Spaces:
Running
on
Zero
Running
on
Zero
Update app.py
Browse files
app.py
CHANGED
@@ -1,71 +1,123 @@
|
|
1 |
-
|
2 |
-
from
|
3 |
import subprocess
|
|
|
|
|
|
|
|
|
4 |
|
5 |
-
|
6 |
-
|
7 |
-
|
8 |
-
|
9 |
-
|
10 |
-
|
11 |
-
|
12 |
-
|
13 |
-
|
14 |
-
|
15 |
-
|
16 |
-
|
17 |
-
|
18 |
-
|
19 |
-
|
20 |
-
|
21 |
-
|
22 |
-
|
23 |
-
|
24 |
-
|
25 |
-
|
26 |
-
|
27 |
-
|
28 |
-
|
29 |
-
|
30 |
-
|
31 |
-
|
32 |
-
|
33 |
-
|
34 |
-
|
35 |
-
|
36 |
-
|
37 |
-
|
38 |
-
|
39 |
-
|
40 |
-
|
41 |
-
|
42 |
-
|
43 |
-
|
44 |
-
|
45 |
-
|
46 |
-
|
47 |
-
|
48 |
-
|
49 |
-
|
50 |
-
|
51 |
-
|
52 |
-
|
53 |
-
|
54 |
-
|
55 |
-
|
56 |
-
|
57 |
-
|
58 |
-
|
59 |
-
|
60 |
-
|
61 |
-
|
62 |
-
|
63 |
-
|
64 |
-
|
65 |
-
|
66 |
-
|
67 |
-
|
68 |
-
|
69 |
-
|
70 |
-
|
71 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
from huggingface_hub import hf_hub_download
|
3 |
import subprocess
|
4 |
+
import tempfile
|
5 |
+
import shutil
|
6 |
+
import os
|
7 |
+
import spaces
|
8 |
|
9 |
+
from transformers import T5ForConditionalGeneration, T5Tokenizer
|
10 |
+
import os
|
11 |
+
|
12 |
+
subprocess.run('pip install flash-attn --no-build-isolation', env={'FLASH_ATTENTION_SKIP_CUDA_BUILD': "TRUE"}, shell=True)
|
13 |
+
|
14 |
+
|
15 |
+
def install_apex():
|
16 |
+
# Install Apex in editable mode from the specified GitHub repository
|
17 |
+
cmd = [
|
18 |
+
'pip', 'install', '--no-cache-dir', '--no-build-isolation',
|
19 |
+
'--config-settings', 'build-option=--cpp_ext', '--config-settings',
|
20 |
+
'build-option=--cuda_ext', '-e', 'git+https://github.com/NVIDIA/apex.git'
|
21 |
+
]
|
22 |
+
subprocess.run(cmd, check=True)
|
23 |
+
|
24 |
+
try:
|
25 |
+
import apex
|
26 |
+
except ModuleNotFoundError:
|
27 |
+
print("Apex not found, installing...")
