Dataset
Dataset
type Dataset = {
id: string;
createdAt: string;
name?: Maybe<string>;
description?: Maybe<string>;
metadata: Record<string, any>;
items?: Maybe<DatasetItem[]>;
type?: Maybe<DatasetType>
}
The unique identifier of the dataset. This is an immutable field generated when the dataset is created.
The date and time when the dataset was created, expressed in ISO 8601 format.
A dictionary containing additional data associated with the dataset. This can include user-defined key-value pairs that provide more context or details about the dataset.
An optional name for the dataset. This can be used to give a descriptive, human-readable title to the dataset for easier identification.
An optional description field that can provide more detailed information about the dataset, such as its purpose, contents, or specific characteristics.
An optional list of items contained within the dataset. Each item in the list is a dictionary representing a specific data entry in the dataset.
An optional type of dataset. Defaults to “key_value”, other option is “generation”.
Create a dataset
const dataset = await literalClient.api.createDataset({
name: "Foo",
description: "A dataset to store samples.",
metadata: { isDemo: true },
});
Params
A partial representation of a dataset.
An optional name for the dataset. This can be used to give a descriptive, human-readable title to the dataset for easier identification.
An optional description field that can provide more detailed information about the dataset, such as its purpose, contents, or specific characteristics.
A dictionary containing additional data associated with the dataset. This can include user-defined key-value pairs that provide more context or details about the dataset.
Response
Return a Dataset
Update a dataset
const dataset = await literalClient.api.updateDataset("dataset_id", {
name: "Baz",
});
Params
The unique identifier of the dataset. This is an immutable field generated when the dataset is created.
A partial representation of a dataset.
An optional name for the dataset. This can be used to give a descriptive, human-readable title to the dataset for easier identification.
An optional description field that can provide more detailed information about the dataset, such as its purpose, contents, or specific characteristics.
A dictionary containing additional data associated with the dataset. This can include user-defined key-value pairs that provide more context or details about the dataset.
Response
Return a Dataset
Get a dataset
const dataset = await literalClient.api.getDataset("dataset_id")
Params
The unique identifier of the dataset. This is an immutable field generated when the dataset is created.
Response
Return a Dataset
Delete a dataset
await literalClient.api.deleteDataset("dataset_id")
Params
The unique identifier of the dataset. This is an immutable field generated when the dataset is created.
Response
Return a Dataset
DatasetItem
type DatasetItem = {
id: string;
createdAt: string;
datasetId: string;
metadata: Record<string, any>;
input: Record<string, any>;
expectedOutput?: Maybe<Record<string, any>>;
intermediarySteps: Record<string, any>[];
}
The unique identifier of the dataset item. Each item within a dataset has its own unique ID.
The date and time when the dataset item was created, expressed in ISO 8601 format.
The identifier of the dataset to which this item belongs. Links the item to its parent dataset.
A dictionary containing additional data associated with this dataset item. This can include specific details relevant to the item.
The input data for this dataset item. Typically contains the parameters or information that forms the LLM request.
The optional output data for this dataset item. Represents the result or outcome associated with the item’s input.
An optional list of intermediary steps involved in the output. This can include any processing steps or intermediate results. They are automatically generated when using “steps” to build a dataset.
Create a dataset item
const datasetItem = await literalClient.api.createDatasetItem("dataset_id", {
input: { content: "What is Literal AI?" },
expectedOutput: { content: "Literal AI is an observability solution." },
});
# Or directly through a Dataset
const datasetItem = await dataset.createDatasetItem("dataset_id", {
input: { content: "What is Literal AI?" },
expectedOutput: { content: "Literal AI is an observability solution." },
});
Params
The identifier of the dataset to which this item belongs.
A partial representation of a dataset item.
The input data for this dataset item. Typically contains the parameters or information that forms the LLM request.
The optional output data for this dataset item. Represents the result or outcome associated with the item’s input.
A dictionary containing additional data associated with this dataset item. This can include specific details relevant to the item.
Response
Return a DatasetItem
Add a step to a dataset
const datasetItem = await literalClient.api.addStepToDataset("dataset_id", "step_id");
# Or directly through a Dataset
const datasetItem = await dataset.addStep("step_id");
Params
The identifier of the dataset to which this item belongs.
The identifier of the step created beforehand.
A dictionary containing additional data associated with this dataset item. This can include specific details relevant to the item.
Response
Return a DatasetItem
Add a generation to a dataset
const datasetItem = await literalClient.api.addGeneationToDataset("dataset_id", "generation_id");
# Or directly through a Dataset
const datasetItem = await dataset.addGeneration("generation_id");
Params
The identifier of the dataset to which this item belongs.
The identifier of the generation created beforehand.
A dictionary containing additional data associated with this dataset item. This can include specific details relevant to the item.
Response
Return a DatasetItem
Get a dataset item
const datasetItem = await literalClient.api.getDatasetItem("dataset_item_id");
Params
The unique identifier of the dataset item. Each item within a dataset has its own unique ID.
Response
Return a DatasetItem
Delete a dataset item
await literalClient.api.deleteDatasetItem("dataset_item_id");
# Or directly through a Dataset
await dataset.deleteDatasetItem("dataset_item_id");
Params
The unique identifier of the dataset item. Each item within a dataset has its own unique ID.
Response
Return a DatasetItem
Was this page helpful?