add msturing 100k runbook

harsha-simhadri · Feb 4, 2025 · 2b54f40 · 2b54f40
1 parent f3a7ad9
commit 2b54f40
Show file tree

Hide file tree

Showing 4 changed files with 937 additions and 6 deletions.
diff --git a/benchmark/datasets.py b/benchmark/datasets.py
@@ -433,16 +433,15 @@ def distance(self):
 # 1M slice of MSTuring dataset, with ground truth corresponding to the 10K query set
 # this is needed for backwards compatibility with the streaming code
 class MSTuringANNSPQ(BillionScaleDatasetCompetitionFormat):
-    def __init__(self, nb_M=1):
-        self.nb_M = nb_M
-        self.nb = 10**6 * nb_M  
+    def __init__(self, nb):
+        self.nb = nb
         self.d = 100
         self.nq = 10000
         self.dtype = "float32"
         self.ds_fn = "base1b.fbin"
         self.qs_fn = "testQuery10K.fbin"
         self.gt_fn = (
-            "msturing-1M-private-gt100" if self.nb_M == 1 else
+            "msturing-1M-private-gt100" if self.nb == 1000000 else
             None
         )
         self.base_url = "https://comp21storage.z5.web.core.windows.net/comp21/MSFT-TURING-ANNS/"
@@ -1309,7 +1308,9 @@ def short_name(self):
     'msturing-10M': lambda : MSTuringANNS(10),
     'msturing-1M': lambda : MSTuringANNS(1),
 
-    'msturingpq-1M': lambda : MSTuringANNSPQ(1),
+    'msturingpq-100K': lambda : MSTuringANNSPQ(100000),
+    'msturingpq-1M': lambda : MSTuringANNSPQ(1000000),
+    'msturingpq-10M': lambda : MSTuringANNSPQ(10000000),
 
     'msturing-10M-clustered': lambda: MSTuringClustered10M(),
     'msturing-30M-clustered': lambda: MSTuringClustered30M(),

diff --git a/neurips23/runbooks/gen_expiration_time_runbook.py b/neurips23/runbooks/gen_expiration_time_runbook.py
@@ -217,3 +217,14 @@ def gen_exp_time_runbook(dataset_name, dataset_size, max_t, runbook_filename, ra
 max_points = 270000
 gen_exp_time_runbook(dataset_name, dataset_size, max_t, dataset_file, ratios, timesteps, seed, True, gt_url, max_points)
 
+ratios = (0, 4, 18)
+timesteps = (0, 50, 5)
+seed = 5554
+dataset_file = 'msturingpq-100K_expiration_time_runbook.yaml'
+dataset_name = 'msturingpq-100K'
+dataset_size = 100000
+max_t = 100
+gt_url = None
+max_points = 28000
+gen_exp_time_runbook(dataset_name, dataset_size, max_t, dataset_file, ratios, timesteps, seed, False, gt_url, True, max_points)
+
diff --git a/neurips23/runbooks/msturing-10M_slidingwindow_runbook.yaml b/neurips23/runbooks/msturing-10M_slidingwindow_runbook.yaml
@@ -1,4 +1,4 @@
-msturing-10M:
+msturingpq-10M:
   1:
     end: 50000
     operation: insert