---
license: mit
datasets:
- sms1097/self_rag_tokens_train_data
---
# Relevant Model

This generates the `is_relevant` token as descirbed in Self-RAG. 

We are testing to see if a retrieved document is relevant to the user input of our language model. 

The expected input to the model is:

```
Instruction:\n{instruction}\nContext:\n{doc}
```