Upload folder using huggingface_hub

Dockerfile

@@ -1,7 +1,8 @@
 FROM node:18-slim
 
 # Install dependencies for building node-llama-cpp
-RUN apt-get update && apt-get install -y python3 make g++ curl
+# ADDED: cmake is required for compiling llama.cpp
+RUN apt-get update && apt-get install -y python3 make g++ curl cmake
 
 WORKDIR /app
 
@@ -9,6 +10,7 @@ WORKDIR /app
 COPY package*.json ./
 
 # Install npm dependencies
+# Note: node-llama-cpp might try to build during install
 RUN npm install
 
 # Copy source code
@@ -17,8 +19,7 @@ COPY . .
 # Create models directory
 RUN mkdir -p models
 
-# Download the model during build (so it's baked into the image)
-# Using direct download URL for speed if possible, or use node-llama-cpp pull
+# Download the model during build
 RUN npx --no node-llama-cpp pull --dir ./models hf:Qwen/Qwen3-1.7B-GGUF:Q8_0 --filename Qwen3-1.7B-Q8_0.gguf
 
 # Expose the port HF expects