Update README.md

README.md

@@ -10,4 +10,37 @@ widget:
   example_title: "Example 1"
 - text: "Incongruent Headlines: Yet Another Way to Mislead Your Readers </s> Emotion Cause Extraction - A Review of Various Methods and Corpora"
   example_title: "Example 2"
----
+---
+
+# Bibtex classification using RoBERTa
+
+## Model Description
+This model is a text classification tool designed to predict the likelihood of a given context paper being cited by a query paper. It processes concatenated titles of context and query papers and outputs a binary prediction: `1` indicates a potential citation relationship (though not necessary), and `0` suggests no such relationship.
+
+### Intended Use
+- **Primary Use**: To extract a subset of bibtex from ACL Anthology to make it < 50 MB.
+
+### Model Training
+- **Data Description**: The model was trained on a ACL Anthology dataset [cestwc/anthology](https://huggingface.co/datasets/cestwc/anthology) comprising pairs of paper titles. Each pair was annotated to indicate whether the context paper could potentially be cited by the query paper.
+
+### Performance
+- **Metrics**: [Include performance metrics like accuracy, precision, recall, F1-score, etc.]
+
+## How to Use
+```python
+from transformers import AutoModelForSequenceClassification, AutoTokenizer
+
+model_name = "cestwc/roberta-base-bib"
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForSequenceClassification.from_pretrained(model_name)
+
+def predict_citation(context_title, query_title):
+    inputs = tokenizer.encode_plus(f"{context_title} </s> {query_title}", return_tensors="pt")
+    outputs = model(**inputs)
+    prediction = outputs.logits.argmax(-1).item()
+    return "include" if prediction == 1 else "not include"
+
+# Example
+context_title = "Evaluating and Enhancing the Robustness of Neural Network-based Dependency Parsing Models with Adversarial Examples"
+query_title = "Assessing Hidden Risks of LLMs: An Empirical Study on Robustness, Consistency, and Credibility"
+print(predict_citation(context_title, query_title))