Summarization Dataset (JSONL)
An abstractive-summarization dataset in JSON Lines — 15 short synthetic news-style documents each paired with a one-sentence summary. A fixture for training and evaluating summarization models and for testing JSONL ingestion.
{"document": "The city council approved a plan to add twelve kilometres of protected bike lanes over the next two years, funded by a regional transport grant. Officials say the network should cut short car trips and improve safety at three busy intersections.", "summary": "The council approved 12 km of new protected bike lanes over two years to improve safety and cut car trips."}
{"document": "A small bakery in the old quarter has switched entirely to renewable energy, installing solar panels on its roof and a battery to store power for its ovens. The owner reports that energy bills have fallen by nearly a third since the change.", "summary": "A bakery went fully solar with rooftop panels and a battery, cutting its energy bills by about a third."}
{"document": "Researchers published a study showing that a short daily walk was associated with better sleep quality among office workers. Participants who walked for twenty minutes reported falling asleep faster than those who did not.", "summary": "A study links a 20-minute daily walk to faster, better sleep among office workers."}
{"document": "The museum extended its popular exhibition on ancient navigation by two months after record attendance. Visitors can see reconstructed instruments and interactive maps that trace historic trade routes across three continents.", "summary": "A museum extended its well-attended ancient-navigation exhibition by two months due to record crowds."}
{"document": "A regional train operator introduced contactless payment across all its stations, letting passengers tap a card or phone instead of buying paper tickets. The rollout follows a successful trial on two of its busiest lines.", "summary": "A train operator rolled out contactless payment at all stations after a successful two-line trial."}
{"document": "Volunteers planted more than five thousand native saplings along a river bank to reduce erosion and restore habitat. The project, run over a single weekend, drew families and school groups from across the county.", "summary": "Volunteers planted 5,000+ native saplings along a river in one weekend to fight erosion and restore habitat."}
{"document": "A start-up unveiled a reusable packaging service that lets shoppers return containers to any partner store. Early data suggests the scheme has already avoided tens of thousands of single-use boxes in its first quarter.", "summary": "A start-up's reusable-packaging returns scheme avoided tens of thousands of single-use boxes in one quarter."}
{"document": "The local library launched a free evening coding club for teenagers, pairing them with volunteer mentors from nearby tech firms. Places filled within days, and organisers are already planning a second weekly session.", "summary": "A library's free teen coding club filled instantly, prompting plans for a second weekly session."}
{"document": "A coastal town completed a flood-defence upgrade that combines a raised sea wall with restored salt marshes. Engineers say the natural marsh buffer should absorb storm surges while supporting local wildlife.", "summary": "A coastal town paired a raised sea wall with restored marshes to buffer storm surges and aid wildlife."}
{"document": "An airline began trialling lighter, recyclable meal trays on short-haul flights as part of a waste-reduction programme. The carrier estimates the change could remove several hundred tonnes of plastic each year.", "summary": "An airline is trialling recyclable meal trays that could cut several hundred tonnes of plastic a year."}
{"document": "A university opened a lab where students can test prototypes on industrial 3D printers at no cost. The facility is shared across the engineering, design, and medicine faculties to encourage cross-disciplinary projects.", "summary": "A university opened a free shared 3D-printing lab to spur cross-disciplinary student projects."}
{"document": "The transit authority added real-time arrival displays at two hundred bus stops, drawing on GPS data from the fleet. Riders can now see waiting times at the stop rather than relying only on a phone app.", "summary": "The transit authority put real-time GPS arrival displays at 200 bus stops."}
{"document": "A community garden converted a disused car park into raised vegetable beds and a small orchard. Produce is shared among members, and surplus is donated to a nearby food bank each week.", "summary": "A disused car park became a community garden that donates surplus produce to a food bank weekly."}
{"document": "A software team open-sourced an internal tool for tracking accessibility issues in web pages. Within a month, outside contributors had added support for two new languages and fixed several reported bugs.", "summary": "A team open-sourced an accessibility-tracking tool that quickly gained contributors, languages, and fixes."}
{"document": "The national weather service upgraded its forecasting model to run at a finer resolution, improving predictions for localized storms. Early evaluations show fewer missed warnings for heavy rainfall in mountainous areas.", "summary": "A finer-resolution forecasting model improved localized-storm and heavy-rain warnings in mountains."}
Specifications
- Records
- 15
- Schema
- document, summary
- Task
- abstractive summarization
What is a .jsonl file?
JSONL (JSON Lines) is a text format where each line is a complete, independent JSON value, allowing records to be streamed and appended without parsing the whole file. It is not itself a JSON array and each line must stand alone. It is common in logging, machine learning datasets, and data pipelines.
How to use this file
Use an example JSONL to test line-by-line streaming parsers, append-and-resume ingestion, and batch pipelines that process one record per line.
Code examples
import json
with open("summarization.jsonl") as f:
rows = [json.loads(line) for line in f]
print(len(rows), rows[0])Related files
- jsonlChat Fine-tuning Dataset — Anthropic Format (JSONL)The same synthetic conversations in the Anthropic Messages JSONL shape — a top-level system prompt plus a messages array of user and assistant turns. The format twin of the OpenAI file, for testing chat-format conversion.

- jsonlChat Fine-tuning Dataset — OpenAI Format (JSONL)A chat fine-tuning dataset in the OpenAI JSONL format — one conversation per line as a messages array with system, user, and assistant turns. Synthetic Q&A content. Paired with an Anthropic-format twin for testing format converters.

- jsonExtractive QA Dataset — SQuAD v2 Format (JSON)An extractive question-answering dataset in the SQuAD v2.0 JSON structure — titled articles with context paragraphs, questions, character-offset answers, and one deliberately unanswerable question. Synthetic content; a fixture for QA model training and SQuAD-format loaders.

- jsonlInstruction-tuning Dataset — Alpaca Format (JSONL)An instruction-tuning dataset in the Alpaca JSONL format — 20 instruction / input / output triples covering small transformations, extraction, and factual answers. Fully synthetic; a fixture for supervised fine-tuning pipelines.

- jsonlNamed-Entity Recognition Dataset — BIO Tags (JSONL)A token-classification dataset in JSON Lines — 16 tokenized sentences with aligned BIO tags for person, organisation, and location entities. All names, companies, and places are fictional. A fixture for NER model training and sequence-labelling tooling.

- jsonlParallel Translation Corpus — EN↔ES (JSONL)An English↔Spanish parallel corpus in JSON Lines — 20 aligned sentence pairs of everyday phrases. A fixture for training and evaluating machine-translation models and for testing UTF-8 handling of accented characters.

Generated by generation/ai_datasets.py. Free for any use, no attribution required — license.