Update Dockerfile

Dockerfile

@@ -3,6 +3,7 @@ FROM python:3.12-slim
 ENV DEBIAN_FRONTEND=noninteractive
 ENV PYTHONUNBUFFERED=1
 ENV PIP_NO_CACHE_DIR=1
+ENV LD_LIBRARY_PATH=/opt/llama
 
 WORKDIR /app
 
@@ -15,8 +16,6 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
     libgomp1 \
     && rm -rf /var/lib/apt/lists/*
 
-
-# Build llama.cpp for CPU.
 RUN git clone --depth 1 https://github.com/ggml-org/llama.cpp.git /tmp/llama.cpp \
     && cmake -S /tmp/llama.cpp -B /tmp/llama.cpp/build \
         -DGGML_NATIVE=OFF \
@@ -29,21 +28,31 @@ RUN git clone --depth 1 https://github.com/ggml-org/llama.cpp.git /tmp/llama.cpp
         --target llama-server \
         -j2 \
     && mkdir -p /opt/llama \
-    && cp /tmp/llama.cpp/build/bin/llama-server /opt/llama/llama-server \
-    && cp /tmp/llama.cpp/build/bin/*.so* /opt/llama/ 2>/dev/null || true \
+    && cp /tmp/llama.cpp/build/bin/llama-server /opt/llama/ \
+    && cp /tmp/llama.cpp/build/bin/*.so* /opt/llama/ \
     && rm -rf /tmp/llama.cpp
 
+RUN pip install huggingface_hub
 
-# Hugging Face downloader.
-RUN pip install --upgrade pip \
-    && pip install huggingface_hub
-
-
-COPY start.sh /app/start.sh
-RUN chmod +x /app/start.sh
-
+RUN python3 -c "\
+from huggingface_hub import hf_hub_download; \
+hf_hub_download( \
+    repo_id='LiquidAI/LFM2.5-2.6B-GGUF', \
+    filename='LFM2.5-2.6B-QAD-Q4_0.gguf', \
+    local_dir='/app/models' \
+)"
 
 EXPOSE 7860
 
-
-CMD ["/app/start.sh"]
+CMD ["/opt/llama/llama-server", \
+     "--model", "/app/models/LFM2.5-2.6B-QAD-Q4_0.gguf", \
+     "--host", "0.0.0.0", \
+     "--port", "7860", \
+     "--alias", "LFM2.5-2.6B", \
+     "--threads", "2", \
+     "--ctx-size", "32768", \
+     "--cache-type-k", "q8_0", \
+     "--cache-type-v", "q8_0", \
+     "--parallel", "1", \
+     "--cont-batching", \
+     "--flash-attn", "auto"]