|
28 |
+
install_apex()
|
29 |
+
# Try to import Apex again after installation
|
30 |
+
import apex
|
31 |
+
|
32 |
+
|
33 |
+
def download_t5_model(model_id, save_directory):
|
34 |
+
# Modelin tokenizer'ını ve modeli indir
|
35 |
+
model = T5ForConditionalGeneration.from_pretrained(model_id)
|
36 |
+
tokenizer = T5Tokenizer.from_pretrained(model_id)
|
37 |
+
|
38 |
+
# Model ve tokenizer'ı belirtilen dizine kaydet
|
39 |
+
if not os.path.exists(save_directory):
|
40 |
+
os.makedirs(save_directory)
|
41 |
+
model.save_pretrained(save_directory)
|
42 |
+
tokenizer.save_pretrained(save_directory)
|
43 |
+
|
44 |
+
# Model ID ve kaydedilecek dizin
|
45 |
+
model_id = "DeepFloyd/t5-v1_1-xxl"
|
46 |
+
save_directory = "pretrained_models/t5_ckpts/t5-v1_1-xxl"
|
47 |
+
|
48 |
+
# Modeli indir
|
49 |
+
download_t5_model(model_id, save_directory)
|
50 |
+
|
51 |
+
def download_model(repo_id, model_name):
|
52 |
+
model_path = hf_hub_download(repo_id=repo_id, filename=model_name)
|
53 |
+
return model_path
|
54 |
+
|
55 |
+
import glob
|
56 |
+
|
57 |
+
@spaces.GPU
|
58 |
+
def run_inference(model_name, prompt_text):
|
59 |
+
repo_id = "hpcai-tech/Open-Sora"
|
60 |
+
|
61 |
+
# Map model names to their respective configuration files
|
62 |
+
config_mapping = {
|
63 |
+
"OpenSora-v1-16x256x256.pth": "configs/opensora/inference/16x256x256.py",
|
64 |
+
"OpenSora-v1-HQ-16x256x256.pth": "configs/opensora/inference/16x512x512.py",
|
65 |
+
"OpenSora-v1-HQ-16x512x512.pth": "configs/opensora/inference/64x512x512.py"
|
66 |
+
}
|
67 |
+
|
68 |
+
config_path = config_mapping[model_name]
|
69 |
+
ckpt_path = download_model(repo_id, model_name)
|
70 |
+
|
71 |
+
# Save prompt_text to a temporary text file
|
72 |
+
prompt_file = tempfile.NamedTemporaryFile(delete=False, suffix=".txt", mode='w')
|
73 |
+
prompt_file.write(prompt_text)
|
74 |
+
prompt_file.close()
|
75 |
+
|
76 |
+
with open(config_path, 'r') as file:
|
77 |
+
config_content = file.read()
|
78 |
+
config_content = config_content.replace('prompt_path = "./assets/texts/t2v_samples.txt"', f'prompt_path = "{prompt_file.name}"')
|
79 |
+
|
80 |
+
with tempfile.NamedTemporaryFile('w', delete=False, suffix='.py') as temp_file:
|
81 |
+
temp_file.write(config_content)
|
82 |
+
temp_config_path = temp_file.name
|
83 |
+
|
84 |
+
cmd = [
|
85 |
+
"torchrun", "--standalone", "--nproc_per_node", "1",
|
86 |
+
"scripts/inference.py", temp_config_path,
|
87 |
+
"--ckpt-path", ckpt_path
|
88 |
+
]
|
89 |
+
subprocess.run(cmd)
|
90 |
+
|
91 |
+
save_dir = "./outputs/samples/" # Örneğin, inference.py tarafından kullanılan kayıt dizini
|
92 |
+
list_of_files = glob.glob(f'{save_dir}/*')
|
93 |
+
if list_of_files:
|
94 |
+
latest_file = max(list_of_files, key=os.path.getctime)
|
95 |
+
return latest_file
|
96 |
+
else:
|
97 |
+
print("No files found in the output directory.")
|
98 |
+
return None
|
99 |
+
|
100 |
+
# Clean up the temporary files
|
101 |
+
os.remove(temp_file.name)
|
102 |
+
os.remove(prompt_file.name)
|
103 |
+
|
104 |
+
def main():
|
105 |
+
gr.Interface(
|
106 |
+
fn=run_inference,
|
107 |
+
inputs=[
|
108 |
+
gr.Dropdown(choices=[
|
109 |
+
"OpenSora-v1-16x256x256.pth",
|
110 |
+
"OpenSora-v1-HQ-16x256x256.pth",
|
111 |
+
"OpenSora-v1-HQ-16x512x512.pth"
|
112 |
+
],
|
113 |
+
value="OpenSora-v1-16x256x256.pth",
|
114 |
+
label="Model Selection"),
|
115 |
+
gr.Textbox(label="Prompt Text", value="Enter prompt text here")
|
116 |
+
],
|
117 |
+
outputs=gr.Video(label="Output Video"),
|
118 |
+
title="Open-Sora Inference",
|
119 |
+
description="Run Open-Sora Inference with Custom Parameters",
|
120 |
+
).launch()
|
121 |
+
|
122 |
+
if __name__ == "__main__":
|
123 |
+
main()
|