Update curated.py

curated.py

@@ -535,7 +535,7 @@ def get_se_data(data_source: str = "StackExchange", doc_id: int = 3, target: str
         target=target,
     )
 
-se_examples = DV2("data/curated_samples/stackexchange_raw.json", "data/curated_samples/stackexchange_extract.json", 0)
+se_examples = DV2("data/curated_samples/stackexchange_raw.json", "data/curated_samples/stackexchange_extract.json", 3)
 
 def get_phil_data(data_source: str = "PhilPapers", doc_id: int = 3, target: str = "foo"):
     doc_id = max(0, min(int(doc_id), 9))
@@ -558,7 +558,7 @@ def get_phil_data(data_source: str = "PhilPapers", doc_id: int = 3, target: str
         target=target,
     )
 
-phil_examples = DV("data/curated_samples/philpapers_raw.json", 0, "PhilPapers")
+phil_examples = DV("data/curated_samples/philpapers_raw.json", 2, "PhilPapers")
 
 def get_arx_data(data_source: str = "Arxiv", doc_id: int = 3, target: str = "foo"):
     doc_id = max(0, min(int(doc_id), 9))
@@ -582,7 +582,7 @@ def get_arx_data(data_source: str = "Arxiv", doc_id: int = 3, target: str = "foo
         target=target,
     )
 
-arx_examples = DV("data/curated_samples/arxiv_extract.json", 0, "Arxiv")
+arx_examples = DV2("data/curated_samples/arxiv_raw.json", "data/curated_samples/arxiv_extract.json", 3)
 
 def get_S2ORC_data(data_source: str = "S2ORC", doc_id: int = 3, target: str = "foo"):
     doc_id = max(0, min(int(doc_id), 9))
@@ -652,7 +652,7 @@ def get_pubmed_data(data_source: str = "Pubmed", doc_id: int = 3, target: str =
         target=target,
     )
 
-pubmed_examples = DV("data/curated_samples/pubmed_extract.json", 0, "PubMed")
+pubmed_examples = DV2("data/curated_samples/pubmed_raw.json", "data/curated_samples/pubmed_extract.json", 3)
 
 def get_dmm_data(data_source: str = "DM Math", doc_id: int = 3, target: str = "foo"):
     doc_id = max(0, min(int(doc_id), 9))
@@ -676,7 +676,7 @@ def get_dmm_data(data_source: str = "DM Math", doc_id: int = 3, target: str = "f
         target=target,
     )
 
-dmm_examples = DV("data/curated_samples/dm_maths_extract.json", 0, "DM Math")
+dmm_examples = DV("data/curated_samples/dm_maths_raw.json", "data/curated_samples/dm_maths_extract.json", 3)
 
 def get_pg19_data(data_source: str = "PG19", doc_id: int = 3, target: str = "foo"):
     doc_id = max(0, min(int(doc_id), 9))