> ## Documentation Index
> Fetch the complete documentation index at: https://docs.openlayer.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Publish records
> Publish records to a data source (formerly known as "inference pipeline").
Use this endpoint to stream individual inference data points to Openlayer.
If you want to upload many inferences in one go, please use the [batch upload method](https://github.com/openlayer-ai/openlayer-python/blob/main/examples/monitoring/upload_batch_data.py) instead.
## OpenAPI
````yaml post /inference-pipelines/{inferencePipelineId}/data-stream
openapi: 3.0.3
info:
contact:
email: support@openlayer.com
name: Openlayer
url: https://openlayer.com/
description: API for interacting with the Openlayer server.
title: Openlayer API
version: '1.0'
x-logo:
url: https://logo.clearbit.com/openlayer.com
servers:
- url: https://api.openlayer.com/v1
description: Our prod backend
security:
- bearerAuth: []
paths:
/inference-pipelines/{inferencePipelineId}/data-stream:
post:
tags:
- Monitoring
summary: Publish inference
description: Publish an inference data point to an inference pipeline.
operationId: streamData
parameters:
- $ref: '#/components/parameters/inferencePipelineId'
requestBody:
content:
application/json:
schema:
type: object
properties:
rows:
type: array
description: A list of inference data points with inputs and outputs
example:
- user_query: what is the meaning of life?
output: '42'
tokens: 7
cost: 0.02
timestamp: 1620000000
nullable: false
items:
type: object
additionalProperties: true
config:
oneOf:
- $ref: '#/components/schemas/LLMData'
- $ref: '#/components/schemas/TabularClassificationData'
- $ref: '#/components/schemas/TabularRegressionData'
- $ref: '#/components/schemas/TextClassificationData'
example:
prompt:
- role: user
content: '{{ user_query }}'
inputVariableNames:
- user_query
outputColumnName: output
timestampColumnName: timestamp
costColumnName: cost
numOfTokenColumnName: tokens
description: >-
Configuration for the data stream. Depends on your
**Openlayer project task type**.
required:
- rows
- config
responses:
'200':
description: Status OK.
content:
application/json:
schema:
type: object
required:
- success
properties:
success:
type: boolean
enum:
- true
'500':
$ref: '#/components/responses/UnexpectedError'
x-codeSamples:
- lang: python
source: >
from openlayer import Openlayer
# Let's say we want to stream the following row, which represents a
model prediction:
data = {
"user_query": "what's the meaning of life?",
"output": "42",
"tokens": 7,
"cost": 0.02,
"timestamp": 1620000000
}
# Prepare the config for the data, which depends on your project's
task type. In this
# case, we have an LLM project:
from openlayer.types.inference_pipelines import data_stream_params
config = data_stream_params.ConfigLlmData(
input_variable_names=["user_query"],
output_column_name="output",
num_of_token_column_name="tokens",
cost_column_name="cost",
timestamp_column_name="timestamp",
prompt=[{"role": "user", "content": "{{ user_query }}"}],
)
client = Openlayer()
data_stream_response = client.inference_pipelines.data.stream(
inference_pipeline_id="182bd5e5-6e1a-4fe4-a799-aa6d9a6ab26e",
rows=[data],
config=config,
)
- lang: typescript
source: |
import Openlayer from 'openlayer';
const openlayer = new Openlayer();
await openlayer.inferencePipelines.data.stream(
{
inferencePipelineId: '182bd5e5-6e1a-4fe4-a799-aa6d9a6ab26e',
config: {
inputVariableNames: ['user_query'],
outputColumnName: 'output',
numOfTokenColumnName: 'tokens',
costColumnName: 'cost',
timestampColumnName: 'timestamp',
prompt: [{ role: 'user', content: '{{ user_query }}' }],
},
rows: [
{
user_query: "what's the meaning of life?",
output: '42',
tokens: 7,
cost: 0.02,
timestamp: 1620000000,
},
],
},
);
- lang: go
source: >
package main
import (
"context"
"github.com/openlayer-ai/openlayer-go"
"github.com/openlayer-ai/openlayer-go/option"
)
client := openlayer.NewClient()
inferencePipelineDataStreamResponse, err :=
client.InferencePipelines.Data.Stream(
context.TODO(),
inferencePipelineId: "182bd5e5-6e1a-4fe4-a799-aa6d9a6ab26e",
openlayer.InferencePipelineDataStreamParams{
Config: openlayer.F[openlayer.InferencePipelineDataStreamParamsConfigUnion](openlayer.InferencePipelineDataStreamParamsConfigLlmData{
