.jsonl File — JSON Lines

JSONL (JSON Lines, also called newline-delimited JSON or NDJSON) is a text format where each line is a valid JSON value, typically a JSON object. The format is designed for streaming structured data, log files, and processing large datasets line by line without loading the entire file into memory. Each line is independently parseable, making JSONL ideal for append-only data, streaming APIs, Unix pipeline processing with tools like jq, and parallel processing where different workers handle different lines. JSONL is widely used in machine learning for training data (OpenAI's GPT fine-tuning format, Hugging Face datasets), logging systems (structured JSON logging with one event per line), data engineering pipelines (BigQuery, Elasticsearch bulk API), and analytics data export. The format combines JSON's self-describing structure with the line-oriented processing model of Unix tools. Unlike JSON arrays, JSONL files don't need a complete parse to extract individual records, making them practical for files that are gigabytes or terabytes in size. JSONL files can be compressed with gzip for efficient storage and transfer while maintaining line-oriented streaming decompression.