[Admin maintenance] Migrate to ZeroGPU

README.md

@@ -4,7 +4,8 @@ emoji: 📚
 colorFrom: red
 colorTo: red
 sdk: gradio
-sdk_version: 4.10.0
+sdk_version: 5.49.1
+python_version: '3.10'
 app_file: app.py
 pinned: false
 license: mit

app.py

@@ -1,6 +1,15 @@
 import os
 import sys
 
+import spaces
+
+# Re-export init helpers that mmcv 2.x moved to mmengine.model, so the remote
+# trust_remote_code modules (MykolaL/evp_depth, MykolaL/evp_refer) keep working.
+import mmcv.cnn as _mmcv_cnn
+from mmengine.model import constant_init as _constant_init, normal_init as _normal_init
+_mmcv_cnn.constant_init = _constant_init
+_mmcv_cnn.normal_init = _normal_init
+
 sys.path.append(os.path.abspath(os.path.join(os.path.dirname(__file__), 'depth')))
 sys.path.append(os.path.abspath(os.path.join(os.path.dirname(__file__), 'refer')))
 sys.path.append(os.path.abspath(os.path.join(os.path.dirname(__file__), 'stable-diffusion')))
@@ -49,7 +58,8 @@ def create_depth_demo(model, device):
         depth_image = gr.Image(label="Depth Map", elem_id='img-display-output')
     raw_file = gr.File(label="16-bit raw depth, multiplier:256")
     submit = gr.Button("Submit")
-    
+
+    @spaces.GPU
     def on_submit(image):
         transform = transforms.ToTensor()
         image = transform(image).unsqueeze(0).to(device)
@@ -85,7 +95,8 @@ def create_refseg_demo(model, tokenizer, device):
         refseg_image = gr.Image(label="Output Mask", elem_id='img-display-output')
     input_text = gr.Textbox(label='Prompt', placeholder='Please upload your image first', lines=2)
     submit = gr.Button("Submit")
-    
+
+    @spaces.GPU
     def on_submit(image, text):
         # Convert PIL -> np array
         transform = transforms.ToTensor()
@@ -163,7 +174,7 @@ def main():
         gr.HTML('''<br><br><br><center>You can duplicate this Space to skip the queue:<a href="https://huggingface.co/spaces/MykolaL/evp?duplicate=true"><img src="https://bit.ly/3gLdBN6" alt="Duplicate Space"></a><br>
                 <p><img src="https://visitor-badge.glitch.me/badge?page_id=MykolaL/evp" alt="visitors"></p></center>''')
 
-    demo.queue().launch(share=True)
+    demo.queue().launch()
 
 
 if __name__ == '__main__':

depth/models_depth/model.py

@@ -9,8 +9,8 @@
 import torch
 import torch.nn as nn
 from timm.models.layers import trunc_normal_, DropPath
-from mmcv.cnn import (build_conv_layer, build_norm_layer, build_upsample_layer,
-                      constant_init, normal_init)
+from mmcv.cnn import build_conv_layer, build_norm_layer, build_upsample_layer
+from mmengine.model import constant_init, normal_init
 from omegaconf import OmegaConf
 from ldm.util import instantiate_from_config
 import torch.nn.functional as F

depth/models_depth/model_vpd.py

@@ -9,8 +9,8 @@
 import torch
 import torch.nn as nn
 from timm.models.layers import trunc_normal_, DropPath
-from mmcv.cnn import (build_conv_layer, build_norm_layer, build_upsample_layer,
-                      constant_init, normal_init)
+from mmcv.cnn import build_conv_layer, build_norm_layer, build_upsample_layer
+from mmengine.model import constant_init, normal_init
 from omegaconf import OmegaConf
 from ldm.util import instantiate_from_config
 import torch.nn.functional as F

depth/utils.py

@@ -18,7 +18,7 @@ from pathlib import Path
 
 import torch
 import torch.distributed as dist
-from torch._six import inf
+from math import inf
 
 from tensorboardX import SummaryWriter
 

requirements.txt

@@ -1,30 +1,20 @@
---extra-index-url https://download.pytorch.org/whl/cu113
-torch==1.13.1
-torchvision==0.14.1
-salesforce-lavis
-bitsandbytes==0.37.0
-transformers==4.35.2 
-diffusers 
-invisible-watermark
+torch
+torchvision
+transformers==4.35.2
+diffusers
 accelerate
-albumentations==0.4.3
-opencv-python==4.8.0.76
-pudb==2019.2
-imageio==2.9.0
-imageio-ffmpeg==0.4.2
-pytorch-lightning==1.4.2
-omegaconf==2.1.1
-test-tube>=0.7.5
-streamlit>=0.73.1
+albumentations
+opencv-python
+imageio
+imageio-ffmpeg
+pytorch-lightning
+omegaconf
 einops==0.7.0
-torch-fidelity==0.3.0
-torchmetrics==0.6.0
-kornia==0.6
-streamlit-drawable-canvas==0.8
-mmcv-full==1.7.1
-tensorboardX==2.6.2.2
-pydantic==2.8.2
-pydantic-core==2.20.1
-fastapi==0.112.4
+torchmetrics
+kornia
+mmcv-lite
+mmengine
+tensorboardX
+timm
 -e git+https://github.com/openai/CLIP.git@main#egg=clip
--e git+https://github.com/CompVis/taming-transformers.git#egg=taming-transformers
+-e git+https://github.com/CompVis/taming-transformers.git#egg=taming-transformers

stable-diffusion/ldm/models/diffusion/ddpm.py

@@ -16,7 +16,7 @@ from contextlib import contextmanager
 from functools import partial
 from tqdm import tqdm
 from torchvision.utils import make_grid
-from pytorch_lightning.utilities.distributed import rank_zero_only
+from pytorch_lightning.utilities import rank_zero_only
 
 from ldm.util import log_txt_as_img, exists, default, ismap, isimage, mean_flat, count_params, instantiate_from_config
 from ldm.modules.ema import LitEma