Update GPT4 FewShot assets to new 'splits' API

qcri · Sep 17, 2023 · 2e6edc3 · 2e6edc3
1 parent b2e1c49
commit 2e6edc3
Show file tree

Hide file tree

Showing 56 changed files with 80 additions and 399 deletions.
diff --git a/assets/ar/QA/ARCD_GPT4_FewShot.py b/assets/ar/QA/ARCD_GPT4_FewShot.py
@@ -17,10 +17,6 @@ def config():
         "model_args": {
             "max_tries": 30,
         },
-        "general_args": {
-            "data_path": "data/QA/ARCD/arcd-test.json",
-            "fewshot": {"train_data_path": "data/QA/ARCD/arcd-train.json"},
-        },
     }
 
 

diff --git a/assets/ar/QA/MLQA_GPT4_FewShot.py b/assets/ar/QA/MLQA_GPT4_FewShot.py
@@ -17,12 +17,7 @@ def config():
         "model_args": {
             "max_tries": 30,
         },
-        "general_args": {
-            "data_path": "data/QA/MLQA/test/test-context-ar-question-ar.json",
-            "fewshot": {
-                "train_data_path": "data/QA/MLQA/dev/dev-context-ar-question-ar.json"
-            },
-        },
+        "general_args": {"fewshot": {"train_split": "dev"}},
     }
 
 

diff --git a/assets/ar/QA/TyDiQA_GPT4_FewShot.py b/assets/ar/QA/TyDiQA_GPT4_FewShot.py
@@ -17,10 +17,7 @@ def config():
         "model_args": {
             "max_tries": 30,
         },
-        "general_args": {
-            "data_path": "data/QA/tydiqa/tydiqa-goldp-dev-arabic.json",
-            "fewshot": {"train_data_path": "data/QA/ARCD/arcd-train.json"},
-        },
+        "general_args": {"test_split": "dev"},
     }
 
 

diff --git a/assets/ar/QA/XQuAD_GPT4_FewShot.py b/assets/ar/QA/XQuAD_GPT4_FewShot.py
@@ -17,10 +17,6 @@ def config():
         "model_args": {
             "max_tries": 30,
         },
-        "general_args": {
-            "data_path": "data/QA/xquad/xquad.ar.json",
-            "fewshot": {"train_data_path": "data/QA/ARCD/arcd-train.json"},
-        },
     }
 
 

diff --git a/assets/ar/demographic_attributes/gender/ArapTweet_GPT4_FewShot.py b/assets/ar/demographic_attributes/gender/ArapTweet_GPT4_FewShot.py
@@ -14,12 +14,6 @@ def config():
             "class_labels": ["Female", "Male"],
             "max_tries": 30,
         },
-        "general_args": {
-            "data_path": "data/demographic_attributes/gender/test-ARAP-unique.txt",
-            "fewshot": {
-                "train_data_path": "data/demographic_attributes/gender/train-wajdi.tsv",
-            },
-        },
     }
 
 

diff --git a/assets/ar/demographic_attributes/location/Location_GPT4_FewShot.py b/assets/ar/demographic_attributes/location/Location_GPT4_FewShot.py
@@ -38,9 +38,7 @@ def config():
             "max_tries": 30,
         },
         "general_args": {
-            "data_path": "data/demographic_attributes/location/arab+others.txt",
             "fewshot": {
-                "train_data_path": "data/demographic_attributes/location/arab+others.txt",  # TODO need to change the file
                 "deduplicate": False,
             },
         },

diff --git a/assets/ar/demographic_attributes/name_info/NameInfo_GPT4_FewShot.py b/assets/ar/demographic_attributes/name_info/NameInfo_GPT4_FewShot.py
@@ -116,9 +116,7 @@ def config():
             "max_tries": 30,
         },
         "general_args": {
-            "data_path": "data/demographic_attributes/name_info/wikidata_test.txt",
             "fewshot": {
-                "train_data_path": "data/demographic_attributes/name_info/wikidata_test.txt",  # TODO need to change the file
                 "deduplicate": False,
             },
         },

