Ali2206 commited on
Commit
6b182f5
·
verified ·
1 Parent(s): 6943e92

Update Dockerfile

Browse files
Files changed (1) hide show
  1. Dockerfile +14 -2
Dockerfile CHANGED
@@ -20,6 +20,18 @@ RUN apt-get update && apt-get install -y \
20
  # Create and set working directory
21
  WORKDIR /app
22
 
 
 
 
 
 
 
 
 
 
 
 
 
23
  # Create necessary directories
24
  RUN mkdir -p /data/hf_cache/txagent_models \
25
  /data/hf_cache/tool_cache \
@@ -27,7 +39,7 @@ RUN mkdir -p /data/hf_cache/txagent_models \
27
  /data/hf_cache/reports \
28
  /tmp/matplotlib
29
 
30
- # Install Python dependencies
31
  COPY requirements.txt .
32
  RUN pip install --no-cache-dir -r requirements.txt
33
 
@@ -37,5 +49,5 @@ COPY . .
37
  # Expose port
38
  EXPOSE 7860
39
 
40
- # Command to run the application
41
  CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]
 
20
  # Create and set working directory
21
  WORKDIR /app
22
 
23
+ # Install specific versions compatible with A100 and CUDA 12.1
24
+ RUN pip install --no-cache-dir \
25
+ torch==2.1.2+cu121 \
26
+ --extra-index-url https://download.pytorch.org/whl/cu121
27
+
28
+ # Install optimized vLLM for A100
29
+ RUN pip install --no-cache-dir \
30
+ vllm==0.3.0 \
31
+ xformers==0.0.22.post7 \
32
+ sentence-transformers==2.2.2 \
33
+ flash-attn==2.3.3
34
+
35
  # Create necessary directories
36
  RUN mkdir -p /data/hf_cache/txagent_models \
37
  /data/hf_cache/tool_cache \
 
39
  /data/hf_cache/reports \
40
  /tmp/matplotlib
41
 
42
+ # Copy requirements first to leverage Docker cache
43
  COPY requirements.txt .
44
  RUN pip install --no-cache-dir -r requirements.txt
45
 
 
49
  # Expose port
50
  EXPOSE 7860
51
 
52
+ # Command to run the application with optimal settings for A100
53
  CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]