Gemini JSONL Dataset Generator

A lightweight Python script that uses the Google Gemini API to generate structured datasets in JSONL format.
It reads the base prompt from a .txt file defined in your .env file and writes each generated object line-by-line into a .jsonl file.

---

Installation

git clone https://github.com/erkerem2/gemini-jsonl-generator.git
cd gemini-jsonl-generator
pip install -r requirements.txt

Create a .env file

GEMINI_KEY="your_api_key"
PROMPT_FILE="prompt.txt path"

Usage

python main.py
--total 100
--per_call 20
--outfile output.jsonl
--temperature 0.7

Arguments

Argument	Description	Default
--api_key	Gemini API key (can be read from .env)	.env:GEMINI_KEY
--model	Model name	gemini-2.5-flash-lite
--total	Total number of JSON objects to generate	6
--per_call	Number of objects per API call	2
--outfile	Output file name	output.jsonl
--temperature	Sampling temperature	0.8