OutputColumnName: openlayer.F("output"),
NumOfTokenColumnName: openlayer.F("tokens"),
CostColumnName: openlayer.F("cost"),
TimestampColumnName: openlayer.F("timestamp"),
}),
Rows: openlayer.F([]map[string]interface{}{map[string]interface{}{
"user_query": "what's the meaning of life?",
"output": "42",
"tokens": 7,
"cost": 0.02,
"timestamp": 1620000000,
}}),
},
)
if err != nil {
panic(err.Error())
}
- lang: java
source: >
import com.openlayer.api.client.OpenlayerClient;
import com.openlayer.api.client.okhttp.OpenlayerOkHttpClient;
import com.openlayer.api.models.InferencePipelineDataStreamParams;
import com.openlayer.api.models.InferencePipelineDataStreamResponse;
import java.util.List;
OpenlayerClient client = OpenlayerOkHttpClient.fromEnv();
// Let's say this is the row with the relevant fields
InferencePipelineDataStreamParams.Row row =
InferencePipelineDataStreamParams.Row.builder()
.putAdditionalProperty("user_query", JsonString.of("what's the meaning of life?"))
.putAdditionalProperty("output", JsonString.of("42"))
.putAdditionalProperty("tokens", JsonNumber.of(7))
.putAdditionalProperty("cost", JsonNumber.of(0.02))
.build();
// Create Inference Pipeline Data Stream Parameters
InferencePipelineDataStreamParams params =
InferencePipelineDataStreamParams.builder()
.inferencePipelineId("182bd5e5-6e1a-4fe4-a799-aa6d9a6ab26e")
.config(InferencePipelineDataStreamParams.Config.ofLlmData(InferencePipelineDataStreamParams.Config.LlmData.builder()
.outputColumnName("output")
.costColumnName("cost")
.inputVariableNames(List.of("user_query"))
.numOfTokenColumnName("tokens")
.timestampColumnName("timestamp")
.build()))
.row(List.of(InferencePipelineDataStreamParams.Row.builder().build()))
.build();
// Make the request
InferencePipelineDataStreamResponse
inferencePipelineDataStreamResponse =
client.inferencePipelines().data().stream(params);
- lang: curl
source: |
curl --request POST \
--url https://api.openlayer.com/v1/inference-pipelines/{inferencePipelineId}/data-stream \
--header 'Authorization: Bearer ' \
--header 'Content-Type: application/json' \
--data '{
"rows": [
{
"user_query": "what is the meaning of life?",
"output": "42",
"tokens": 7,
"cost": 0.02,
"timestamp": 1620000000
}
],
"config": {
"prompt": [
{
"role": "user",
"content": "{{ user_query }}"
}
],
"inputVariableNames": [
"user_query"
],
"outputColumnName": "output",
"timestampColumnName": "timestamp",
"costColumnName": "cost",
"numOfTokenColumnName": "tokens"
}
}'
components:
parameters:
inferencePipelineId:
name: inferencePipelineId
in: path
description: The inference pipeline id (a UUID).
required: true
schema:
type: string
format: uuid
schemas:
LLMData:
title: LLM
type: object
properties:
numOfTokenColumnName:
type: string
description: Name of the column with the total number of tokens.
example: num_tokens
nullable: true
contextColumnName:
type: string
description: >-
Name of the column with the context retrieved. Applies to RAG use
cases. Providing the context enables RAG-specific metrics.
example: context
costColumnName:
type: string
description: Name of the column with the cost associated with each row.
example: cost
groundTruthColumnName:
type: string
description: Name of the column with the ground truths.
example: ground_truth
inferenceIdColumnName:
type: string
description: >-
Name of the column with the inference ids. This is useful if you
want to update rows at a later point in time. If not provided, a
unique id is generated by Openlayer.
example: id
inputVariableNames:
type: array
description: >-
Array of input variable names. Each input variable should be a
dataset column.
example:
- user_query
items:
type: string
latencyColumnName:
type: string
description: Name of the column with the latencies.
example: latency
metadata:
type: object
description: Object with metadata.
outputColumnName:
type: string
description: Name of the column with the model outputs.
example: output
prompt:
type: array
description: Prompt for the LLM.
example:
- role: user
content: '{{ user_query }}'
items:
type: object
properties:
role:
type: string
description: Role of the prompt.
example: user
content:
type: string
description: Content of the prompt.
example: '{{ user_query }}'
questionColumnName:
type: string
description: >-
Name of the column with the questions. Applies to RAG use cases.
