Update README.md

README.md

@@ -44,7 +44,7 @@ Example:
 ```python
 from transformers import AutoTokenizer, AutoModelForSequenceClassification, pipeline
 
-model_id = "YOUR_ORG/YOUR_MODEL_NAME"
+model_id = "ChangeIsKey/denotational-roberta-classifier"
 
 classifier = pipeline(
     "text-classification",
@@ -71,9 +71,8 @@ The model was fine-tuned on a combined denotational-relation dataset referred to
 
 * ChainNet for metaphor, metonymy, and homonymy relations;
 * UniMet for additional metonymy examples;
-* WordNet-based hypernym/hyponym pairs for generalization and specialization;
-* auto-antonymy examples where available in the fine-tuning data.
-
+* WordNet for generalization, specialization, and auto-antonymy examples.
+  
 These resources use definition or synset-pair information as proxies for relations between word senses.
 
 ## Training Procedure
@@ -104,23 +103,6 @@ The model was evaluated on:
 
 On the SenseRel denotational dataset, this model achieved the best reported score among the evaluated systems.
 
-## Limitations
-
-The model should be used with the following limitations in mind:
-
-* It operates on definition pairs, not full diachronic corpus evidence.
-* Training data uses lexical-resource relations as proxies for sense-level semantic change.
-* The distinction between `generalization` and `specialization` is direction-sensitive and depends on the order of the input senses.
-* Taxonomical relations such as `generalization` and `specialization` may be confused with one another.
-* Figurative relations such as `metaphor` and `metonymy` can be difficult to separate, especially when senses are abstract or weakly related.
-* `homonymy` is challenging because the model must distinguish true unrelatedness from distant semantic extension.
-* The model was developed and evaluated for English noun sense definitions; performance may degrade on other parts of speech, languages, domains, or informal text.
-
-## Bias, Risks, and Ethical Considerations
-
-This model reflects the structure and assumptions of the lexical resources and annotation schemes used for training and evaluation. It may encode biases present in dictionary definitions, lexical databases, and historical semantic descriptions. Its predictions should be treated as linguistic classifications, not as definitive etymological or historical claims.
-
-For scholarly use, model outputs should be validated by domain experts, especially when used to support claims about semantic change.
 
 ## Citation