feat: stable final version with pinned deps and MODEL_URL env var

- Add requirements.txt with pinned llama-cpp-python[server]==0.2.90, pydantic<2
- Dockerfile uses requirements.txt + cmake/build-essential, start.sh as entrypoint
- start.sh downloads via MODEL_URL env var with curl, skips if exists
- Add .huggingface.yml with sdk: docker

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

.huggingface.yml

@@ -0,0 +1 @@
+sdk: docker

Dockerfile

@@ -2,35 +2,22 @@ FROM python:3.10-slim
 
 ENV DEBIAN_FRONTEND=noninteractive
 ENV PYTHONUNBUFFERED=1
+ENV PIP_NO_CACHE_DIR=1
 
 RUN apt-get update && apt-get install -y --no-install-recommends \
     curl \
-    && rm -rf /var/lib/apt/lists/*
+    build-essential \
+    cmake \
+ && rm -rf /var/lib/apt/lists/*
 
-RUN pip install --no-cache-dir \
-    huggingface-hub \
-    uvicorn \
-    fastapi \
-    sse-starlette \
-    pydantic-settings \
-    starlette-context
+WORKDIR /app
 
-RUN pip install --no-cache-dir \
-    llama-cpp-python \
-    --index-url https://abetlen.github.io/llama-cpp-python/whl/cpu
+COPY requirements.txt .
+RUN pip install -r requirements.txt
+
+COPY start.sh .
+RUN chmod +x start.sh
 
 EXPOSE 7860
 
-CMD bash -c "mkdir -p /models && \
-    echo 'Downloading model...' && \
-    curl -L \
-        -H 'Authorization: Bearer '\"${HF_TOKEN}\" \
-        'https://huggingface.co/bldeaw/guardrails_typhoon25_q4_k_m/resolve/main/guardrails_typhoon25_q4_k_m.gguf' \
-        -o /models/model.gguf && \
-    echo 'Starting server...' && \
-    python -m llama_cpp.server \
-        --model /models/model.gguf \
-        --host 0.0.0.0 \
-        --port 7860 \
-        --n_ctx 512 \
-        --n_threads 2"
+CMD ["./start.sh"]

requirements.txt

@@ -0,0 +1,4 @@
+llama-cpp-python[server]==0.2.90
+pydantic<2
+fastapi<0.111
+uvicorn<0.30

start.sh

@@ -1,18 +1,24 @@
 #!/bin/bash
 set -e
 
-mkdir -p /models
+MODEL_DIR=/models
+MODEL_FILE=typhoon.gguf
 
-echo "Downloading model..."
-huggingface-cli download bldeaw/guardrails_typhoon25_q4_k_m \
-  guardrails_typhoon25_q4_k_m.gguf \
-  --local-dir /models \
-  --token "${HF_TOKEN}"
+mkdir -p $MODEL_DIR
+
+if [ ! -f "$MODEL_DIR/$MODEL_FILE" ]; then
+  echo "Downloading model..."
+  curl -L "$MODEL_URL" -o "$MODEL_DIR/$MODEL_FILE"
+  echo "✓ Downloaded"
+else
+  echo "✓ Model already exists"
+fi
+
+echo "Starting server..."
 
-echo "Starting llama-cpp-python server..."
 python -m llama_cpp.server \
-  --model /models/guardrails_typhoon25_q4_k_m.gguf \
+  --model $MODEL_DIR/$MODEL_FILE \
   --host 0.0.0.0 \
   --port 7860 \
-  --n_ctx 512 \
+  --n_ctx 1024 \
   --n_threads 2