mteb 2.7.3__py3-none-any.whl → 2.7.5__py3-none-any.whl

mteb/tasks/retrieval/eng/chat_doctor_retrieval.py

@@ -36,7 +36,7 @@ class ChatDoctorRetrieval(AbsTaskRetrieval):
         },
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/eng/fin_qa_retrieval.py

@@ -37,7 +37,7 @@ class FinQARetrieval(AbsTaskRetrieval):
         },
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/eng/finance_bench_retrieval.py

@@ -37,7 +37,7 @@ class FinanceBenchRetrieval(AbsTaskRetrieval):
         },
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/eng/hateful_memes_i2t_retrieval.py

@@ -90,7 +90,7 @@ class HatefulMemesI2TRetrieval(AbsTaskRetrieval):
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
         self.corpus, self.queries, self.relevant_docs = _load_data(

mteb/tasks/retrieval/eng/hateful_memes_t2i_retrieval.py

@@ -90,7 +90,7 @@ class HatefulMemesT2IRetrieval(AbsTaskRetrieval):
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
         self.corpus, self.queries, self.relevant_docs = _load_data(

mteb/tasks/retrieval/eng/hc3_finance_retrieval.py

@@ -37,7 +37,7 @@ class HC3FinanceRetrieval(AbsTaskRetrieval):
         },
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/eng/lemb_narrative_qa_retrieval.py

@@ -55,7 +55,7 @@ Roark, Brian},
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/eng/lemb_needle_retrieval.py

@@ -46,7 +46,7 @@ class LEMBNeedleRetrieval(AbsTaskRetrieval):
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/eng/lemb_passkey_retrieval.py

@@ -46,7 +46,7 @@ class LEMBPasskeyRetrieval(AbsTaskRetrieval):
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/eng/lemb_summ_screen_fd_retrieval.py

@@ -51,7 +51,7 @@ Villavicencio, Aline},
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/eng/lemb_wikim_qa_retrieval.py

@@ -40,7 +40,7 @@ class LEMBWikimQARetrieval(AbsTaskRetrieval):
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/eng/lembqm_sum_retrieval.py

@@ -64,7 +64,7 @@ Zhou, Yichao},
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/eng/lit_search_retrieval.py

@@ -35,7 +35,7 @@ class LitSearchRetrieval(AbsTaskRetrieval):
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
         self.corpus, self.queries, self.relevant_docs = {}, {}, {}

mteb/tasks/retrieval/eng/memotion_i2t_retrieval.py

@@ -113,7 +113,7 @@ class MemotionI2TRetrieval(AbsTaskRetrieval):
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
         self.corpus, self.queries, self.relevant_docs = _load_data(

mteb/tasks/retrieval/eng/memotion_t2i_retrieval.py

@@ -112,7 +112,7 @@ class MemotionT2IRetrieval(AbsTaskRetrieval):
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
         self.corpus, self.queries, self.relevant_docs = _load_data(

mteb/tasks/retrieval/eng/ml_questions.py

@@ -55,7 +55,7 @@ Reddy, Siva},
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
         self.corpus, self.queries, self.relevant_docs = {}, {}, {}

mteb/tasks/retrieval/eng/nano_argu_ana_retrieval.py

@@ -40,7 +40,7 @@ class NanoArguAnaRetrieval(AbsTaskRetrieval):
         adapted_from=["ArguAna"],
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/eng/nano_climate_fever_retrieval.py

@@ -44,7 +44,7 @@ class NanoClimateFeverRetrieval(AbsTaskRetrieval):
         adapted_from=["ClimateFEVER"],
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/eng/nano_db_pedia_retrieval.py

@@ -42,7 +42,7 @@ class NanoDBPediaRetrieval(AbsTaskRetrieval):
         adapted_from=["DBPedia"],
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/eng/nano_fever_retrieval.py

@@ -57,7 +57,7 @@ Stent, Amanda},
         adapted_from=["FEVER"],
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/eng/nano_fi_qa2018_retrieval.py

@@ -43,7 +43,7 @@ class NanoFiQA2018Retrieval(AbsTaskRetrieval):
         adapted_from=["FiQA2018"],
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/eng/nano_hotpot_qa_retrieval.py

@@ -60,7 +60,7 @@ Tsujii, Jun{'}ichi},
         adapted_from=["HotpotQA"],
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/eng/nano_msmarco_retrieval.py

@@ -55,7 +55,7 @@ Li Deng},
         adapted_from=["MSMARCO"],
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/eng/nano_nf_corpus_retrieval.py

@@ -46,7 +46,7 @@ class NanoNFCorpusRetrieval(AbsTaskRetrieval):
         adapted_from=["NFCorpus"],
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/eng/nano_nq_retrieval.py

@@ -46,7 +46,7 @@ Linguistics},
         adapted_from=["NQ"],
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/eng/nano_quora_retrieval.py

