Shorten About section

README.md

@@ -36,28 +36,17 @@ This is the official dataset repository of [WiC is Not WSD: A Study on LLMs and
 ## About
 ![Example of a WiC instance where sense granularity changes the answer](main_figure.png)
 
-Word-in-Context (WiC) remains challenging for language models, despite recent progress on lexical-semantic tasks.
-Given two sentences that both contain the same target word, a model must decide whether the word is used with the
-**same meaning** in both (`True`) or with **different meanings** (`False`). Unlike Word Sense Disambiguation (WSD),
-WiC provides no sense inventory, so the model must also guess *how fine-grained* the intended distinction is.
-In the example above, both uses of *club* refer to an organised group at a coarse level of granularity, giving the
-human label True, but an LLM that separates a sports club from a decision-making body predicts False.
-
-We hypothesise that much of the difficulty of WiC comes from this missing specification of sense granularity rather
-than from comparing two contexts. To study this, the paper evaluates open LLMs on WiC and on traditional WSD under
-comparable settings, and releases the two WiC datasets in this repository:
-
-- **Coarse-grained WiC** (2,798 pairs, 20 target words): built automatically from CoarseWSD-20 by pairing sentences
-  that contain the same target word and comparing their coarse sense labels.
-- **Fine-grained WiC** (1,185 pairs, 924 target words): a newly constructed dataset that follows the construction of
-  the original WiC benchmark but draws on WordNet 3.1 noun and verb entries, with modifications intended to limit
-  memorisation by LLMs. Each instance also carries the WordNet definition of the target word in each sentence.
-
-We find that providing candidate senses, as in traditional WSD, improves WiC performance in all settings, and that
-explicit sense information helps models make more consistent and targeted judgements. Human evaluation further shows
-that many apparent WiC errors reflect label ambiguity or mismatches between model and annotator sense boundaries
-rather than failures of lexical understanding. In particular, LLMs tend to overthink the sense distinction, often
-leading to errors based on overly fine-grained distinctions.
+This repository contains two Word-in-Context (WiC) evaluation datasets. Given two sentences that contain the same
+target word, the task is to decide whether the word has the **same meaning** in both (`True`) or **different
+meanings** (`False`). The two datasets differ in the granularity of the sense distinctions they encode, as
+illustrated above for the word *club*:
+
+- **Coarse-grained WiC**: 2,798 sentence pairs over 20 ambiguous nouns, built from CoarseWSD-20 by pairing
+  sentences with the same target word and comparing their coarse-grained sense labels.
+- **Fine-grained WiC**: 1,185 sentence pairs over 924 target words, newly constructed from WordNet 3.1 noun and verb
+  entries. Each instance also includes the WordNet definition of the target word in each sentence.
+
+Both datasets are evaluation sets and are released as a single `test` split.
 
 ## Dataset