Class: Dataset<IsLegacyDataset>

A dataset is a collection of records, such as model inputs and expected outputs, which represent data you can use to evaluate and fine-tune models. You can log production data to datasets, curate them with interesting examples, edit/delete records, and run evaluations against them.

You should not create Dataset objects directly. Instead, use the braintrust.initDataset() method.

Type parameters

Name	Type
`IsLegacyDataset`	extends `boolean` = typeof `DEFAULT_IS_LEGACY_DATASET`

Hierarchy

ObjectFetcher<DatasetRecord<IsLegacyDataset>>

↳ Dataset

Accessors

id

• get id(): Promise<string>

Returns

Promise<string>

Overrides

ObjectFetcher.id

name

• get name(): Promise<string>

Returns

Promise<string>

project

• get project(): Promise<ObjectMetadata>

Returns

Promise<ObjectMetadata>

Constructors

constructor

• new Dataset<IsLegacyDataset>(state, lazyMetadata, pinnedVersion?, legacy?): Dataset<IsLegacyDataset>

Type parameters

Name	Type
`IsLegacyDataset`	extends `boolean` = `false`

Parameters

Name	Type
`state`	`BraintrustState`
`lazyMetadata`	`LazyValue`<`ProjectDatasetMetadata`>
`pinnedVersion?`	`string`
`legacy?`	`IsLegacyDataset`

Returns

Dataset<IsLegacyDataset>

Overrides

ObjectFetcher<DatasetRecord<IsLegacyDataset>>.constructor

Methods

[asyncIterator]

▸ [asyncIterator](): AsyncIterator<WithTransactionId<DatasetRecord<IsLegacyDataset>>, any, undefined>

Returns

AsyncIterator<WithTransactionId<DatasetRecord<IsLegacyDataset>>, any, undefined>

Inherited from

ObjectFetcher.[asyncIterator]

clearCache

▸ clearCache(): void

Returns

void

Inherited from

ObjectFetcher.clearCache

close

▸ close(): Promise<string>

Returns

Promise<string>

Deprecated

This function is deprecated. You can simply remove it from your code.

delete

▸ delete(id): string

Parameters

Name	Type
`id`	`string`

Returns

string

fetch

▸ fetch(): AsyncGenerator<WithTransactionId<DatasetRecord<IsLegacyDataset>>, any, unknown>

Returns

AsyncGenerator<WithTransactionId<DatasetRecord<IsLegacyDataset>>, any, unknown>

Inherited from

ObjectFetcher.fetch

fetchedData

▸ fetchedData(): Promise<WithTransactionId<DatasetRecord<IsLegacyDataset>>[]>

Returns

Promise<WithTransactionId<DatasetRecord<IsLegacyDataset>>[]>

Inherited from

ObjectFetcher.fetchedData

flush

▸ flush(): Promise<void>

Flush any pending rows to the server.

Returns

Promise<void>

getState

▸ getState(): Promise<BraintrustState>

Returns

Promise<BraintrustState>

Overrides

ObjectFetcher.getState

insert

▸ insert(event): string

Insert a single record to the dataset. The record will be batched and uploaded behind the scenes. If you pass in an id, and a record with that id already exists, it will be overwritten (upsert).

Parameters

Name	Type	Description
`event`	`Object`	The event to log.
`event.expected?`	`unknown`	The output of your application, including post-processing (an arbitrary, JSON serializable object).
`event.id?`	`string`	(Optional) a unique identifier for the event. If you don't provide one, Braintrust will generate one for you.
`event.input?`	`unknown`	The argument that uniquely define an input case (an arbitrary, JSON serializable object).
`event.metadata?`	`Record`<`string`, `unknown`>	(Optional) a dictionary with additional data about the test example, model outputs, or just about anything else that's relevant, that you can use to help find and analyze examples later. For example, you could log the `prompt`, example's `id`, or anything else that would be useful to slice/dice later. The values in `metadata` can be any JSON-serializable type, but its keys must be strings.
`event.output?`	`unknown`	-
`event.tags?`	`string`[]	(Optional) a list of strings that you can use to filter and group records later.

Returns

string

The id of the logged record.

summarize

▸ summarize(options?): Promise<DatasetSummary>

Summarize the dataset, including high level metrics about its size and other metadata.

Parameters

Name	Type
`options`	`Object`
`options.summarizeData?`	`boolean`

Returns

Promise<DatasetSummary>

DatasetSummary

A summary of the dataset.

update

▸ update(event): string

Update fields of a single record in the dataset. The updated fields will be batched and uploaded behind the scenes. You must pass in an id of the record to update. Only the fields provided will be updated; other fields will remain unchanged.

Parameters

Name	Type	Description
`event`	`Object`	The fields to update in the record.
`event.expected?`	`unknown`	(Optional) The new expected output value for the record (an arbitrary, JSON serializable object).
`event.id`	`string`	The unique identifier of the record to update.
`event.input?`	`unknown`	(Optional) The new input value for the record (an arbitrary, JSON serializable object).
`event.metadata?`	`Record`<`string`, `unknown`>	(Optional) A dictionary to update the metadata of the record. The values in `metadata` can be any JSON-serializable type, but its keys must be strings.
`event.tags?`	`string`[]	(Optional) A list of strings to update the tags of the record.

On this page