Update README.md

README.md

@@ -1,3 +1,31 @@
----
-license: mit
----
+---
+license: mit
+language:
+- en
+base_model:
+- FacebookAI/roberta-base
+---
+# Ambiguity-aware RoBERTa
+
+This model is trained on SemEval2007 Task 14 Affective Text dataset and is capable of representing the ambiguity occurring in emotion analysis tasks as an accurate distribution (i.e., softmax output). It was introduced in the following paper: ["Deep Model Compression Also Helps Models Capture Ambiguity"](https://aclanthology.org/2023.acl-long.381.pdf) (ACL 2023).
+
+# Usage
+
+```python
+from transformers import RobertaTokenizer, RobertaForSequenceClassification
+tokenizer = RobertaTokenizer.from_pretrained('hancheolp/ambiguity-aware-roberta-emotion')
+model = RobertaForSequenceClassification.from_pretrained('hancheolp/ambiguity-aware-roberta-emotion')
+news_headline = "Amateur rocket scientists reach for space."
+encoded_input = tokenizer(news_headline, return_tensors='pt')
+output = model(**encoded_input)
+distribution = output.logits.softmax(dim=-1)
+print(distribution)
+```
+
+Each index of the output vector represents the following:
+* 0: anger
+* 1: disgust
+* 2: fear
+* 3: joy
+* 4: sadness
+* 5: surprise