README.md

metadata

base_model: TinyLlama/TinyLlama-1.1B-Chat-v1.0
library_name: peft
pipeline_tag: text-generation
license: mit
language:
  - en
datasets:
  - spider
tags:
  - base_model:adapter:TinyLlama/TinyLlama-1.1B-Chat-v1.0
  - lora
  - sft
  - transformers
  - trl
  - text-to-sql
  - sql
  - natural-language-processing
metrics:
  - loss

Text-to-SQL TinyLlama LoRA Adapter

A fine-tuned LoRA adapter that converts natural language questions into SQL queries. Built on top of TinyLlama-1.1B-Chat-v1.0 using Supervised Fine-Tuning (SFT) on the Spider benchmark dataset.

Model Details

Model Description

This is a LoRA (Low-Rank Adaptation) adapter fine-tuned to generate SQL queries from natural language questions. Only 0.10% of the base model's parameters were trained, making it extremely lightweight (4.5 MB) while still achieving strong results.

• Developed by: Rj18
• Model type: Causal Language Model (LoRA Adapter)
• Language(s): English
• License: MIT
• Fine-tuned from: TinyLlama/TinyLlama-1.1B-Chat-v1.0

Model Sources

• Repository: https://github.com/18-RAJAT/Interactive-Production-text2sql-Pipeline

How to Use

import torch from transformers import AutoModelForCausalLM, AutoTokenizer from peft import PeftModel

Load base model and tokenizer

base_model = "TinyLlama/TinyLlama-1.1B-Chat-v1.0" adapter = "Rj18/text-to-sql-tinyllama-lora"

tokenizer = AutoTokenizer.from_pretrained(adapter) model = AutoModelForCausalLM.from_pretrained(base_model, torch_dtype=torch.float16) model = PeftModel.from_pretrained(model, adapter) model.eval()

Generate SQL

question = "How many employees are in each department?" prompt = f"[INST] Generate SQL for the following question.\nQuestion: {question} [/INST]\n"

inputs = tokenizer(prompt, return_tensors="pt") with torch.no_grad(): outputs = model.generate(**inputs, max_new_tokens=128, temperature=0.1)

sql = tokenizer.decode(outputs[0], skip_special_tokens=True) print(sql)