Update Dockerfile

Dockerfile

@@ -2,15 +2,15 @@ FROM ghcr.io/ggml-org/llama.cpp:server
 
 RUN apt update && apt install wget -y && rm -rf /var/lib/apt/lists/*
 
-# Download the Q4_K_M model for faster CPU inference
+# The main model (Q4 for speed)
 RUN wget "https://huggingface.co/unsloth/gemma-4-E2B-it-GGUF/resolve/main/gemma-4-E2B-it-Q4_K_M.gguf" -O /model.gguf
 
-# Download the multimodal projector (required for Image & Audio)
-RUN wget "https://huggingface.co/unsloth/gemma-4-E2B-it-GGUF/resolve/main/mmproj-gemma-4-E2B-it-Q4_K_M.gguf" -O /mmproj.gguf
+# The multimodal projector (Corrected filename)
+RUN wget "https://huggingface.co/unsloth/gemma-4-E2B-it-GGUF/resolve/main/mmproj-model-f16.gguf" -O /mmproj.gguf
 
-# -t 2 matches your 2vCPU resources exactly
-# --mmproj enables the vision/audio features
+# Matching your 2vCPU exactly
+# Added --ctx-size 8192 to keep memory low but functional
 CMD ["-m", "/model.gguf", \
      "--mmproj", "/mmproj.gguf", \
      "--port", "7860", "--host", "0.0.0.0", \
-     "-t", "2", "--mlock", "-b", "512"]
+     "-t", "2", "--mlock", "-c", "8192"]