cache SentenceTransformer model and serve app with gunicorn

Dockerfile

@@ -34,9 +34,12 @@ USER user
 ENV HOST=0.0.0.0
 EXPOSE 5001
 
+# Download data from Google Drive
 RUN gdown 1In9qAzV5t--rMmEH2R5miWpZ4IQStgFu \
 	&& unzip data.zip \
-	&& rm data.zip
+	&& rm data.zip \
+	# Download Sentence Transformers model
+	&& python3 cache_model.py 
 
 ENTRYPOINT bash run_app.sh
 

cache_model.py

@@ -0,0 +1,11 @@
+""""
+Intented for use by Dockerfile; downloads pytorch model and caches it 
+in the image, so the download doesn't have to run when app starts.
+"""
+
+from sentence_transformers import SentenceTransformer
+
+model_name = "paraphrase-MiniLM-L6-v2"
+model = SentenceTransformer(model_name)
+
+

requirements.txt

@@ -4,6 +4,7 @@ altair_transform==0.2.0
 bertopic==0.11.0
 Flask==2.2.2
 gdown==4.7.1
+gunicorn==21.2.0
 matplotlib==3.5.3
 mne==1.1.1
 numpy==1.22.4

run_app.sh

@@ -3,7 +3,7 @@
 cd indie_label_svelte
 npm start &
 cd ..
-python3 server.py
+gunicorn -b 0.0.0.0:5001 -w 4 --limit-request-line 0 server:app