Spaces:
Sleeping
Sleeping
Create Dockerfile
Browse files- Dockerfile +18 -0
Dockerfile
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# 1. Use the highly optimized C++ llama.cpp server
|
| 2 |
+
FROM ghcr.io/ggerganov/llama.cpp:server
|
| 3 |
+
|
| 4 |
+
# Switch to root temporarily to install missing packages
|
| 5 |
+
USER root
|
| 6 |
+
RUN apt-get update && apt-get install -y wget
|
| 7 |
+
|
| 8 |
+
# 2. Expose the port Hugging Face expects for web traffic
|
| 9 |
+
EXPOSE 7860
|
| 10 |
+
|
| 11 |
+
# 3. Download the 4-bit quantized Gemma 2 (2B) model during the build phase
|
| 12 |
+
RUN wget -O gemma-2b.gguf https://huggingface.co/bartowski/gemma-2-2b-it-GGUF/resolve/main/gemma-2-2b-it-Q4_K_M.gguf
|
| 13 |
+
|
| 14 |
+
# Switch back to a non-root user (Hugging Face Spaces requires running as User 1000)
|
| 15 |
+
USER 1000
|
| 16 |
+
|
| 17 |
+
# 4. Boot the server
|
| 18 |
+
CMD ["-m", "gemma-2b.gguf", "--host", "0.0.0.0", "--port", "7860", "-t", "2", "-c", "2048"]
|