Instructions to use yashcse21/text2sql-llama1b-adapters-caution with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use yashcse21/text2sql-llama1b-adapters-caution with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] hf download yashcse21/text2sql-llama1b-adapters-caution --local-dir text2sql-llama1b-adapters-caution
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Atomic Chat
File size: 723 Bytes
04c3f04 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 | ---
license: llama3.2
base_model: mlx-community/Llama-3.2-1B-Instruct-bf16
tags: [text-to-sql, lora, mlx, reasoning]
library_name: mlx
---
# Text-to-SQL · adapters_caution
MLX LoRA adapter fine-tuned from
`mlx-community/Llama-3.2-1B-Instruct-bf16` to turn a natural-language question + a `CREATE TABLE` schema into a
schema-faithful SQLite `SELECT` (reasoning models emit `Reasoning: … / SQL: …`).
```python
# download the adapter, then:
from mlx_lm import load
model, tok = load("mlx-community/Llama-3.2-1B-Instruct-bf16", adapter_path="<downloaded-adapter-dir>")
```
Pair generation with a deterministic AST guardrail (single read-only SELECT).
See the project repo for the data pipeline, study, and evaluation.
|