updated docu

tacred.py

@@ -15,7 +15,6 @@
 
 """The TACRED dataset for English Relation Classification"""
 
-
 import json
 import os
 
@@ -67,6 +66,12 @@ TACRED is a large-scale relation extraction dataset with 106,264 examples built
 Note: There is currently a label-corrected version of the TACRED dataset, which you should consider using instead of
 the original version released in 2017. For more details on this new version, see the TACRED Revisited paper
 published at ACL 2020.
+
+Note 2: This Datasetreader changes the offsets of the following fields, to conform with standard Python usage (see
+#_generate_examples()):
+- subj_end to subj_end + 1 (make end offset exclusive)
+- oj_end to oj_end + 1 (make end offset exclusive)
+- stanford_head to stanford_head - 1 (make head offsets 0-based)
 """
 
 _HOMEPAGE = "https://nlp.stanford.edu/projects/tacred/"
@@ -82,6 +87,8 @@ _PATCH_URLs = {
     "test": "https://raw.githubusercontent.com/DFKI-NLP/tacrev/master/patch/test_patch.json",
 }
 
+_VERSION = datasets.Version("1.0.0")
+
 _CLASS_LABELS = [
     "no_relation",
     "org:alternate_names",
@@ -146,11 +153,11 @@ class Tacred(datasets.GeneratorBasedBuilder):
 
     BUILDER_CONFIGS = [
         datasets.BuilderConfig(
-            name="original", version=datasets.Version("1.0.0"), description="The original TACRED."
+            name="original", version=_VERSION, description="The original TACRED."
         ),
         datasets.BuilderConfig(
             name="revised",
-            version=datasets.Version("1.0.0"),
+            version=_VERSION,
             description="The revised TACRED (corrected labels in dev and test split).",
         ),
     ]