.

README.md

@@ -161,7 +161,7 @@ MRR = 0.311 implies that there's a good change that the target book will be in r
 1. I rewrote the snippets from `z_evaluate.py` to `app.py` with minor changes to expriment with view. 
 2. To make the HF space faster (previously: 234s; **NOW: 15s**) I added `gr.NO_RELOAD` context block in the module files when loading embedding and generation model.
   Reference: https://www.gradio.app/guides/developing-faster-with-reload-mode
-
+3. DONT set `debug=True` for gradio in HF space, else it doesn't start.
 
 
 

app.py

@@ -12,8 +12,11 @@ N_RECOMMENDS = 5
 # generator_model = pipeline('text-generation', model=TRAINED_CASUAL_MODEL)
 
 import gradio as gr
-from z_similarity import computes_similarity_w_hypothetical
-from z_hypothetical_summary import generate_summaries
+
+if gr.NO_RELOAD:
+    from z_similarity import computes_similarity_w_hypothetical
+    from z_hypothetical_summary import generate_summaries
+    
 from z_utils import get_dataframe
 
 books_df = get_dataframe(CLEAN_DF_UNIQUE_TITLES)