@@ -45,7 +45,7 @@ class NanoQuoraRetrieval(AbsTaskRetrieval):
         adapted_from=["QuoraRetrieval"],
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/eng/nano_sci_fact_retrieval.py

@@ -42,7 +42,7 @@ class NanoSciFactRetrieval(AbsTaskRetrieval):
         adapted_from=["SciFact"],
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/eng/nano_scidocs_retrieval.py

@@ -44,7 +44,7 @@ class NanoSCIDOCSRetrieval(AbsTaskRetrieval):
         adapted_from=["SCIDOCS"],
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/eng/nano_touche2020_retrieval.py

@@ -53,7 +53,7 @@ Questions}},
         adapted_from=["Touche2020"],
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/eng/narrative_qa_retrieval.py

@@ -43,7 +43,7 @@ class NarrativeQARetrieval(AbsTaskRetrieval):
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/eng/r2_med_retrieval.py

@@ -70,7 +70,7 @@ class R2MEDBiologyRetrieval(AbsTaskRetrieval):
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 
@@ -114,7 +114,7 @@ class R2MEDBioinformaticsRetrieval(AbsTaskRetrieval):
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 
@@ -158,7 +158,7 @@ class R2MEDMedicalSciencesRetrieval(AbsTaskRetrieval):
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 
@@ -202,7 +202,7 @@ class R2MEDMedXpertQAExamRetrieval(AbsTaskRetrieval):
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 
@@ -246,7 +246,7 @@ class R2MEDMedQADiagRetrieval(AbsTaskRetrieval):
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 
@@ -290,7 +290,7 @@ class R2MEDPMCTreatmentRetrieval(AbsTaskRetrieval):
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 
@@ -334,7 +334,7 @@ class R2MEDPMCClinicalRetrieval(AbsTaskRetrieval):
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 
@@ -378,7 +378,7 @@ class R2MEDIIYiClinicalRetrieval(AbsTaskRetrieval):
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/eng/sci_mmir_i2t_retrieval.py

@@ -88,7 +88,7 @@ class SciMMIRI2TRetrieval(AbsTaskRetrieval):
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
         self.corpus, self.queries, self.relevant_docs = _load_data(

mteb/tasks/retrieval/eng/sci_mmir_t2i_retrieval.py

@@ -88,7 +88,7 @@ class SciMMIRT2IRetrieval(AbsTaskRetrieval):
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
         self.corpus, self.queries, self.relevant_docs = _load_data(

mteb/tasks/retrieval/eng/vidore_bench_retrieval.py

@@ -95,7 +95,7 @@ class VidoreArxivQARetrieval(AbsTaskRetrieval):
         prompt={"query": "Find a screenshot that relevant to the user's question."},
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         self.corpus, self.queries, self.relevant_docs = _load_data(
             path=self.metadata.dataset["path"],
             splits=self.metadata.eval_splits,
@@ -138,7 +138,7 @@ class VidoreDocVQARetrieval(AbsTaskRetrieval):
         prompt={"query": "Find a screenshot that relevant to the user's question."},
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         self.corpus, self.queries, self.relevant_docs = _load_data(
             path=self.metadata.dataset["path"],
             splits=self.metadata.eval_splits,
@@ -181,7 +181,7 @@ class VidoreInfoVQARetrieval(AbsTaskRetrieval):
         prompt={"query": "Find a screenshot that relevant to the user's question."},
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         self.corpus, self.queries, self.relevant_docs = _load_data(
             path=self.metadata.dataset["path"],
             splits=self.metadata.eval_splits,
@@ -224,7 +224,7 @@ class VidoreTabfquadRetrieval(AbsTaskRetrieval):
         prompt={"query": "Find a screenshot that relevant to the user's question."},
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         self.corpus, self.queries, self.relevant_docs = _load_data(
             path=self.metadata.dataset["path"],
             splits=self.metadata.eval_splits,
@@ -267,7 +267,7 @@ class VidoreTatdqaRetrieval(AbsTaskRetrieval):
         prompt={"query": "Find a screenshot that relevant to the user's question."},
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         self.corpus, self.queries, self.relevant_docs = _load_data(
             path=self.metadata.dataset["path"],
             splits=self.metadata.eval_splits,
@@ -310,7 +310,7 @@ class VidoreShiftProjectRetrieval(AbsTaskRetrieval):
         prompt={"query": "Find a screenshot that relevant to the user's question."},
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         self.corpus, self.queries, self.relevant_docs = _load_data(
             path=self.metadata.dataset["path"],
             splits=self.metadata.eval_splits,
@@ -354,7 +354,7 @@ class VidoreSyntheticDocQAAIRetrieval(AbsTaskRetrieval):
         adapted_from=["VidoreDocVQARetrieval"],
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         self.corpus, self.queries, self.relevant_docs = _load_data(
             path=self.metadata.dataset["path"],
             splits=self.metadata.eval_splits,
@@ -398,7 +398,7 @@ class VidoreSyntheticDocQAEnergyRetrieval(AbsTaskRetrieval):
         adapted_from=["VidoreDocVQARetrieval"],
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         self.corpus, self.queries, self.relevant_docs = _load_data(
             path=self.metadata.dataset["path"],
             splits=self.metadata.eval_splits,
@@ -442,7 +442,7 @@ class VidoreSyntheticDocQAGovernmentReportsRetrieval(AbsTaskRetrieval):
         adapted_from=["VidoreDocVQARetrieval"],
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         self.corpus, self.queries, self.relevant_docs = _load_data(
             path=self.metadata.dataset["path"],
             splits=self.metadata.eval_splits,
@@ -486,7 +486,7 @@ class VidoreSyntheticDocQAHealthcareIndustryRetrieval(AbsTaskRetrieval):
         adapted_from=["VidoreDocVQARetrieval"],
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         self.corpus, self.queries, self.relevant_docs = _load_data(
             path=self.metadata.dataset["path"],
             splits=self.metadata.eval_splits,