diff --git a/...r/factuality_disinformation_harmful_content/adult_content_detection/Adult_GPT4_FewShot.py b/...r/factuality_disinformation_harmful_content/adult_content_detection/Adult_GPT4_FewShot.py
@@ -15,9 +15,7 @@ def config():
             "max_tries": 30,
         },
         "general_args": {
-            "data_path": "data/factuality_disinformation_harmful_content/adult/adult-test.tsv",
             "fewshot": {
-                "train_data_path": "data/factuality_disinformation_harmful_content/adult/adult-train.tsv",
                 "deduplicate": True,
             },
         },

diff --git a/...uality_disinformation_harmful_content/attentionworthy/CT22Attentionworthy_GPT4_FewShot.py b/...uality_disinformation_harmful_content/attentionworthy/CT22Attentionworthy_GPT4_FewShot.py
@@ -24,12 +24,7 @@ def config():
             ],
             "max_tries": 30,
         },
-        "general_args": {
-            "data_path": "data/factuality_disinformation_harmful_content/attentionworthy/CT22_arabic_1D_attentionworthy_test_gold.tsv",
-            "fewshot": {
-                "train_data_path": "data/factuality_disinformation_harmful_content/attentionworthy/CT22_arabic_1D_attentionworthy_train.tsv",
-            },
-        },
+        "general_args": {"test_split": "ar", "fewshot": {"train_split": "ar"}},
     }
 
 

diff --git a/...uality_disinformation_harmful_content/checkworthyness/CT22Checkworthiness_GPT4_FewShot.py b/...uality_disinformation_harmful_content/checkworthyness/CT22Checkworthiness_GPT4_FewShot.py
@@ -16,12 +16,7 @@ def config():
             "class_labels": ["0", "1"],
             "max_tries": 30,
         },
-        "general_args": {
-            "data_path": "data/factuality_disinformation_harmful_content/checkworthyness/arabic/CT22_arabic_1A_checkworthy_test_gold.tsv",
-            "fewshot": {
-                "train_data_path": "data/factuality_disinformation_harmful_content/checkworthyness/arabic/CT22_arabic_1A_checkworthy_train.tsv",
-            },
-        },
+        "general_args": {"test_split": "ar", "fewshot": {"train_split": "ar"}},
     }
 
 

diff --git a/...ts/ar/factuality_disinformation_harmful_content/claim_detection/CT22Claim_GPT4_FewShot.py b/...ts/ar/factuality_disinformation_harmful_content/claim_detection/CT22Claim_GPT4_FewShot.py
@@ -13,12 +13,7 @@ def config():
         "model_args": {
             "max_tries": 3,
         },
-        "general_args": {
-            "data_path": "data/factuality_disinformation_harmful_content/claim_covid19/CT22_arabic_1B_claim_test_gold.tsv",
-            "fewshot": {
-                "train_data_path": "data/factuality_disinformation_harmful_content/claim_covid19/CT22_arabic_1B_claim_train.tsv"
-            },
-        },
+        "general_args": {"test_split": "ar", "fewshot": {"train_split": "ar"}},
     }
 
 

diff --git a/assets/ar/factuality_disinformation_harmful_content/factuality/ANSFactuality_GPT4_FewShot.py b/assets/ar/factuality_disinformation_harmful_content/factuality/ANSFactuality_GPT4_FewShot.py
@@ -13,12 +13,6 @@ def config():
         "model_args": {
             "max_tries": 3,
         },
-        "general_args": {
-            "data_path": "data/factuality_disinformation_harmful_content/factuality_stance_khouja/claim/test.csv",
-            "fewshot": {
-                "train_data_path": "data/factuality_disinformation_harmful_content/factuality_stance_khouja/claim/train.csv"
-            },
-        },
     }
 
 

diff --git a/...ar/factuality_disinformation_harmful_content/factuality/COVID19Factuality_GPT4_FewShot.py b/...ar/factuality_disinformation_harmful_content/factuality/COVID19Factuality_GPT4_FewShot.py
@@ -14,12 +14,6 @@ def config():
             "class_labels": ["yes", "no"],
             "max_tries": 30,
         },
-        "general_args": {
-            "data_path": "data/factuality_disinformation_harmful_content/factuality_covid19/covid19_infodemic_arabic_data_factuality_binary_test.tsv",
-            "fewshot": {
-                "train_data_path": "data/factuality_disinformation_harmful_content/factuality_covid19/covid19_infodemic_arabic_data_factuality_binary_train.tsv",
-            },
-        },
     }
 
