switch to text-only dataset for offline build

cache_ds.py

@@ -1,5 +1,12 @@
 # cache_ds.py
+# cache_ds.py  –- build-time pre-cache (runs ONCE on HF builder)
 from datasets import load_dataset
-# downloads & caches the parquet branch into /code/.cache
-load_dataset("NimrodDev/LD_Events2", revision="refs/convert/parquet", split="train")
-print("Dataset cached.")
+from sentence_transformers import SentenceTransformer
+
+# 1. plain JSON dataset – zero custom features
+ds = load_dataset("NimrodDev/LD_Events_TEXT", split="train", keep_in_memory=True)
+print("✓ Text dataset cached at build time")
+
+# 2. embedding model
+SentenceTransformer("sentence-transformers/all-MiniLM-L6-v2")
+print("✓ Embedding model cached at build time")