Deploy 99fe36d9744820d798e0bf26993fb1f25c454346

.env

@@ -0,0 +1 @@
+LLAMA_API_KEY=sk-llama-b1415053335fd1d9ce87c3ff0a71f25c8ff8b46b9857cb62b67c278959bbbb75

.env.example

@@ -0,0 +1,3 @@
+# llama-server API key (env LLAMA_API_KEY). Copy to .env for local builds.
+# CI writes the real .env from the GitHub secret of the same name.
+LLAMA_API_KEY=

Dockerfile

@@ -0,0 +1,29 @@
+# OpenAI Chat — MiniCPM5-1B (Q8_0) on llama-server, port 7860
+FROM ghcr.io/ggml-org/llama.cpp:server
+
+# pre-build: bake the official Q8_0 GGUF into the image (1.1 GB)
+RUN mkdir -p /models && \
+    curl -fL --retry 3 -o /models/MiniCPM5-1B-Q8_0.gguf \
+      "https://huggingface.co/openbmb/MiniCPM5-1B-GGUF/resolve/main/MiniCPM5-1B-Q8_0.gguf" && \
+    ls -lh /models
+
+# key lives in .env (llama-server reads the LLAMA_API_KEY env var natively)
+COPY .env /app/.env
+
+EXPOSE 7860
+
+HEALTHCHECK --interval=30s --timeout=5s --start-period=120s --retries=6 \
+  CMD curl -fsS http://127.0.0.1:7860/health || exit 1
+
+# source .env unless the key is already set (Space secret / --env-file win)
+ENTRYPOINT ["/bin/sh", "-c", "if [ -z \"$LLAMA_API_KEY\" ]; then set -a; . /app/.env; set +a; fi; exec /app/llama-server \"$@\"", "--"]
+
+CMD ["-m", "/models/MiniCPM5-1B-Q8_0.gguf", \
+     "--host", "0.0.0.0", \
+     "--port", "7860", \
+     "-c", "16384", \
+     "-np", "4", \
+     "-t", "2", \
+     "-tb", "4", \
+     "--jinja", \
+     "--alias", "MiniCPM5-1B"]

README.md

@@ -0,0 +1,9 @@
+---
+title: OpenAI Chat
+emoji: 🤖
+colorFrom: purple
+colorTo: green
+sdk: docker
+app_file: main.py
+pinned: false
+---