Local LLMs for SQL Generation: Build Your Own Text-to-SQL

Text-to-SQL is one of the most practical business applications of local LLMs. Connecting a language model to your database schema lets non-technical users query data in plain English — without sending your sensitive business data to OpenAI's servers. Here's how to build it with locally-run models.

Why local matters here: Your database schema reveals your entire business structure. Sending it to a cloud API is a significant data exposure risk. Local LLMs keep everything on your infrastructure. For hardware capable of running SQL-focused models, LLMHardware.io has recommendations starting under $500.

Best Models for SQL Generation

SQLCoder-7B (from Defog) is fine-tuned specifically for SQL and outperforms GPT-4 on Text-to-SQL benchmarks for complex queries. Llama 3.1 8B Instruct with schema-in-context performs reasonably well. Codestral 22B excels at multi-table joins and subqueries. For production use, always fine-tune on your specific schema — even a small LoRA adapter trained on 100 examples dramatically improves accuracy.

Building the Pipeline

The basic pipeline: (1) extract schema as CREATE TABLE statements, (2) inject into system prompt, (3) take user question, (4) generate SQL, (5) execute against database, (6) return results. Use LangChain's SQLDatabaseChain or build it directly with the Ollama Python API. Always sandbox SQL execution — use a read-only database user and validate queries for dangerous keywords (DROP, DELETE, TRUNCATE) before execution.

Improving Accuracy

Schema clarity matters enormously. Add descriptive comments to your CREATE TABLE statements — columns named "c1", "c2" will produce poor SQL regardless of model quality. Include sample data rows in the prompt for ambiguous columns. Use few-shot examples of common queries. Implement a verification step where the model checks its own SQL before execution. Chain-of-thought prompting ("first identify relevant tables, then construct the query") improves accuracy by 15-25% on complex joins.

Comments

Post a Comment

Popular posts from this blog

AMD RX 9070 XT for Local LLMs: ROCm Performance Guide (2026)

The AMD RX 9070 XT launched in early 2026 at $599 with 16GB GDDR6 and 648 GB/s memory bandwidth. For local LLM users who want a mid-range GPU without NVIDIA prices, this changes the calculus significantly. Here is what it actually delivers. RX 9070 XT Specs vs Competition GPU VRAM Bandwidth Price LLM Score RX 9070 XT 16 GB 648 GB/s $599 ~60 tok/s RTX 4070 Ti Super 16 GB 672 GB/s $799 ~62 tok/s RTX 4080 Super 16 GB 736 GB/s $999 ~68 tok/s RTX 4090 24 GB 1,008 GB/s ~$1,100 ~90 tok/s At $599 with 16GB VRAM, the RX 9070 XT is the best value GPU for local LLMs in the mid-range tier in 2026. See the full comparison at LLMHardware.io's AMD GPU guide . ROCm Setup on Linux (2026) # Ubuntu 22.04/24.04 sudo apt install rocm-hip-libraries hip-base # or use the AMD installer: wget https://repo.radeon.com/amdgpu-install/6.3/ubuntu/jammy/amdgpu-install_6.3.60300-1_all.deb sudo dpkg -i amdgpu-install*.deb sudo amdgpu-install --usecase=rocm,lrt,openclsdk # Verify ROCm sees your GPU r...
Read more

llama.cpp flash attention: how to enable and benchmark

Image
Flash attention computes attention with far less memory traffic. The math is identical; the memory movement is not. In llama.cpp it matters most in two situations: long context windows and tight video memory (VRAM). With flash attention enabled, the key-value cache shrinks substantially. A 32k-token context window becomes realistic on a 16GB card, and the prompt-processing phase speeds up, so long documents get ingested faster. The catch is compatibility: not every model and quantization combination supports it, and on some setups the speedup is modest. If you run short chats on a strong GPU, you may barely notice. Where it shines is retrieval-augmented generation (RAG) and long-document work on mid-range cards. Enable it, compare your tokens per second, and keep it if your workload benefits. For the common case of pasting a long PDF and asking questions, it is close to free performance. See how context length changes VRAM needs .
Read more

SillyTavern with Local LLMs: Complete Setup for AI Companions

Image
SillyTavern is the most feature-rich frontend for character-based AI conversations. What began as an AI roleplay tool has evolved into a powerful platform with persona management, memory systems, extensive prompt engineering tools, and connections to every major local inference backend. Here's how to set it up with local models. Quick connect: SillyTavern connects to Ollama, KoboldCpp, llama.cpp server, and TabbyAPI (ExLlamaV2) out of the box. Any of the hardware setups at LLMHardware.io provides a solid backend for SillyTavern. Installation Install via git: git clone https://github.com/SillyTavern/SillyTavern && cd SillyTavern && npm install then node server.js . Access at localhost:8000. For persistent installation, use the included start.sh script. SillyTavern uses Node.js and runs on any platform. The companion app (STMP) enables mobile access to your SillyTavern instance via Tailscale or local network. Character Cards and the TavernAI Format Chara...
Read more