AlioLeuchtmann/ALIO-SQL-7B

TEXT GENERATIONPricing:Input $0.4 / Cached $0.08 / Output $0.8Concurrent Unit Cost:1Model Size:7.6BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Mar 30, 2025License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

AlioLeuchtmann/ALIO-SQL-7B is a 7.6 billion parameter language model fine-tuned from Qwen2.5-Coder-7B, specifically optimized for text-to-SQL generation in SQLite dialect. It achieves 57% on the BIRD private test set and 60% on the BIRD dev greedy set, outperforming larger universal models like GPT-4o and Gemini 1.5 Pro in its specialized domain. This model excels at converting natural language questions into SQL queries, leveraging a context length of 32768 tokens.

Loading preview...

ALIO-SQL-7B: Specialized Text-to-SQL Model

ALIO-SQL-7B is a 7.6 billion parameter model, fine-tuned from Qwen2.5-Coder-7B, designed exclusively for generating SQLite SQL queries from natural language. This model demonstrates strong performance on text-to-SQL benchmarks, particularly within the BIRD dataset.

Key Capabilities

  • High Accuracy on SQL Generation: Achieves 57% on the BIRD private test set and 60% on the BIRD dev greedy set, surpassing its teacher model, Qwen2.5-Coder-32B, and larger general-purpose models like GPT-4o and Gemini 1.5 Pro in this specific task.
  • Domain-Specific Optimization: Trained on a curated dataset combining BIRD and Spider, utilizing knowledge distillation with CoT prompting from Qwen2.5-Coder-32B, ensuring high-quality and correct SQL outputs.
  • SQLite Dialect Focus: Exclusively trained for and restricted to generating SQL in the SQLite dialect.
  • Contextual Understanding: Utilizes a prompt structure that incorporates schema, external knowledge, and natural language questions to generate precise SQL.

Good For

  • Applications Requiring SQLite Query Generation: Ideal for systems that need to translate user questions into executable SQLite commands.
  • Benchmarking Text-to-SQL Performance: Provides a strong baseline for specialized text-to-SQL tasks, particularly against the BIRD and Spider datasets.
  • Constrained Domain SQL Tasks: Excels in scenarios where the focus is solely on generating SQL for database interactions, without requiring other general language model capabilities.

Restrictions

  • SQLite Dialect Only: The model is trained exclusively on SQLite and will not perform well with other SQL dialects.
  • English Language Only: Optimized for English language input for natural language questions.
  • Specialized Skill Set: This model's training focused solely on text-to-SQL, and it does not retain general conversational or other broad language model skills.

Finetunes
1 models