Fuzzify addresses challenges faced by police officers and others managing large public databases, where variations in names create inconsistencies. For example, "Laxmi" can be written as Lakshmi, Lakxmy, or Lackshmy, making database searches difficult.
Fuzzify leverages a fine-tuned Llama 3.2 1B model, optimized for lightweight performance, to predict all possible pronunciations of a given name in English Latin script. It outputs these variations in IPA (International Phonetic Alphabet) notation, effectively capturing the full range of pronunciations.
- Name Input: User enters a name to search
- IPA Generation: Fine-tuned Llama 3.2 1B model predicts all possible pronunciations in IPA notation
- Vector Embedding: Custom embedder converts IPA representations into vectors
- Vector Storage: Vectors stored in Chroma DB vector database
- Similarity Search: Cosine similarity algorithm matches and retrieves relevant results
- Model: Fine-tuned Llama 3.2 1B on Unsloth
- Vector Database: Chroma DB
- Frontend: Flutter mobile app
- Backend: FastAPI server
- Search Algorithm: Cosine similarity
Developed for SIH Grand Finale 2024, Fuzzify enables efficient name matching across large public databases, reducing search time and improving accuracy for law enforcement and database administrators. 