model_19072.pt - trained using fineweb data for 19072 steps
model_19072-nq_00149.pt - fine tuned using sentence-transformers/natural-questions. The fine tunning data followed the format user: <question>\nassistant: <answer>
Code
model.py - The model
load_model.py - A simple example of loading the model from hugging face
example.py - An example that takes a user input and generates a model output
