> ## Documentation Index
> Fetch the complete documentation index at: https://docs.fiddler.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# items_from_jsonl

> Build one dataset item per line of a JSON Lines file (empty lines are skipped).

**New in version 0.8.0**

## Parameters

**file\_path** `str | Path` *required*

Path to the JSON Lines file, one JSON object per line.

**input\_keys** `list[str]` *required*

Keys to use as inputs; required, since JSON objects have no
fixed columns.

**expected\_output\_keys** `list[str] | None` (default: None)

Keys holding the expected outputs.

**metadata\_keys** `list[str] | None` (default: None)

Keys to store as metadata.

**extras\_keys** `list[str] | None` (default: None)

Keys to store as extras.

**id\_key** `str` (default: "id")

Key holding each item's id. Defaults to "id". Lines without one
get a new id.

**source\_name\_key** `str` (default: "source\_name")

Key holding the source name. Defaults to "source\_name".

**source\_id\_key** `str` (default: "source\_id")

Key holding the source id. Defaults to "source\_id".

## Returns

One item per non-empty line, in file order.

## Raises

**ValueError** – If the file has no items, `input_keys` is empty, or a line
has no input values.

## Example

```python theme={null}
items = items_from_jsonl("tests.jsonl", input_keys=["question"])
dataset = dataset.update_items(items=items).dataset
```