError: CUDA out of memory. Tried to allocate 1024.00 MiB (GPU 0; 8.00 GiB total capacity; 5.88 GiB already allocated; 0 bytes free; 6.39 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
Hi, it's my understanding that this uses the model hosted by Huggingface and therefore needs an access token, is it possible to point to a local model (4~GB) and circumvent the access token requirement?
Please feel free to close this once answered, I know questions shouldn't go into the issues section, sorry!
I use a random seed so I can just run the same prompt several times until I get one I like, but once I do I would like the seed for it so I can generate it again in the future (or iterate on it with other settings). However, it doesn't appear the random seed is presented to the user anywhere.
Please describe how to run it on GPU with less than 8 GB VRAM
Hi!
I'm utilising a system with multiple GPUs. Running multiple instances of the docker container means I see errors as shown below. How can I target specific GPUs on startup?
loaded models after: 2023-07-06T08:39:39.643731 Traceback (most recent call last): File "/usr/local/bin/docker-entrypoint.py", line 307, in <module> main() File "/usr/local/bin/docker-entrypoint.py", line 303, in main stable_diffusion_inference(pipeline) File "/usr/local/bin/docker-entrypoint.py", line 153, in stable_diffusion_inference result = p.pipeline(**remove_unused_args(p)) File "/usr/local/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context return func(*args, **kwargs) File "/usr/local/lib/python3.10/site-packages/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py", line 688, in __call__ prompt_embeds = self._encode_prompt( File "/usr/local/lib/python3.10/site-packages/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py", line 382, in _encode_prompt prompt_embeds = self.text_encoder( File "/usr/local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl return forward_call(*args, **kwargs) File "/usr/local/lib/python3.10/site-packages/transformers/models/clip/modeling_clip.py", line 822, in forward return self.text_model( File "/usr/local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl return forward_call(*args, **kwargs) File "/usr/local/lib/python3.10/site-packages/transformers/models/clip/modeling_clip.py", line 740, in forward encoder_outputs = self.encoder( File "/usr/local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl return forward_call(*args, **kwargs) File "/usr/local/lib/python3.10/site-packages/transformers/models/clip/modeling_clip.py", line 654, in forward layer_outputs = encoder_layer( File "/usr/local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl return forward_call(*args, **kwargs) File "/usr/local/lib/python3.10/site-packages/transformers/models/clip/modeling_clip.py", line 383, in forward hidden_states, attn_weights = self.self_attn( File "/usr/local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl return forward_call(*args, **kwargs) File "/usr/local/lib/python3.10/site-packages/transformers/models/clip/modeling_clip.py", line 272, in forward query_states = self.q_proj(hidden_states) * self.scale File "/usr/local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl return forward_call(*args, **kwargs) File "/usr/local/lib/python3.10/site-packages/torch/nn/modules/linear.py", line 114, in forward return F.linear(input, self.weight, self.bias) RuntimeError: CUDA error: CUBLAS_STATUS_NOT_INITIALIZED when calling cublasCreate(handle)``
After a recent git pull running the same command to generate an image from an image no longer works:
./build.sh run --image img.png "<desc>"
I do see the log line loaded image from img.png: 2024-02-01T18:00:40.999099 yet nothing resembling the image ever comes out (like it did before) despite playing with strength and image-scale options
Hi there,
I'm trying to run the docker, and running into an issue. Not quite sure why numpy wouldn't have 'float'.
the docker build went just fine, but when trying the example, it failed with the following:
Any idea how to diagnose that?
