mteb 2.0.4__py3-none-any.whl → 2.1.0__py3-none-any.whl

mteb/descriptive_stats/Image/ZeroShotClassification/PatchCamelyonZeroShot.json

@@ -1,23 +1,37 @@
 {
     "test": {
         "num_samples": 32768,
-        "unique_num_labels": 2,
-        "min_image_width": 96,
-        "average_image_width": 96.0,
-        "max_image_width": 96,
-        "min_image_height": 96,
-        "average_image_height": 96.0,
-        "max_image_height": 96,
-        "min_label_text_length": 35,
-        "average_label_text_length": 52.0,
-        "max_label_text_length": 69,
-        "labels": {
-            "0": {
-                "count": 16391
-            },
-            "1": {
-                "count": 16377
+        "number_of_characters": null,
+        "text_statistics": null,
+        "image_statistics": {
+            "min_image_width": 96,
+            "average_image_width": 96.0,
+            "max_image_width": 96,
+            "min_image_height": 96,
+            "average_image_height": 96.0,
+            "max_image_height": 96,
+            "unique_images": 29383
+        },
+        "label_statistics": {
+            "min_labels_per_text": 1,
+            "average_label_per_text": 1.0,
+            "max_labels_per_text": 1,
+            "unique_labels": 2,
+            "labels": {
+                "0": {
+                    "count": 16391
+                },
+                "1": {
+                    "count": 16377
+                }
             }
+        },
+        "candidates_labels_text_statistics": {
+            "total_text_length": 103,
+            "min_text_length": 34,
+            "average_text_length": 51.5,
+            "max_text_length": 69,
+            "unique_texts": 2
         }
     }
 }

mteb/descriptive_stats/MultilabelClassification/CovidDisinformationNLMultiLabelClassification.json

@@ -0,0 +1,84 @@
+{
+    "test": {
+        "num_samples": 252,
+        "number_texts_intersect_with_train": null,
+        "text_statistics": {
+            "total_text_length": 48963,
+            "min_text_length": 35,
+            "average_text_length": 194.29761904761904,
+            "max_text_length": 337,
+            "unique_texts": 252
+        },
+        "image_statistics": null,
+        "label_statistics": {
+            "min_labels_per_text": 0,
+            "average_label_per_text": 2.115079365079365,
+            "max_labels_per_text": 6,
+            "unique_labels": 7,
+            "labels": {
+                "1": {
+                    "count": 183
+                },
+                "None": {
+                    "count": 50
+                },
+                "3": {
+                    "count": 102
+                },
+                "5": {
+                    "count": 83
+                },
+                "0": {
+                    "count": 56
+                },
+                "2": {
+                    "count": 62
+                },
+                "4": {
+                    "count": 47
+                }
+            }
+        }
+    },
+    "train": {
+        "num_samples": 929,
+        "number_texts_intersect_with_train": null,
+        "text_statistics": {
+            "total_text_length": 174673,
+            "min_text_length": 33,
+            "average_text_length": 188.02260495156082,
+            "max_text_length": 714,
+            "unique_texts": 929
+        },
+        "image_statistics": null,
+        "label_statistics": {
+            "min_labels_per_text": 0,
+            "average_label_per_text": 1.9730893433799785,
+            "max_labels_per_text": 6,
+            "unique_labels": 7,
+            "labels": {
+                "1": {
+                    "count": 709
+                },
+                "3": {
+                    "count": 377
+                },
+                "0": {
+                    "count": 164
+                },
+                "2": {
+                    "count": 183
+                },
+                "4": {
+                    "count": 138
+                },
+                "5": {
+                    "count": 262
+                },
+                "None": {
+                    "count": 168
+                }
+            }
+        }
+    }
+}

mteb/descriptive_stats/MultilabelClassification/VABBMultiLabelClassification.json

