jastorj/couchmind-v5.8.2_cold_start-cw-26K-16bit

TEXT GENERATIONPricing:Input $0.4 / Cached $0.08 / Output $0.8Concurrent Unit Cost:1Model Size:7.6BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Aug 19, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

The jastorj/couchmind-v5.8.2_cold_start-cw-26K-16bit model is a 7.6 billion parameter instruction-tuned language model, fine-tuned from Snowflake/Arctic-Text2SQL-R1-7B. It specializes in Text-to-SQL generation, specifically for Couchbase SQL++ queries, using the NL2SQL++ v5.8.2_cold_start dataset with code-with-thought reasoning. This model excels at converting natural language questions into syntactically valid SQL++ queries based on provided database schemas.

Loading preview...

Model Overview

This model, jastorj/couchmind-v5.8.2_cold_start-cw-26K-16bit, is a specialized 7.6 billion parameter language model derived from Snowflake/Arctic-Text2SQL-R1-7B. It has been supervised fine-tuned (SFT) using LoRA (Low-Rank Adaptation) with Unsloth, specifically for the task of Text-to-SQL generation.

Key Capabilities

  • Text-to-SQL Generation: Translates natural language questions into precise Couchbase SQL++ queries.
  • Schema-Aware Querying: Utilizes provided database schemas, including bucket, scope, and collection names, to generate syntactically valid and contextually accurate SQL++.
  • Code-with-Thought Reasoning: Trained on a dataset that incorporates an internal monologue reasoning process, allowing it to explain its query generation steps.
  • 16-bit Quantization: Features 16-bit merged weights for efficient deployment and inference.

Training Details

The model was fine-tuned on the NL2SQL++ v5.8.2_cold_start dataset, which includes 1148 examples. A notable aspect of the training data is the inclusion of 'code-with-thought' reasoning, where the model is exposed to the logical steps taken to arrive at a SQL query from a natural language question. This enhances its ability to generate well-reasoned and accurate SQL++ outputs.

Good For

  • Developers and data professionals working with Couchbase databases who need to quickly generate SQL++ queries from natural language.
  • Applications requiring automated SQL query generation based on user input and defined database schemas.
  • Use cases where understanding the reasoning behind the generated SQL query is beneficial.

Finetunes
1 models