 

diff --git a/.../factuality_disinformation_harmful_content/factuality/UnifiedFCFactuality_GPT4_FewShot.py b/.../factuality_disinformation_harmful_content/factuality/UnifiedFCFactuality_GPT4_FewShot.py
@@ -14,9 +14,7 @@ def config():
             "max_tries": 3,
         },
         "general_args": {
-            "data_path": "data/factuality_disinformation_harmful_content/factuality_stance_ramy/ramy_arabic_fact_checking.tsv",
             "fewshot": {
-                "train_data_path": "data/factuality_disinformation_harmful_content/factuality_stance_khouja/claim/train.csv",
                 "deduplicate": False,  # N-fold evaluation
             },
         },

diff --git a/...ctuality_disinformation_harmful_content/harmfulness_detection/CT22Harmful_GPT4_FewShot.py b/...ctuality_disinformation_harmful_content/harmfulness_detection/CT22Harmful_GPT4_FewShot.py
@@ -13,12 +13,7 @@ def config():
         "model_args": {
             "max_tries": 3,
         },
-        "general_args": {
-            "data_path": "data/factuality_disinformation_harmful_content/harmful/CT22_arabic_1C_harmful_test_gold.tsv",
-            "fewshot": {
-                "train_data_path": "data/factuality_disinformation_harmful_content/harmful/CT22_arabic_1C_harmful_train.tsv"
-            },
-        },
+        "general_args": {"test_split": "ar", "fewshot": {"train_split": "ar"}},
     }
 
 

diff --git a/...s/ar/factuality_disinformation_harmful_content/hate_speech/OSACT4SubtaskB_GPT4_FewShot.py b/...s/ar/factuality_disinformation_harmful_content/hate_speech/OSACT4SubtaskB_GPT4_FewShot.py
@@ -13,12 +13,6 @@ def config():
         "model_args": {
             "max_tries": 3,
         },
-        "general_args": {
-            "data_path": "data/factuality_disinformation_harmful_content/hate_speech/OSACT2020-sharedTask-test-tweets-labels.txt",
-            "fewshot": {
-                "train_data_path": "data/factuality_disinformation_harmful_content/hate_speech/OSACT2020-sharedTask-train_HS.txt",  # TO_DO
-            },
-        },
     }
 
 

diff --git a/...ctuality_disinformation_harmful_content/offensive_language/OSACT4SubtaskA_GPT4_FewShot.py b/...ctuality_disinformation_harmful_content/offensive_language/OSACT4SubtaskA_GPT4_FewShot.py
@@ -14,12 +14,6 @@ def config():
             "class_labels": ["OFF", "NOT_OFF"],
             "max_tries": 3,
         },
-        "general_args": {
-            "data_path": "data/factuality_disinformation_harmful_content/offensive_language/OSACT2020-sharedTask-test-tweets-labels.txt",
-            "fewshot": {
-                "train_data_path": "data/factuality_disinformation_harmful_content/offensive_language/OSACT2020-sharedTask-train_OFF.txt",  # TO_DO
-            },
-        },
     }
 
 

diff --git a/...ar/factuality_disinformation_harmful_content/propaganda/WANLP22Propaganda_GPT4_FewShot.py b/...ar/factuality_disinformation_harmful_content/propaganda/WANLP22Propaganda_GPT4_FewShot.py
@@ -38,12 +38,6 @@ def config():
             ],
             "max_tries": 30,
         },
-        "general_args": {
-            "data_path": "data/factuality_disinformation_harmful_content/propaganda/task1_test_gold_label_final.json",
-            "fewshot": {
-                "train_data_path": "data/factuality_disinformation_harmful_content/propaganda/task1_train.json",
-            },
-        },
     }
 
