microsoft / DiskANN

Graph-structured Indices for Scalable, Fast, Fresh and Filtered Approximate Nearest Neighbor Search
Other
1.16k stars 227 forks source link

Update Workflows #603

Open ah89 opened 4 days ago

ah89 commented 4 days ago

Some of the app arguments, along with the example descriptions and commands in the workflows, are outdated.

For example:

> compute_groundtruth -h
            Arguments:
              -h [ --help ]         Print information on arguments
              --data_type arg       data type <int8/uint8/float>
              --dist_fn arg         distance function <l2/mips/cosine>
              --base_file arg       File containing the base vectors in binary format
              --query_file arg      File containing the query vectors in binary format
              --gt_file arg         File name for the writing ground truth in binary 
                                    format, please don' append .bin at end if no 
                                    filter_label or filter_label_file is provided it will 
                                    save the file with '.bin' at end.else it will save the 
                                    file as filename_label.bin
              --K arg               Number of ground truth nearest neighbors to compute
              --tags_file arg       File containing the tags in binary format

how ever in the Sift example shown: build/apps/utils/compute_groundtruth --data_type float --dist_fn l2 --base_file siftsmall/siftsmall_base.bin --query_file siftsmall/siftsmall_query.bin --gt_file siftsmall/siftsmall_gt_35.bin --K 100 --label_file ./rand_labels_50_10K.txt --filter_label 35 --universal_label 0

Where label_file, filter_label, universal_label need to be deleted also tags_file argument is missing.