mteb/tasks/retrieval/fra/f_qu_ad_retrieval.py

@@ -49,7 +49,7 @@ Liu, Yang},
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
         dataset_raw = datasets.load_dataset(

mteb/tasks/retrieval/fra/syntec_retrieval.py

@@ -38,7 +38,7 @@ class SyntecRetrieval(AbsTaskRetrieval):
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
         # fetch both subsets of the dataset

mteb/tasks/retrieval/hun/hun_sum2.py

@@ -43,7 +43,7 @@ class HunSum2AbstractiveRetrieval(AbsTaskRetrieval):
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
         self.corpus, self.queries, self.relevant_docs = {}, {}, {}

mteb/tasks/retrieval/kat/georgian_faq_retrieval.py

@@ -33,7 +33,7 @@ class GeorgianFAQRetrieval(AbsTaskRetrieval):
         bibtex_citation="",
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/multilingual/cross_lingual_semantic_discrimination_wmt19.py

@@ -53,7 +53,7 @@ class CrossLingualSemanticDiscriminationWMT19(AbsTaskRetrieval):
     )
     num_of_distractors = 4
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         """Generic data loader function for original clsd datasets with the format shown in "hf_dataset_link".
         Loading the hf dataset, it populates the following three variables to be used for retrieval evaluation.
 

mteb/tasks/retrieval/multilingual/cross_lingual_semantic_discrimination_wmt21.py

@@ -54,7 +54,7 @@ class CrossLingualSemanticDiscriminationWMT21(AbsTaskRetrieval):
 
     num_of_distractors = 4
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         """Generic data loader function for original clsd datasets with the format shown in "hf_dataset_link".
         Loading the hf dataset, it populates the following three variables to be used for retrieval evaluation.
 

mteb/tasks/retrieval/multilingual/cur_ev1_retrieval.py

@@ -111,7 +111,7 @@ class CUREv1Retrieval(AbsTaskRetrieval):
 
         return queries
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/multilingual/jina_vdr_bench_retrieval.py

@@ -148,7 +148,7 @@ def _load_data(
     return corpus, queries, relevant_docs
 
 
-def load_data(self) -> None:
+def load_data(self, num_proc: int = 1, **kwargs) -> None:
     if self.data_loaded:
         return
 

mteb/tasks/retrieval/multilingual/miracl_vision_retrieval.py

@@ -143,7 +143,7 @@ class MIRACLVisionRetrieval(AbsTaskRetrieval):
         prompt={"query": "Find a screenshot that is relevant to the user's query."},
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/multilingual/mr_tidy_retrieval.py

@@ -108,7 +108,7 @@ class MrTidyRetrieval(AbsTaskRetrieval):
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/multilingual/public_health_qa_retrieval.py

@@ -97,7 +97,7 @@ class PublicHealthQARetrieval(AbsTaskRetrieval):
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/multilingual/ru_sci_bench_retrieval.py

@@ -103,7 +103,7 @@ class RuSciBenchCiteRetrieval(AbsTaskRetrieval):
         },
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 
@@ -161,7 +161,7 @@ class RuSciBenchCociteRetrieval(AbsTaskRetrieval):
         },
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/multilingual/statcan_dialogue_dataset_retrieval.py

@@ -96,7 +96,7 @@ de Vries, Harm},
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return
 

mteb/tasks/retrieval/multilingual/vdr_multilingual_retrieval.py

@@ -126,7 +126,7 @@ class VDRMultilingualRetrieval(AbsTaskRetrieval):
 """,
     )
 
-    def load_data(self) -> None:
+    def load_data(self, num_proc: int = 1, **kwargs) -> None:
         if self.data_loaded:
             return