 

diff --git a/...r/factuality_disinformation_harmful_content/subjectivity/CT23Subjectivity_GPT4_FewShot.py b/...r/factuality_disinformation_harmful_content/subjectivity/CT23Subjectivity_GPT4_FewShot.py
@@ -14,12 +14,7 @@ def config():
             "class_labels": ["SUBJ", "OBJ"],
             "max_tries": 30,
         },
-        "general_args": {
-            "data_path": "data/factuality_disinformation_harmful_content/subjectivity/dev_ar.tsv",
-            "fewshot": {
-                "train_data_path": "data/factuality_disinformation_harmful_content/subjectivity/train_ar.tsv"
-            },
-        },
+        "general_args": {"test_split": "ar/dev", "fewshot": {"train_split": "ar"}},
     }
 
 

diff --git a/assets/ar/news_categorization/ASND_GPT4_FewShot.py b/assets/ar/news_categorization/ASND_GPT4_FewShot.py
@@ -27,12 +27,6 @@ def config():
             ],
             "max_tries": 30,
         },
-        "general_args": {
-            "data_path": "data/news_categorization/Arabic_Social_Media_News_Dataset_ASND/sm_news_ar_tst.csv",
-            "fewshot": {
-                "train_data_path": "data/news_categorization/Arabic_Social_Media_News_Dataset_ASND/sm_news_ar_trn.csv"
-            },
-        },
     }
 
 

diff --git a/assets/ar/news_categorization/SANADAkhbarona_GPT4_FewShot.py b/assets/ar/news_categorization/SANADAkhbarona_GPT4_FewShot.py
@@ -27,12 +27,6 @@ def config():
             ],
             "max_tries": 30,
         },
-        "general_args": {
-            "data_path": "data/news_categorization/SANAD_akhbarona_news_cat_test.tsv",
-            "fewshot": {
-                "train_data_path": "data/news_categorization/SANAD_akhbarona_news_cat_train.tsv"
-            },
-        },
     }
 
 

diff --git a/assets/ar/news_categorization/SANADAlArabiya_GPT4_FewShot.py b/assets/ar/news_categorization/SANADAlArabiya_GPT4_FewShot.py
@@ -27,12 +27,6 @@ def config():
             ],
             "max_tries": 30,
         },
-        "general_args": {
-            "data_path": "data/news_categorization/SANAD_alarabiya_news_cat_test.tsv",
-            "fewshot": {
-                "train_data_path": "data/news_categorization/SANAD_alarabiya_news_cat_train.tsv"
-            },
-        },
     }
 
 

diff --git a/assets/ar/news_categorization/SANADAlKhaleej_GPT4_FewShot.py b/assets/ar/news_categorization/SANADAlKhaleej_GPT4_FewShot.py
@@ -27,12 +27,6 @@ def config():
             ],
             "max_tries": 30,
         },
-        "general_args": {
-            "data_path": "data/news_categorization/SANAD_alkhaleej_news_cat_test.tsv",
-            "fewshot": {
-                "train_data_path": "data/news_categorization/SANAD_alkhaleej_news_cat_train.tsv"
-            },
-        },
     }
 
 

diff --git a/assets/ar/semantics/NLI/XNLI_GPT4_FewShot.py b/assets/ar/semantics/NLI/XNLI_GPT4_FewShot.py
@@ -13,10 +13,7 @@ def config():
         "model_args": {
             "max_tries": 3,
         },
-        "general_args": {
-            "data_path": "data/XNLI/xnli.test.ar.tsv",
-            "fewshot": {"train_data_path": "data/XNLI/xnli.dev.tsv"},
-        },
+        "general_args": {"fewshot": {"train_split": "dev"}},
     }
 
 

diff --git a/assets/ar/semantics/STS/Q2QSim_GPT4_FewShot.py b/assets/ar/semantics/STS/Q2QSim_GPT4_FewShot.py
@@ -15,12 +15,6 @@ def config():
         "model_args": {
             "max_tries": 3,
         },
-        "general_args": {
-            "data_path": "data/STS/nsurl-2019-task8/test.tsv",
-            "fewshot": {
-                "train_data_path": "data/STS/nsurl-2019-task8/train.tsv",
-            },
-        },
     }
 
