From 4cc0697b66dc5be0756976afe041712ae0d90169 Mon Sep 17 00:00:00 2001 From: Weves Date: Fri, 3 Nov 2023 00:25:42 -0700 Subject: [PATCH] Standard --- .../document_index/vespa/app_config/schemas/danswer_chunk.sd | 2 +- backend/danswer/document_index/vespa/index.py | 4 ++-- 2 files changed, 3 insertions(+), 3 deletions(-) diff --git a/backend/danswer/document_index/vespa/app_config/schemas/danswer_chunk.sd b/backend/danswer/document_index/vespa/app_config/schemas/danswer_chunk.sd index f56010d37..c1c6d7dd5 100644 --- a/backend/danswer/document_index/vespa/app_config/schemas/danswer_chunk.sd +++ b/backend/danswer/document_index/vespa/app_config/schemas/danswer_chunk.sd @@ -157,7 +157,7 @@ schema danswer_chunk { global-phase { expression: (normalize_linear(closeness(field, embeddings)) + normalize_linear(bm25(content))) / 2 * document_boost * recency_bias - rerank-count: 250 + rerank-count: 1000 } # Cannot pass normalize_linear features in match-features diff --git a/backend/danswer/document_index/vespa/index.py b/backend/danswer/document_index/vespa/index.py index a93597d61..c83649f72 100644 --- a/backend/danswer/document_index/vespa/index.py +++ b/backend/danswer/document_index/vespa/index.py @@ -484,7 +484,7 @@ class VespaIndex(DocumentIndex): f"{HIDDEN}, " f"{DOC_UPDATED_AT}, " f"{METADATA} " - # f"{CONTENT_SUMMARY} " + f"{CONTENT_SUMMARY} " f"from {DOCUMENT_INDEX_NAME} where " ) @@ -690,7 +690,7 @@ class VespaIndex(DocumentIndex): decay_multiplier = FAVOR_RECENT_DECAY_MULTIPLIER if favor_recent else 1 vespa_where_clauses = _build_vespa_filters(filters) # Needs to be at least as much as the value set in Vespa schema config - target_hits = max(10 * num_to_retrieve, 250) + target_hits = max(10 * num_to_retrieve, 1000) yql = ( VespaIndex.yql_base + vespa_where_clauses