FILE 02 · AI / NLP2024.04 — 2024.08RESEARCH PROJECT
ADAPTIVE RESPONSES CHATBOTCHATBOT SERVER
A language-learning chatbot that grades before it answers. A BERT model first scores the learner's Chinese fluency; that score drives a dynamic prompt builder, so GPT-4's reply difficulty and style adapt to the learner's actual level instead of answering everyone the same way.
- 01 → Learner writes a Chinese sentence
- 02 → BERT evaluates fluency & level
- 03 → Dynamic prompt assembled per score (Django REST)
- 04 → GPT-4 replies at the learner's level
PythonDjango REST FrameworkBERTGPT-4 / LLM
DEMO VIDEO / CODE — AVAILABLE ON REQUEST