Added Streamlit autocomplete UI for bigram name generation

[amit03wit]

Summary
This PR adds a lightweight web UI (bigram_autocomplete.py) that lets users interact with the bigram model in real time. As the user types, the app shows the top 5 most likely next characters based on the bigram transition probabilities learned from names.txt.

Motivation
The notebook currently generates full names from the bigram model, but there is no interactive way to explore the learned character-by-character transitions. This UI makes the model more intuitive to understand and demo.

Changes

1. Added bigram_autocomplete.py in the project root
2. Loads names.txt, builds the same 27 × 27 bigram count matrix, and converts it to probabilities
3. Provides a text input that uses the last typed character as context
4. Displays the top 5 predicted next characters with their probabilities

How to run

pip install streamlit torch
streamlit run bigram_autocomplete.py

UI