Update Dockerfile

Dockerfile

@@ -1,11 +1,15 @@
 FROM huggingface/zero-gpu-gradio:latest
 
-COPY requirements.txt /tmp/requirements.txt
-RUN pip install --no-cache-dir -r /tmp/requirements.txt
-
+# --- STEP 1: llama-cpp-python prebuilt CUDA wheel FIRST ---
+# This MUST happen before any other pip install or it compiles from source
 RUN pip install --no-cache-dir llama-cpp-python \
   --extra-index-url https://abetlen.github.io/llama-cpp-python/whl/cu124
 
+# --- STEP 2: everything else ---
+COPY requirements.txt /tmp/requirements.txt
+RUN pip install --no-cache-dir -r /tmp/requirements.txt
+
+# --- STEP 3: file parsers ---
 RUN pip install --no-cache-dir python-docx pdfplumber chardet
 
 COPY app.py /home/user/app/app.py