tweak

Dockerfile

@@ -1,9 +1,25 @@
+# Use the Python 3.11 base image
 FROM python:3.11
 
+# Set the working directory
 WORKDIR /usr/src/app
+
+# Copy the application code into the container
 COPY . .
+
+# Install the Python dependencies from requirements.txt
 RUN pip install -r requirements.txt
+
+# Install huggingface-cli
+RUN pip install huggingface-cli
+
+# Expose the port on which your Gradio app runs
 EXPOSE 7860
+
+# Set an environment variable for the Gradio server name
 ENV GRADIO_SERVER_NAME="0.0.0.0"
 
-CMD ["python", "app.py"]
+# Define an entrypoint script to run both commands
+COPY entrypoint.sh /usr/src/app/entrypoint.sh
+RUN chmod +x /usr/src/app/entrypoint.sh
+ENTRYPOINT ["/usr/src/app/entrypoint.sh"]

app.py

@@ -8,7 +8,7 @@ from huggingface_hub import hf_hub_download
 # huggingface-cli download LoneStriker/OpenBioLLM-Llama3-8B-GGUF --local-dir ./llama3-gguf
 llm = Llama(
     # model_path="./Phi-3-mini-4k-instruct-q4.gguf",
-    model_path="LoneStriker/OpenBioLLM-Llama3-8B-Q5_K_M.gguf",
+    model_path="./llama3-gguf/OpenBioLLM-Llama3-8B-Q5_K_M.gguf",
     n_ctx=2048,
     n_gpu_layers=50, # change n_gpu_layers if you have more or less VRAM 
 ) 

entrypoint.sh

@@ -0,0 +1,7 @@
+#!/bin/bash
+
+# Download the Hugging Face model
+huggingface-cli repo download LoneStriker/OpenBioLLM-Llama3-8B-GGUF --model-id Llama3-8B-Q5_K_M.gguf --target-dir ./llama3-gguf
+
+# Start your Python application (replace "app.py" with your actual script)
+python app.py