View Parquet, Excel & other binary files online — no upload

Most data viewers make you upload a file to a server. Yellorn doesn't. Drag a binary file onto the editor — or onto the + button to open it in a new tab — and it's decoded locally, in a background worker, into a JSON preview you can read, search, and convert. The file never leaves your machine.

The format is detected from the file extension, so there's nothing to configure — just drop and read.

• Parquet and Arrow / Feather — columnar data becomes an array of row objects.
• Excel (.xlsx / .xlsm / .xls) — one sheet becomes a row array; a multi-sheet workbook becomes an object keyed by sheet name. Dates read as ISO strings.
• MessagePack, CBOR, and BSON — decoded to whatever they hold (an object, an array, or a stream of records).
• Protobuf — decoded without the .proto schema, so fields appear keyed by their field number.
• gzip — decompressed to its underlying text (gzipped JSON comes back as JSON).

Whatever the container, the short version is the same: you get clean, browsable JSON.

Once a file is open you have the full editor: the tree, graph, and table views; JMESPath search to query deep into the data; and one-click conversion to JSON, CSV, YAML, and more.

Binary formats are read-only: you're viewing a decoded snapshot, so you can convert the contents out to a text format, but Yellorn doesn't write binary files back.

Very large files are capped to keep the editor responsive. When that happens, a notice reads “Large file: showing the first N of M rows” so you always know you're looking at a preview, not the whole dataset.

← All help articles · Inspect AI model files · Convert between formats