diff --git a/docs/benchmarks.md b/docs/benchmarks.md index a7e65a1d..29fd9207 100644 --- a/docs/benchmarks.md +++ b/docs/benchmarks.md @@ -35,7 +35,8 @@ Available datasets: | `repliqa` | ~30MB | | `hotpotqa` | ~331MB | | `wix` | ~511MB | -| `open_rag_bench` | ~14GB | +| `open_rag_bench` (text embedder + VLM picture descriptions) | ~15GB | +| `open_rag_bench` (multimodal embedder, override with `--db`) | ~16GB | After downloading, run benchmarks with `--skip-db` to use the pre-built database: diff --git a/evaluations/evaluations/datasets/open_rag_bench.py b/evaluations/evaluations/datasets/open_rag_bench.py index f11be906..35f07204 100644 --- a/evaluations/evaluations/datasets/open_rag_bench.py +++ b/evaluations/evaluations/datasets/open_rag_bench.py @@ -218,7 +218,7 @@ def is_multimodal_query(source: str) -> bool: OPEN_RAG_BENCH_SPEC = DatasetSpec( key="orb", - db_filename="open_rag_bench.lancedb", + db_filename="open_rag_bench_text.lancedb", document_loader=load_orb_corpus, document_mapper=map_orb_document, qa_loader=load_orb_qa,