Overview

The tomasonjo/text2cypher-demo-16bit is an 8 billion parameter language model, developed by tomasonjo, specifically fine-tuned for generating Cypher queries from natural language questions. It is based on the unsloth/llama-3-8b-Instruct model and leverages Unsloth and Huggingface's TRL library for efficient training, resulting in a 2x speed improvement during the finetuning process.

Key Capabilities

• Natural Language to Cypher Conversion: Translates user questions into executable Cypher queries, given a Neo4j graph schema.
• Optimized Inference: Utilizes Unsloth's FastLanguageModel.for_inference for enhanced inference speed.
• Llama-3 Chat Template: Employs the Llama-3 chat template for structured input and output.

Good For

• Graph Database Interaction: Ideal for applications that require users to query Neo4j graph databases using natural language.
• Developer Tools: Can be integrated into tools that automate Cypher query generation, reducing manual query writing.
• Educational Purposes: Useful for demonstrating how LLMs can interface with structured data query languages like Cypher.