testing gradio caching

z_embedding.py

@@ -11,7 +11,11 @@ CACHE_SUMMARY_EMB_NPY = "app_cache/summary_vectors.npy"
 
 # Load Model
 #   setting this at global level because entire runtime will continue to use this model.
-model = SentenceTransformer(EMB_MODEL)
+
+import gradio as gr
+
+if gr.NO_RELOAD:
+	model = SentenceTransformer(EMB_MODEL)
 
 
 

z_hypothetical_summary.py

@@ -8,8 +8,10 @@ from transformers import pipeline, set_seed
 set_seed(42)
 TRAINED_CASUAL_MODEL = "LunaticMaestro/gpt2-book-summary-generator"
 
+import gradio as gr
 
-generator_model = pipeline('text-generation', model=TRAINED_CASUAL_MODEL)
+if gr.NO_RELOAD:
+	generator_model = pipeline('text-generation', model=TRAINED_CASUAL_MODEL)
 
 
 def generate_summaries(book_title: str, genre: Optional[str] = None, n_samples=2, top_k = 50, top_p = 0.85, ) -> list[str]: