diff --git a/hadith-ingestion/scripts/generate_embeddings.py b/hadith-ingestion/scripts/generate_embeddings.py index c266302..5796c2d 100755 --- a/hadith-ingestion/scripts/generate_embeddings.py +++ b/hadith-ingestion/scripts/generate_embeddings.py @@ -28,8 +28,8 @@ DB_CONFIG = { 'password': 'hadith_ingest' # UPDATE THIS } -# BATCH_SIZE = 8 # Process 32 hadiths at a time -BATCH_SIZE = 32 # Process 32 hadiths at a time +BATCH_SIZE = 8 # Process 32 hadiths at a time +# BATCH_SIZE = 32 # Process 32 hadiths at a time MAX_TEXT_LENGTH = 1500 # Truncate individual texts to avoid issues COLLECTION_NAME = "hadith_embeddings" VERIFY_SSL = False # Ignore SSL certificate verification