artificialguybr/llama3-8b-sql-create-context

TEXT GENERATIONConcurrency Cost:1Model Size:8BQuant:FP8Ctx Length:8kPublished:Apr 23, 2024Architecture:Transformer0.0K Cold

The artificialguybr/llama3-8b-sql-create-context model is an 8 billion parameter Llama 3 based language model, fine-tuned by artificialguybr. It is specifically optimized for generating SQL queries from natural language questions, using a provided SQL CREATE TABLE statement as context. This model excels at preventing hallucination of column and table names by relying on explicit schema context, making it highly effective for text-to-SQL tasks where the database schema is known.

Loading preview...

LLAMA 3 8B SQL CREATE CONTEXT: Text-to-SQL Model

This model, developed by artificialguybr, is a fine-tuned version of the NousResearch/Meta-Llama-3-8B base model. It specializes in converting natural language questions into SQL queries, leveraging a provided SQL CREATE TABLE statement for context. This approach is designed to significantly reduce the hallucination of column and table names, a common challenge in text-to-SQL systems.

Key Capabilities

  • Text-to-SQL Generation: Translates natural language questions into accurate SQL queries.
  • Schema-Aware Querying: Utilizes CREATE TABLE statements to understand database schema, ensuring contextually relevant and precise SQL output.
  • Reduced Hallucination: Trained on a combined WikiSQL and Spider dataset (78,577 examples) specifically to prevent incorrect column and table name generation.
  • Adaptable: Can be fine-tuned for specific use cases and various SQL dialects.

Intended Uses

  • Generating SQL queries from natural language inputs when the table schema is known.
  • Supporting data analysis and visualization tasks.
  • Integration into data analysis pipelines.

Limitations

  • Performance is dependent on the accuracy and completeness of the provided CREATE TABLE statement.
  • May not generalize well to entirely new SQL dialects or highly complex, multi-join queries.
  • Not designed for queries requiring external knowledge or common sense reasoning beyond the provided schema.