compute FID only on selected samples, following the exp settings in Aug-PE

Author: fjxmlzn

example/text/openreview_huggingface/main.py

@@ -25,6 +25,7 @@
 from pe.callback import SaveTextToCSV
 from pe.logger import CSVPrint
 from pe.logger import LogPrint
+from pe.constant.data import VARIATION_API_FOLD_ID_COLUMN_NAME
 
 import pandas as pd
 import os
@@ -57,7 +58,9 @@
     )
 
     save_checkpoints = SaveCheckpoints(os.path.join(exp_folder, "checkpoint"))
-    compute_fid = ComputeFID(priv_data=data, embedding=embedding)
+    compute_fid = ComputeFID(
+        priv_data=data, embedding=embedding, filter_criterion={VARIATION_API_FOLD_ID_COLUMN_NAME: -1}
+    )
     save_text_to_csv = SaveTextToCSV(output_folder=os.path.join(exp_folder, "synthetic_text"))
 
     csv_print = CSVPrint(output_folder=exp_folder)

example/text/openreview_openai/main.py

@@ -48,6 +48,7 @@
 from pe.callback import SaveTextToCSV
 from pe.logger import CSVPrint
 from pe.logger import LogPrint
+from pe.constant.data import VARIATION_API_FOLD_ID_COLUMN_NAME
 
 import pandas as pd
 import os
@@ -87,7 +88,9 @@
     )
 
     save_checkpoints = SaveCheckpoints(os.path.join(exp_folder, "checkpoint"))
-    compute_fid = ComputeFID(priv_data=data, embedding=embedding)
+    compute_fid = ComputeFID(
+        priv_data=data, embedding=embedding, filter_criterion={VARIATION_API_FOLD_ID_COLUMN_NAME: -1}
+    )
     save_text_to_csv = SaveTextToCSV(output_folder=os.path.join(exp_folder, "synthetic_text"))
 
     csv_print = CSVPrint(output_folder=exp_folder)

example/text/pubmed_huggingface/main.py

@@ -25,6 +25,7 @@
 from pe.callback import SaveTextToCSV
 from pe.logger import CSVPrint
 from pe.logger import LogPrint
+from pe.constant.data import VARIATION_API_FOLD_ID_COLUMN_NAME
 
 import pandas as pd
 import os
@@ -57,7 +58,9 @@
     )
 
     save_checkpoints = SaveCheckpoints(os.path.join(exp_folder, "checkpoint"))
-    compute_fid = ComputeFID(priv_data=data, embedding=embedding)
+    compute_fid = ComputeFID(
+        priv_data=data, embedding=embedding, filter_criterion={VARIATION_API_FOLD_ID_COLUMN_NAME: -1}
+    )
     save_text_to_csv = SaveTextToCSV(output_folder=os.path.join(exp_folder, "synthetic_text"))
 
     csv_print = CSVPrint(output_folder=exp_folder)

example/text/pubmed_openai/main.py

@@ -48,6 +48,7 @@
 from pe.callback import SaveTextToCSV
 from pe.logger import CSVPrint
 from pe.logger import LogPrint
+from pe.constant.data import VARIATION_API_FOLD_ID_COLUMN_NAME
 
 import pandas as pd
 import os
@@ -87,7 +88,9 @@
     )
 
     save_checkpoints = SaveCheckpoints(os.path.join(exp_folder, "checkpoint"))
-    compute_fid = ComputeFID(priv_data=data, embedding=embedding)
+    compute_fid = ComputeFID(
+        priv_data=data, embedding=embedding, filter_criterion={VARIATION_API_FOLD_ID_COLUMN_NAME: -1}
+    )
     save_text_to_csv = SaveTextToCSV(output_folder=os.path.join(exp_folder, "synthetic_text"))
 
     csv_print = CSVPrint(output_folder=exp_folder)

example/text/yelp_huggingface/main.py

@@ -25,6 +25,7 @@
 from pe.callback import SaveTextToCSV
 from pe.logger import CSVPrint
 from pe.logger import LogPrint
+from pe.constant.data import VARIATION_API_FOLD_ID_COLUMN_NAME
 
 import pandas as pd
 import os
@@ -57,7 +58,9 @@
     )
 
     save_checkpoints = SaveCheckpoints(os.path.join(exp_folder, "checkpoint"))
-    compute_fid = ComputeFID(priv_data=data, embedding=embedding)
+    compute_fid = ComputeFID(
+        priv_data=data, embedding=embedding, filter_criterion={VARIATION_API_FOLD_ID_COLUMN_NAME: -1}
+    )
     save_text_to_csv = SaveTextToCSV(output_folder=os.path.join(exp_folder, "synthetic_text"))
 
     csv_print = CSVPrint(output_folder=exp_folder)

example/text/yelp_openai/main.py

@@ -48,6 +48,7 @@
 from pe.callback import SaveTextToCSV
 from pe.logger import CSVPrint
 from pe.logger import LogPrint
+from pe.constant.data import VARIATION_API_FOLD_ID_COLUMN_NAME
 
 import pandas as pd
 import os
@@ -87,7 +88,9 @@
     )
 
     save_checkpoints = SaveCheckpoints(os.path.join(exp_folder, "checkpoint"))
-    compute_fid = ComputeFID(priv_data=data, embedding=embedding)
+    compute_fid = ComputeFID(
+        priv_data=data, embedding=embedding, filter_criterion={VARIATION_API_FOLD_ID_COLUMN_NAME: -1}
+    )
     save_text_to_csv = SaveTextToCSV(output_folder=os.path.join(exp_folder, "synthetic_text"))
 
     csv_print = CSVPrint(output_folder=exp_folder)