Optimize model loading with cache (#83)

.github/workflows/ci-cd.yaml

@@ -21,6 +21,7 @@ on:
 
 jobs:
   sync-to-huggingface-space:
+    if: github.ref == 'refs/heads/master'
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v3

app.py

@@ -44,7 +44,7 @@ with col2:
     ###### ➠ If you want to translate the subtitles to English, select the task as "Translate" 
     ###### I recommend starting with the base model and then experimenting with the larger models, the small and medium models often work well. """)
 
-
+@st.cache_resource
 def change_model(current_size, size):
     if current_size != size:
         loaded_model = whisper.load_model(size)