Thanks!
stable-diffusion-docker$ ./build.sh run --device cpu 'Andromeda galaxy in a bottle'
load pipeline start: 2023-01-04T20:22:12.012404
Downloading: 100%|██████████| 604/604 [00:00<00:00, 1.16MB/s]
Downloading: 100%|██████████| 342/342 [00:00<00:00, 158kB/s]
Downloading: 100%|██████████| 1.22G/1.22G [00:51<00:00, 23.4MB/s]
Downloading: 100%|██████████| 289/289 [00:00<00:00, 155kB/s]s/it]
Downloading: 100%|██████████| 492M/492M [00:21<00:00, 22.6MB/s]t]
Downloading: 100%|██████████| 525k/525k [00:00<00:00, 1.42MB/s]t]
Downloading: 100%|██████████| 472/472 [00:00<00:00, 259kB/s]s/it]
Downloading: 100%|██████████| 917/917 [00:00<00:00, 501kB/s]s/it]
Downloading: 100%|██████████| 1.06M/1.06M [00:00<00:00, 2.26MB/s]
Downloading: 100%|██████████| 778k/778k [00:00<00:00, 2.45MB/s]t]
Fetching 13 files: 77%|███████▋ | 10/13 [01:21<00:10, 3.42s/itrla
Downloading: 100%|██████████| 3.44G/3.44G [02:25<00:00, 23.7MB/s]
Downloading: 100%|██████████| 198M/198M [00:08<00:00, 22.4MB/s]it]
Downloading: 100%|██████████| 137M/137M [00:06<00:00, 22.6MB/s]it]
Fetching 13 files: 100%|██████████| 13/13 [04:04<00:00, 18.84s/it]
loaded models after: 2023-01-04T20:26:26.394072
Traceback (most recent call last):
File "/usr/local/bin/docker-entrypoint.py", line 263, in <module>
main()
File "/usr/local/bin/docker-entrypoint.py", line 259, in main
stable_diffusion_inference(pipeline)
File "/usr/local/bin/docker-entrypoint.py", line 118, in stable_diffusion_inference
result = p.pipeline(**remove_unused_args(p))
File "/usr/local/lib/python3.10/site-packages/diffusers/pipelines/stable_diffusion/pipeline_onnx_stable_diffusion.py", line 250, in __call__
latents = latents * np.float(self.scheduler.init_noise_sigma)
File "/usr/local/lib/python3.10/site-packages/numpy/__init__.py", line 284, in __getattr__
raise AttributeError("module {!r} has no attribute "
AttributeError: module 'numpy' has no attribute 'float'. Did you mean: 'cfloat'?
setup is ok, but error when run
./build.sh run 'Andromeda galaxy in a bottle' docker: Error response from daemon: could not select device driver "" with capabilities: [[gpu]].
I'm using Ubuntu 22.04.02 LTS
Core i7-1195G7
RAM 16GB
Graphics Mesa Intel® Xe Graphics (TGL GT2)
Disk Capacity 512,1GB
Please help!
Thanks!
.\build.sh run 'magical castle' --W 256 --H 256 --half --seed 42
Potential NSFW content was detected in one or more images. A black image will be returned instead. Try again with a different prompt and/or seed.
I am using 256*256 resolution with half set since my GPU only has 4GB VRAM.
I have tried multiple prompts, including the one used in README.md of this project, but all prompts give the same issue. I have tried playing with the seed value too but no change.
my command is the following and it throws an error
sudo bash ./build.sh run --token hf_sagfdcnhfxvbgc --device cpu --onnx 'Andromeda galaxy in a bottle'
Question: is it possible to use a local safetensors file for the model instead of a hosted one?
Hello!
I was curious if there would be an addition to change what samplers you would be able to use at ./build.sh run. I am particularly interested in adding the following sampler: (DPM++ 2S a Karras). The reason why I ask is so that I could minimize the step count while maintaining a high-fidelity image as this would allow for a quicker rendering of these images.
Thank you for this docker file!
Hi there!
Thank you for creating this, it's really nice to have a self-contained, easily deployable and replicable image to work from.
I'm from a network engineering, not software dev, so please excuse me if this is a silly question. :)
I was just wondering if you are using txt2img.py in this, and if so, is there any way to pass txt2img.py script options when running build.sh?
I've tried specifying things like height (--H), width (--W), samples (--n_samples) etc, both inside and outside the single quotes, but it does not seem to have the usual effect.
