VectorDBBench icon indicating copy to clipboard operation
VectorDBBench copied to clipboard

Is there any naming rules for external dataset file names ?

Open cydrain opened this issue 2 months ago • 0 comments

  1. download "bioasq_medium_1m" and "bioasq_large_10m" dataset from "s3://assets.zilliz.com/benchmark"
  2. save above dataset to local dir, and treat them as customized dataset
  3. do test with following command Image
  4. vectordb-bench try to get the file with name "train.parquet" but failed Image
  5. if I add a symbol link "train.parquet", and let it link to "shuffle_train.parquet", the test can run smoothly
  6. but for "bioasq_large_10m", the train data file name is like "shuffle_train-00-of-10.parquet", vectordb-bench still want to get the file with name "train.parquet" and fail

Is there any naming rules for external dataset file names ?

cydrain avatar Nov 06 '25 02:11 cydrain