mknolan commited on
Commit
1959bc7
·
verified ·
1 Parent(s): 4b8ace7

Fix Dockerfile CMD to point to app.py

Browse files

Updated the Dockerfile to correctly use app.py instead of app_internvl2.py

Files changed (1) hide show
  1. Dockerfile +78 -21
Dockerfile CHANGED
@@ -1,34 +1,91 @@
1
- FROM nvidia/cuda:12.1.0-runtime-ubuntu22.04
2
 
3
- # Set working directory
4
- WORKDIR /app
 
 
 
 
 
 
 
 
 
 
 
 
 
5
 
6
  # Install system dependencies
7
- RUN apt-get update && apt-get install -y \
8
- python3 \
9
- python3-pip \
10
  git \
11
- ffmpeg \
12
- libsm6 \
13
- libxext6 \
 
 
14
  && rm -rf /var/lib/apt/lists/*
15
-
16
- # Upgrade pip
17
- RUN python3 -m pip install --upgrade pip
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
18
 
19
  # Copy requirements file
20
  COPY requirements.txt .
21
 
22
- # Install Python dependencies
23
- RUN pip install --no-cache-dir -r requirements.txt
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
24
 
25
- # Copy the rest of the application
26
  COPY . .
27
 
28
- # Set environment variables
29
- ENV PYTHONPATH=/app
30
- ENV PYTHONUNBUFFERED=1
31
- ENV PORT=7860
 
 
 
 
 
 
 
32
 
33
- # Set default command
34
- CMD ["python3", "app.py"]
 
1
+ FROM nvidia/cuda:11.8.0-cudnn8-runtime-ubuntu22.04
2
 
3
+ # Set environment variables
4
+ ENV DEBIAN_FRONTEND=noninteractive
5
+ ENV PYTHONUNBUFFERED=1
6
+ ENV HF_HOME=/app/.cache/huggingface
7
+ ENV TRANSFORMERS_CACHE=/app/.cache/huggingface/transformers
8
+ ENV MPLCONFIGDIR=/tmp/matplotlib
9
+ # Force PyTorch to use the NCCl backend
10
+ ENV PYTORCH_CUDA_ALLOC_CONF=max_split_size_mb:128
11
+
12
+ # Create necessary directories with proper permissions
13
+ RUN mkdir -p /app/.cache/huggingface/transformers && \
14
+ mkdir -p /tmp/matplotlib && \
15
+ mkdir -p /app/gradio_cached_examples && \
16
+ chmod -R 777 /app && \
17
+ chmod -R 777 /tmp/matplotlib
18
 
19
  # Install system dependencies
20
+ RUN apt-get update && apt-get install -y --no-install-recommends \
21
+ build-essential \
 
22
  git \
23
+ curl \
24
+ ca-certificates \
25
+ python3-pip \
26
+ python3-dev \
27
+ python3-setuptools \
28
  && rm -rf /var/lib/apt/lists/*
29
+
30
+ # Create a working directory
31
+ WORKDIR /app
32
+
33
+ # Add a script to check GPU status at startup
34
+ RUN echo '#!/bin/bash \n\
35
+ echo "Checking NVIDIA GPU status..." \n\
36
+ if ! command -v nvidia-smi &> /dev/null; then \n\
37
+ echo "WARNING: nvidia-smi command not found. NVIDIA driver might not be installed." \n\
38
+ else \n\
39
+ echo "NVIDIA driver found. Running nvidia-smi:" \n\
40
+ nvidia-smi \n\
41
+ fi \n\
42
+ echo "Environment variables for GPU:" \n\
43
+ echo "CUDA_VISIBLE_DEVICES=${CUDA_VISIBLE_DEVICES}" \n\
44
+ echo "NVIDIA_VISIBLE_DEVICES=${NVIDIA_VISIBLE_DEVICES}" \n\
45
+ exec "$@"' > /entrypoint.sh && \
46
+ chmod +x /entrypoint.sh
47
 
48
  # Copy requirements file
49
  COPY requirements.txt .
50
 
51
+ # Upgrade pip and install dependencies in specific order to avoid conflicts
52
+ RUN pip3 install --no-cache-dir --upgrade pip && \
53
+ # Install torch and torchvision first with CUDA support
54
+ pip3 install --no-cache-dir torch==2.0.1+cu118 torchvision==0.15.2+cu118 --extra-index-url https://download.pytorch.org/whl/cu118 && \
55
+ # Install core dependencies
56
+ pip3 install --no-cache-dir numpy==1.24.3 scipy==1.11.3 requests==2.31.0 && \
57
+ # Install typing-extensions first to ensure proper version for other packages
58
+ pip3 install --no-cache-dir typing-extensions==4.10.0 && \
59
+ # Install huggingface dependencies
60
+ pip3 install --no-cache-dir transformers==4.37.2 safetensors==0.4.1 huggingface_hub==0.19.4 && \
61
+ # Install timm for vision models
62
+ pip3 install --no-cache-dir timm==0.9.11 && \
63
+ # Install nest-asyncio for handling nested event loops
64
+ pip3 install --no-cache-dir nest-asyncio==1.5.8 && \
65
+ # Install lmdeploy and its dependencies first
66
+ pip3 install --no-cache-dir "accelerate==0.30.0" && \
67
+ pip3 install --no-cache-dir "lmdeploy==0.5.3" && \
68
+ # Install other acceleration libraries
69
+ pip3 install --no-cache-dir bitsandbytes==0.41.3 && \
70
+ # Install gradio
71
+ pip3 install --no-cache-dir gradio==3.38.0 && \
72
+ # Install any remaining requirements
73
+ pip3 install --no-cache-dir packaging==23.2 pyyaml==6.0.1 tqdm==4.66.1 openai==1.6.1
74
 
75
+ # Copy the application files
76
  COPY . .
77
 
78
+ # Make sure the runtime directories exist and have proper permissions
79
+ RUN mkdir -p gradio_cached_examples && \
80
+ chmod -R 777 gradio_cached_examples && \
81
+ mkdir -p .cache/huggingface/transformers && \
82
+ chmod -R 777 .cache
83
+
84
+ # Make port 7860 available for the app
85
+ EXPOSE 7860
86
+
87
+ # Use our entrypoint script to check GPU status before starting the app
88
+ ENTRYPOINT ["/entrypoint.sh"]
89
 
90
+ # Start the application - FIXED to point to the correct file
91
+ CMD ["python3", "app.py"]