dblincoe

#@ load("/lingtools/neural_ir_train.lib.yaml", "nir_loc")
#@ dag_name = "neural_ir_loc_train_large_voc_hard"
#@ start_date = "2022-03-09"
#@ end_date = "2022-03-13"
#@ model_name = "large_voc"
#@ model_name_short = "nllv"
#@ random_training_epochs = "1"
#@ is_dev = True
--- #@ nir_loc(dag_name, model_name, model_name_short, start_date, end_date=end_date, is_dev=is_dev, random_training_epochs=random_training_epochs)

dblincoe

diff --git a/projects/etsy-batchjobs-prod/services/airflow-search/dags/neural_ir_loc_train_region_language.yaml b/projects/etsy-batchjobs-prod/services/airflow-search/dags/neural_ir_loc_train_hypertuned_region_language.yaml
index 7420f3d6a..56a61d907 100644
--- a/projects/etsy-batchjobs-prod/services/airflow-search/dags/neural_ir_loc_train_region_language.yaml
+++ b/projects/etsy-batchjobs-prod/services/airflow-search/dags/neural_ir_loc_train_hypertuned_region_language.yaml
@@ -1,4 +1,4 @@
-name: neural_ir_loc_train_region_language
+name: neural_ir_loc_train_hypertuned_region_language
 
 compatibility_version: 1.5.5
 

dblincoe

diff --git a/projects/etsy-batchjobs-prod/services/airflow-search/dags/neural_ir_loc_train_region_language.yaml b/projects/etsy-batchjobs-prod/services/airflow-search/dags/neural_ir_loc_train_hypertuned_region_language.yaml
index 7420f3d6a..56a61d907 100644
--- a/projects/etsy-batchjobs-prod/services/airflow-search/dags/neural_ir_loc_train_region_language.yaml
+++ b/projects/etsy-batchjobs-prod/services/airflow-search/dags/neural_ir_loc_train_hypertuned_region_language.yaml
@@ -1,4 +1,4 @@
-name: neural_ir_loc_train_region_language
+name: neural_ir_loc_train_hypertuned_region_language
 
 compatibility_version: 1.5.5
 

dblincoe

neural_ir,region,slice_1.listingCountry.NL.recall_click_100,1,0.0,nan
neural_ir,region,slice_2.userCountry.DE.listingCountry.US.recall_purchase_100,1,0.0,nan
neural_ir,region,slice_2.userCountry.ES.listingCountry.US.recall_purchase_1000,1,0.0,nan
neural_ir,region,slice_2.userCountry.AU.listingCountry.GB.recall_purchase_100,1,0.0,nan
neural_ir,region,slice_2.userCountry.GB.listingCountry.FR.recall_purchase_1000,1,0.0,nan
neural_ir,region,slice_2.userCountry.NL.listingCountry.NL.recall_purchase_1000,1,0.0,nan
neural_ir,region,1_1_purchases.recall_purchase_100,1,0.0,nan
neural_ir,region,slice_1.userCountry.IT.recall_click_1000,1,0.0,nan
neural_ir,region,slice_2.userCountry.DE.listingCountry.ES.recall_purchase_100,1,0.0,nan
neural_ir,region,slice_2.userCountry.AU.listingCountry.GB.recall_click_100,1,0.0,nan

dblincoe

neural_ir,test,top.1.recall_purchase_10,1,0.12247191011235956,nan
neural_ir,test,top.1.recall_purchase_100,1,0.47614629440472134,nan
neural_ir,test,top.1.recall_purchase_1000,1,0.8573816819884233,nan
neural_ir,test,top.1.avg_dot_prod_purchase,1,0.5703236496528054,nan
neural_ir,test,all.recall_purchase_10,1,0.2012913832632438,nan
neural_ir,test,all.recall_purchase_100,1,0.5232166433170595,nan
neural_ir,test,all.recall_purchase_1000,1,0.843845636437557,nan
neural_ir,test,all.avg_dot_prod_purchase,1,0.5424379532936886,nan
neural_ir,test,51_inf_purchases.recall_purchase_10,1,0.08906535668159475,nan
neural_ir,test,51_inf_purchases.recall_purchase_100,1,0.36540456236652086,nan

dblincoe

neural_ir,test,no_bin.recall_purchase_10,1,0.04307894161382025,nan
neural_ir,test,no_bin.recall_purchase_100,1,0.10457030395801444,nan
neural_ir,test,no_bin.recall_purchase_1000,1,0.21337561046723522,nan
neural_ir,test,no_bin.avg_dot_prod_purchase,1,0.30870987761343394,nan
neural_ir,test,all.recall_purchase_10,1,0.034053742991315694,nan
neural_ir,test,all.recall_purchase_100,1,0.10552571926254467,nan
neural_ir,test,all.recall_purchase_1000,1,0.23486283643233122,nan
neural_ir,test,all.avg_dot_prod_purchase,1,0.31746900994696003,nan
neural_ir,test,slice_1.userCountry.AU.recall_purchase_10,1,0.030218068535825548,nan
neural_ir,test,slice_1.userCountry.AU.recall_purchase_100,1,0.08093897276140267,nan

dblincoe

diff --git a/neural_ir/neural_ir/evaluation/create_eval_queries.py b/neural_ir/neural_ir/evaluation/create_eval_queries.py
index 9659a25..4ad9be6 100644
--- a/neural_ir/neural_ir/evaluation/create_eval_queries.py
+++ b/neural_ir/neural_ir/evaluation/create_eval_queries.py
@@ -67,8 +67,12 @@ def create_eval_data(search_ranking_input_paths: List[str],
         def replace_unknown_list(examples, slice_feature_values):
             def replace_unknown_example(example, slice_feature_values):
                 for feature, values in slice_feature_values.items():
-                    if example.get(feature) not in values:
+                    example_feature = feature.split("Slice")[0]

dblincoe

custom trucker hat	en-US,US
she leaves a little sparkle wherever she goes	en-US,US
blonde hair png	en-US,US
not today succa	en-US,US
funny koozie svg	en-US,US
professional open house invites	en-US,US
air plants	en-US,US
safari jungle cake topper baby shower	en-GB,AU
let your light shine bright svg	en-US,US
I cook as good as I look svg	en-US,US

dblincoe

Samsung Galaxy s20 note haunted Mansion case
badge holder cover plastic
chanel charm
dainty fire agate ring
graduation signs for cake
custom address stamp
funny aprons svg
stop sign svg
crochet boob earrings
she leaves a little sparkle wherever she goes

dblincoe

diff --git a/neural_ir/neural_ir/evaluation/create_eval_queries.py b/neural_ir/neural_ir/evaluation/create_eval_queries.py
index da8a8ef..6029dda 100644
--- a/neural_ir/neural_ir/evaluation/create_eval_queries.py
+++ b/neural_ir/neural_ir/evaluation/create_eval_queries.py
@@ -137,26 +137,24 @@ def create_eval_data(search_ranking_input_paths: List[str],
             query_metric_slices = create_slices_from_metrics(metrics_obj)
 
             query_instance2listing2target_counts = defaultdict(dict)
-            query_instance2locale_features = defaultdict(dict)