Kind regards,
Dominic
i run it on a CPU paired with 32Gb of Ram, the problem is that it uses only 50% of the cpu, i have 6 cores/12threads but it uses only 6threads, can i increase that somewhere to use all cores?
I am having this error on build:
./build.sh run 'A high tech solarpunk utopia in the Amazon rainforest'
WARNING: The requested image's platform (linux/amd64) does not match the detected host platform (linux/arm64/v8) and no specific platform was requested
docker: Error response from daemon: could not select device driver "" with capabilities: [[gpu]].
(base)
Not sure if other people will run into this, but I added
RUN rm -rf /usr/local/cuda/lib64/stubs
to the Dockerfile (before USER) to have the CUDA drivers loaded in the container.
Hello!
I am having trouble with line 23 in the file, "Dockerfile" where the following command is this:
Line 23: COPY token.txt /home/huggingface
When I try to:
./build.sh build
I get the following error on Step 11/12" COPY token.txt /home/huggingface":
COPY failed: file not found in build context or excluded by .dockerignore: stat token.txt: file does not exist
I have tried the following:
I feel like I made a mistake somewhere that I cannot figure out how to fix. #
this is my terminal, after install all requirements and testing first command, i get no response, no image displayed. thanks.
"""
(stable) ubuntu@ubuntuu:~/Downloads/stable-diffusion-docker-main$ ./build.sh run 'Andromeda galaxy in a bottle' --device cpu --onnx
load pipeline start: 2023-08-07T00:11:30.714407
loaded models after: 2023-08-07T00:11:41.887265
100%|██████████| 51/51 [03:23<00:00, 3.99s/it]
completed pipeline: 2023-08-07T00:15:12.201066
"""
When I try to launch the container. I get this error.
/usr/bin/env: ‘python\r’: No such file or directoryI'm using a g5.xlarge in AWS (and trying with RTX 4090 locally). text-to-image works while image-to-image fails with:
torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 5.78 GiB (GPU 0; 22.06 GiB total capacity; 19.90 GiB already allocated; 904.38 MiB free; 19.92 GiB reserved in total by PyTorch)
any help would be appreciated! 😄
Pretty much the title. Can anyone help me?
I always get the following output, even using --device cpu.
./build.sh run 'An impressionist painting of a parakeet eating spaghetti in the desert' --device cpu
docker: Error response from daemon: failed to create shim task: OCI runtime create failed: runc create failed: unable to start container process: error during container init: error running hook #0: error running hook: exit status 1, stdout: , stderr: Auto-detected mode as 'legacy'
nvidia-container-cli: initialization error: load library failed: libnvidia-ml.so.1: cannot open shared object file: no such file or directory: unknown.
Just reinstalled Debian and set stable diffusion up, and I've never gotten this error before. Any thoughts that could help? Doing everything properly, running build.sh in sudo
Hello!
I was playing around with trying to switch the checkpoint file, and the method I was using was the following:
./build.sh run --model "andite/anything-v4.0" "andromeda galaxy in a bottle".
However, there was no success in doing so.
Is there a command with which I can implement the following checkpoint file: "https://huggingface.co/andite/anything-v4.0/blob/main/anything-v4.0-pruned-fp16.ckpt"?
I took a look at the following resources prior to this:
Issue 30: #30
Issue 33: #33
Scripts: https://github.com/huggingface/diffusers/blob/main/scripts/convert_original_stable_diffusion_to_diffusers.py
Hi,
I've just been trying out different tools, grabbed both of the 1.4 checkpoint files - and with my seed, steps, scale, image size.... was not able to reproduce images I've been creating in this docker container!
I went to look in the huggingface docker volume mounted, but can only see a folder "blobs" and no ckpt to pull out to test with?
Thanks!
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
Django
The Web framework for perfectionists with deadlines.
Laravel
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
Microsoft
Open source projects and samples from Microsoft.
Google
Google ❤️ Open Source for everyone.
Alibaba
Alibaba Open Source for everyone
D3
Data-Driven Documents codes.
Tencent
China tencent open source team.