 

diff --git a/assets/ar/semantics/STS/SemEval17T1STS_GPT4_FewShot.py b/assets/ar/semantics/STS/SemEval17T1STS_GPT4_FewShot.py
@@ -13,15 +13,6 @@ def config():
         "model_args": {
             "max_tries": 3,
         },
-        "general_args": {
-            "data_path": {
-                "sentences_path": "data/STS/semeval-2017/STS2017.eval.v1.1/STS.input.track1.ar-ar.txt",
-                "gt_data_path": "data/STS/semeval-2017/STS2017.gs/STS.gs.track1.ar-ar.txt",
-            },
-            "fewshot": {
-                "train_data_path": "data/STS/semeval-2017/ar_sts_data_updated/Ar_STS/ar.STS.All.txt",
-            },
-        },
     }
 
 

diff --git a/assets/ar/semantics/STS/SemEval17T2STS_GPT4_FewShot.py b/assets/ar/semantics/STS/SemEval17T2STS_GPT4_FewShot.py
@@ -13,15 +13,6 @@ def config():
         "model_args": {
             "max_tries": 3,
         },
-        "general_args": {
-            "data_path": {
-                "sentences_path": "data/STS/semeval-2017/STS2017.eval.v1.1/STS.input.track2.ar-en.txt",
-                "gt_data_path": "data/STS/semeval-2017/STS2017.gs/STS.gs.track2.ar-en.txt",
-            },
-            "fewshot": {
-                "train_data_path": "data/STS/semeval-2017/ar_sts_data_updated/En_Ar_STS/en_ar.STS.All.txt",
-            },
-        },
     }
 
 

diff --git a/assets/ar/sentiment_emotion_others/emotion/Emotion_GPT4_FewShot.py b/assets/ar/sentiment_emotion_others/emotion/Emotion_GPT4_FewShot.py
@@ -25,12 +25,6 @@ def config():
             ],
             "max_tries": 30,
         },
-        "general_args": {
-            "data_path": "data/sentiment_emotion_others/emotion/test-gold.txt",
-            "fewshot": {
-                "train_data_path": "data/sentiment_emotion_others/emotion/train.txt",
-            },
-        },
     }
 
 

diff --git a/assets/ar/sentiment_emotion_others/sarcasm/ArSarcasm2_GPT4_FewShot.py b/assets/ar/sentiment_emotion_others/sarcasm/ArSarcasm2_GPT4_FewShot.py
@@ -14,12 +14,6 @@ def config():
             "class_labels": ["TRUE", "FALSE"],
             "max_tries": 3,
         },
-        "general_args": {
-            "data_path": "data/sentiment_emotion_others/sarcasm/ArSarcasm2/testing_data.csv",
-            "fewshot": {
-                "train_data_path": "data/sentiment_emotion_others/sarcasm/ArSarcasm2/training_data.csv",
-            },
-        },
     }
 
 

diff --git a/assets/ar/sentiment_emotion_others/sarcasm/ArSarcasm_GPT4_FewShot.py b/assets/ar/sentiment_emotion_others/sarcasm/ArSarcasm_GPT4_FewShot.py
@@ -14,12 +14,6 @@ def config():
             "class_labels": ["TRUE", "FALSE"],
             "max_tries": 3,
         },
-        "general_args": {
-            "data_path": "data/sentiment_emotion_others/sarcasm/ArSarcasm/ArSarcasm_test.csv",
-            "fewshot": {
-                "train_data_path": "data/sentiment_emotion_others/sarcasm/ArSarcasm/ArSarcasm_train.csv",
-            },
-        },
     }
 
 

diff --git a/assets/ar/sentiment_emotion_others/sentiment/ArSAS_GPT4_FewShot.py b/assets/ar/sentiment_emotion_others/sentiment/ArSAS_GPT4_FewShot.py
@@ -16,12 +16,6 @@ def config():
             "class_labels": ["Positive", "Negative", "Neutral", "Mixed"],
             "max_tries": 3,
         },
-        "general_args": {
-            "data_path": "data/sentiment_emotion_others/sentiment/ArSAS-test.txt",
-            "fewshot": {
-                "train_data_path": "data/sentiment_emotion_others/sentiment/ArSAS-train.txt",
-            },
-        },
     }