arviii/nsql-llama-2-7B-bfloat16

TEXT GENERATIONConcurrency Cost:1Model Size:7BQuant:FP8Ctx Length:4kLicense:llama2Architecture:Transformer Open Weights Cold

NSQL-Llama-2-7B-bfloat16 is a 7 billion parameter autoregressive language model developed by arviii, based on Meta's Llama-2 architecture. It is specifically pre-trained on a large dataset of general SQL queries and then fine-tuned on text-to-SQL pairs, making it highly specialized for generating SQL queries from natural language prompts and given table schemas. This model excels at producing SELECT queries and is evaluated on benchmarks like Spider and GeoQuery.

Loading preview...

NSQL-Llama-2-7B: Specialized Text-to-SQL Generation

NSQL-Llama-2-7B is a 7 billion parameter model built upon Meta's Llama-2 architecture, specifically engineered for robust text-to-SQL generation. This model is part of the NSQL family, which focuses on converting natural language prompts and database schemas into accurate SQL queries.

Key Capabilities

  • SQL Generation: Highly optimized for generating SELECT SQL queries from natural language questions and provided table schemas.
  • Specialized Training: Underwent extensive pre-training on 1 million SQL queries from The Stack dataset, followed by fine-tuning on over 20 public text-to-SQL datasets.
  • Llama-2 Foundation: Leverages the strong base capabilities of the Llama-2 7B model, enhanced for domain-specific performance.
  • Evaluation: Performance is rigorously evaluated on standard text-to-SQL benchmarks, including Spider and GeoQuery.

Good for

  • Developers building applications that require converting natural language into SQL queries.
  • Automating database interactions through natural language interfaces.
  • Use cases focused on generating SELECT statements based on provided table structures and user questions.
  • Integrating SQL generation capabilities into chatbots or data analysis tools.

Finetunes
1 models