Upload app.py

app.py

@@ -43,6 +43,16 @@ def get_model():
 with st.spinner('Loading the model...'):
     model, tokenizer = get_model()
 
+st.header('RoBERTa Q&A model')
+
+st.markdown('''
+This app demonstrates the answer-retrieval capabilities of a finetuned RoBERTa (Robustly optimized Bidirectional Encoder Representations from Transformers) model.  The [RoBERTa base model](https://huggingface.co/roberta-base) was fine-tuned on version 2 of the [SQuAD (Stanford Question Answering Dataset) dataset](https://huggingface.co/datasets/squad_v2), a dataset of context-question-answer triples.  The objective of the model is to retrieve the answer to the question from the context paragraph.
+
+Version 2 incorporates the 100,000 samples from Version 1.1, along with 50,000 'unanswerable' questions, i.e. samples in the question cannot be answered using the context given.
+
+Please type or paste a context paragraph and question you'd like to ask about it.  The model will attempt to answer the question, or otherwise will report that it cannot.
+''')
+
 input_container = st.container()
 st.divider()
 response_container = st.container()