Spaces:
Runtime error
Runtime error
Commit
Β·
d410924
1
Parent(s):
5d86663
Add google colab
Browse files
app.py
CHANGED
|
@@ -3,18 +3,27 @@ import gradio as gr
|
|
| 3 |
import torch
|
| 4 |
from torch import autocast
|
| 5 |
from diffusers import DiffusionPipeline
|
| 6 |
-
import streamlit as st
|
| 7 |
from transformers import (
|
| 8 |
pipeline,
|
| 9 |
MBart50TokenizerFast,
|
| 10 |
MBartForConditionalGeneration,
|
| 11 |
)
|
| 12 |
|
|
|
|
|
|
|
| 13 |
device = "cuda" if torch.cuda.is_available() else "cpu"
|
| 14 |
device_dict = {"cuda": 0, "cpu": -1}
|
| 15 |
context = autocast if device == "cuda" else nullcontext
|
| 16 |
dtype = torch.float16 if device == "cuda" else torch.float32
|
| 17 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 18 |
# Add language detection pipeline
|
| 19 |
language_detection_model_ckpt = "papluca/xlm-roberta-base-language-detection"
|
| 20 |
language_detection_pipeline = pipeline("text-classification",
|
|
@@ -27,16 +36,28 @@ trans_model = MBartForConditionalGeneration.from_pretrained("facebook/mbart-larg
|
|
| 27 |
|
| 28 |
model_id = "CompVis/stable-diffusion-v1-4"
|
| 29 |
|
| 30 |
-
|
| 31 |
-
|
| 32 |
-
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
|
| 36 |
-
|
| 37 |
-
|
| 38 |
-
|
| 39 |
-
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 40 |
|
| 41 |
pipe = pipe.to(device)
|
| 42 |
|
|
@@ -228,6 +249,10 @@ with block as demo:
|
|
| 228 |
<p style="margin-bottom: 10px; font-size: 94%">
|
| 229 |
Stable Diffusion Pipeline that supports prompts in 50 different languages.
|
| 230 |
</p>
|
|
|
|
|
|
|
|
|
|
|
|
|
| 231 |
</div>
|
| 232 |
"""
|
| 233 |
)
|
|
@@ -277,4 +302,8 @@ Despite how impressive being able to turn text into image is, beware to the fact
|
|
| 277 |
</div>
|
| 278 |
"""
|
| 279 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
| 280 |
demo.queue(max_size=25).launch()
|
|
|
|
| 3 |
import torch
|
| 4 |
from torch import autocast
|
| 5 |
from diffusers import DiffusionPipeline
|
|
|
|
| 6 |
from transformers import (
|
| 7 |
pipeline,
|
| 8 |
MBart50TokenizerFast,
|
| 9 |
MBartForConditionalGeneration,
|
| 10 |
)
|
| 11 |
|
| 12 |
+
import utils
|
| 13 |
+
|
| 14 |
device = "cuda" if torch.cuda.is_available() else "cpu"
|
| 15 |
device_dict = {"cuda": 0, "cpu": -1}
|
| 16 |
context = autocast if device == "cuda" else nullcontext
|
| 17 |
dtype = torch.float16 if device == "cuda" else torch.float32
|
| 18 |
|
| 19 |
+
# Detect if code is running in Colab
|
| 20 |
+
is_colab = utils.is_google_colab()
|
| 21 |
+
colab_instruction = "" if is_colab else """
|
| 22 |
+
<p>You can skip the queue using Colab:
|
| 23 |
+
<a href="https://colab.research.google.com/drive/1nhXyddThldnxPfIYO2my_bYinlMUW30R?usp=sharing">
|
| 24 |
+
<img data-canonical-src="https://colab.research.google.com/assets/colab-badge.svg" alt="Open In Colab" src="https://colab.research.google.com/assets/colab-badge.svg"></a></p>"""
|
| 25 |
+
device_print = "GPU π₯" if torch.cuda.is_available() else "CPU π₯Ά"
|
| 26 |
+
|
| 27 |
# Add language detection pipeline
|
| 28 |
language_detection_model_ckpt = "papluca/xlm-roberta-base-language-detection"
|
| 29 |
language_detection_pipeline = pipeline("text-classification",
|
|
|
|
| 36 |
|
| 37 |
model_id = "CompVis/stable-diffusion-v1-4"
|
| 38 |
|
| 39 |
+
if is_colab:
|
| 40 |
+
pipe = DiffusionPipeline.from_pretrained(
|
| 41 |
+
model_id,
|
| 42 |
+
custom_pipeline="multilingual_stable_diffusion",
|
| 43 |
+
detection_pipeline=language_detection_pipeline,
|
| 44 |
+
translation_model=trans_model,
|
| 45 |
+
translation_tokenizer=trans_tokenizer,
|
| 46 |
+
revision="fp16",
|
| 47 |
+
torch_dtype=dtype,
|
| 48 |
+
)
|
| 49 |
+
else:
|
| 50 |
+
import streamlit as st
|
| 51 |
+
pipe = DiffusionPipeline.from_pretrained(
|
| 52 |
+
model_id,
|
| 53 |
+
custom_pipeline="multilingual_stable_diffusion",
|
| 54 |
+
use_auth_token=st.secrets["USER_TOKEN"],
|
| 55 |
+
detection_pipeline=language_detection_pipeline,
|
| 56 |
+
translation_model=trans_model,
|
| 57 |
+
translation_tokenizer=trans_tokenizer,
|
| 58 |
+
revision="fp16",
|
| 59 |
+
torch_dtype=dtype,
|
| 60 |
+
)
|
| 61 |
|
| 62 |
pipe = pipe.to(device)
|
| 63 |
|
|
|
|
| 249 |
<p style="margin-bottom: 10px; font-size: 94%">
|
| 250 |
Stable Diffusion Pipeline that supports prompts in 50 different languages.
|
| 251 |
</p>
|
| 252 |
+
<p>
|
| 253 |
+
{colab_instruction}
|
| 254 |
+
Running on <b>{device_print}</b>{(" in a <b>Google Colab</b>." if is_colab else "")}
|
| 255 |
+
</p>
|
| 256 |
</div>
|
| 257 |
"""
|
| 258 |
)
|
|
|
|
| 302 |
</div>
|
| 303 |
"""
|
| 304 |
)
|
| 305 |
+
gr.Markdown('''
|
| 306 |
+
[](https://twitter.com/juancopi81)
|
| 307 |
+

|
| 308 |
+
''')
|
| 309 |
demo.queue(max_size=25).launch()
|
utils.py
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
def is_google_colab():
|
| 2 |
+
try:
|
| 3 |
+
import google.colab
|
| 4 |
+
return True
|
| 5 |
+
except:
|
| 6 |
+
return False
|