@@ -0,0 +1,156 @@
+{
+    "test": {
+        "num_samples": 3235,
+        "number_texts_intersect_with_train": 2265,
+        "text_statistics": {
+            "total_text_length": 2927178,
+            "min_text_length": 47,
+            "average_text_length": 904.8463678516229,
+            "max_text_length": 35167,
+            "unique_texts": 3235
+        },
+        "image_statistics": null,
+        "label_statistics": {
+            "min_labels_per_text": 1,
+            "average_label_per_text": 1.8593508500772797,
+            "max_labels_per_text": 8,
+            "unique_labels": 19,
+            "labels": {
+                "Social Sciences (General)": {
+                    "count": 472
+                },
+                "Sociology": {
+                    "count": 366
+                },
+                "Other Disciplines": {
+                    "count": 557
+                },
+                "Educational Sciences": {
+                    "count": 302
+                },
+                "Economics & Business": {
+                    "count": 393
+                },
+                "Social Health Sciences": {
+                    "count": 255
+                },
+                "Political Sciences": {
+                    "count": 201
+                },
+                "Communication Studies": {
+                    "count": 155
+                },
+                "History": {
+                    "count": 287
+                },
+                "Humanities (General)": {
+                    "count": 398
+                },
+                "Law": {
+                    "count": 999
+                },
+                "Philosophy": {
+                    "count": 233
+                },
+                "Art History": {
+                    "count": 349
+                },
+                "Theology": {
+                    "count": 58
+                },
+                "Psychology": {
+                    "count": 174
+                },
+                "Criminology": {
+                    "count": 284
+                },
+                "Linguistics": {
+                    "count": 254
+                },
+                "Literature": {
+                    "count": 140
+                },
+                "Archaeology": {
+                    "count": 138
+                }
+            }
+        }
+    },
+    "train": {
+        "num_samples": 6471,
+        "number_texts_intersect_with_train": null,
+        "text_statistics": {
+            "total_text_length": 5847151,
+            "min_text_length": 38,
+            "average_text_length": 903.5931077113274,
+            "max_text_length": 35167,
+            "unique_texts": 6471
+        },
+        "image_statistics": null,
+        "label_statistics": {
+            "min_labels_per_text": 1,
+            "average_label_per_text": 1.8822438572090867,
+            "max_labels_per_text": 11,
+            "unique_labels": 19,
+            "labels": {
+                "Social Sciences (General)": {
+                    "count": 1021
+                },
+                "Sociology": {
+                    "count": 660
+                },
+                "Other Disciplines": {
+                    "count": 1124
+                },
+                "Educational Sciences": {
+                    "count": 541
+                },
+                "Economics & Business": {
+                    "count": 756
+                },
+                "Linguistics": {
+                    "count": 501
+                },
+                "Humanities (General)": {
+                    "count": 995
+                },
+                "Political Sciences": {
+                    "count": 380
+                },
+                "Communication Studies": {
+                    "count": 327
+                },
+                "Psychology": {
+                    "count": 375
+                },
+                "Philosophy": {
+                    "count": 434
+                },
+                "Art History": {
+                    "count": 703
+                },
+                "History": {
+                    "count": 522
+                },
+                "Law": {
+                    "count": 2150
+                },
+                "Theology": {
+                    "count": 116
+                },
+                "Social Health Sciences": {
+                    "count": 489
+                },
+                "Criminology": {
+                    "count": 576
+                },
+                "Literature": {
+                    "count": 275
+                },
+                "Archaeology": {
+                    "count": 235
+                }
+            }
+        }
+    }
+}

mteb/descriptive_stats/PairClassification/SICKNLPairClassification.json

@@ -0,0 +1,35 @@
+{
+    "test": {
+        "num_samples": 2115,
+        "number_of_characters": 194569,
+        "unique_pairs": 2115,
+        "text1_statistics": {
+            "total_text_length": 100505,
+            "min_text_length": 10,
+            "average_text_length": 47.520094562647756,
+            "max_text_length": 158,
+            "unique_texts": 1769
+        },
+        "text2_statistics": {
+            "total_text_length": 94064,
+            "min_text_length": 10,
+            "average_text_length": 44.47470449172577,
+            "max_text_length": 158,
+            "unique_texts": 1738
+        },
+        "labels_statistics": {
+            "min_labels_per_text": 1,
+            "average_label_per_text": 1.0,
+            "max_labels_per_text": 1,
+            "unique_labels": 2,
+            "labels": {
+                "1": {
+                    "count": 1404
+                },
+                "0": {
+                    "count": 711
+                }
+            }
+        }
+    }
+}

mteb/descriptive_stats/PairClassification/XLWICNLPairClassification.json

@@ -0,0 +1,35 @@
+{
+    "test": {
+        "num_samples": 1004,
+        "number_of_characters": 103768,
+        "unique_pairs": 1004,
+        "text1_statistics": {
+            "total_text_length": 52390,
+            "min_text_length": 15,
+            "average_text_length": 52.18127490039841,
+            "max_text_length": 139,
+            "unique_texts": 851
+        },
+        "text2_statistics": {
+            "total_text_length": 51378,
+            "min_text_length": 9,
+            "average_text_length": 51.17330677290837,
+            "max_text_length": 131,
+            "unique_texts": 863
+        },
+        "labels_statistics": {
+            "min_labels_per_text": 1,
+            "average_label_per_text": 1.0,
+            "max_labels_per_text": 1,
+            "unique_labels": 2,
+            "labels": {
+                "1": {
+                    "count": 502
+                },
+                "0": {
+                    "count": 502
+                }
+            }
+        }
+    }
+}