Providing the question enables RAG-specific metrics.
example: question
timestampColumnName:
type: string
description: >-
Name of the column with the timestamps. Timestamps must be in UNIX
sec format. If not provided, the upload timestamp is used.
example: timestamp
userIdColumnName:
type: string
description: Name of the column with the user id.
nullable: true
example: user_id
sessionIdColumnName:
type: string
description: Name of the column with the session id.
nullable: true
example: session_id
required:
- outputColumnName
TabularClassificationData:
title: Tabular classification
type: object
properties:
categoricalFeatureNames:
type: array
description: >-
Array with the names of all categorical features in the dataset.
E.g. ["Age", "Geography"].
example:
- Geography
items:
type: string
classNames:
type: array
description: >-
List of class names indexed by label integer in the dataset. E.g.
["Retained", "Exited"] when 0, 1 are in your label column.
example:
- Retained
- Exited
items:
type: string
featureNames:
type: array
description: Array with all input feature names.
example:
- Age
- Geography
items:
type: string
inferenceIdColumnName:
type: string
description: >-
Name of the column with the inference ids. This is useful if you
want to update rows at a later point in time. If not provided, a
unique id is generated by Openlayer.
example: id
labelColumnName:
type: string
description: >-
Name of the column with the labels. The data in this column must be
**zero-indexed integers**, matching the list provided in
`classNames`.
example: label
latencyColumnName:
type: string
description: Name of the column with the latencies.
example: latency
metadata:
type: object
description: Object with metadata.
predictionsColumnName:
type: string
description: >-
Name of the column with the model's predictions as **zero-indexed
integers**.
example: prediction
predictionScoresColumnName:
type: string
description: >-
Name of the column with the model's predictions as **lists of class
probabilities**.
example: prediction_scores
timestampColumnName:
type: string
description: >-
Name of the column with the timestamps. Timestamps must be in UNIX
sec format. If not provided, the upload timestamp is used.
example: timestamp
required:
- classNames
TabularRegressionData:
title: Tabular regression
type: object
properties:
categoricalFeatureNames:
type: array
description: >-
Array with the names of all categorical features in the dataset.
E.g. ["Gender", "Geography"].
example:
- Gender
- Geography
items:
type: string
featureNames:
type: array
description: Array with all input feature names.
items:
type: string
inferenceIdColumnName:
type: string
description: >-
Name of the column with the inference ids. This is useful if you
want to update rows at a later point in time. If not provided, a
unique id is generated by Openlayer.
example: id
latencyColumnName:
type: string
description: Name of the column with the latencies.
example: latency
metadata:
type: object
description: Object with metadata.
predictionsColumnName:
type: string
description: Name of the column with the model's predictions.
example: prediction
targetColumnName:
type: string
description: Name of the column with the targets (ground truth values).
example: target
timestampColumnName:
type: string
description: >-
Name of the column with the timestamps. Timestamps must be in UNIX
sec format. If not provided, the upload timestamp is used.
example: timestamp
TextClassificationData:
title: Text classification
type: object
properties:
classNames:
type: array
description: >-
List of class names indexed by label integer in the dataset. E.g.
["Retained", "Exited"] when 0, 1 are in your label column.
example:
- Retained
- Exited
items:
type: string
inferenceIdColumnName:
type: string
description: >-
Name of the column with the inference ids. This is useful if you
want to update rows at a later point in time. If not provided, a
unique id is generated by Openlayer.
example: id
labelColumnName:
type: string
description: >-
Name of the column with the labels. The data in this column must be
**zero-indexed integers**, matching the list provided in
`classNames`.
example: label
latencyColumnName:
type: string
description: Name of the column with the latencies.
example: latency
metadata:
type: object
description: Object with metadata.
predictionsColumnName:
type: string
description: >-
Name of the column with the model's predictions as **zero-indexed
integers**.
example: prediction
predictionScoresColumnName:
type: string
description: >-
Name of the column with the model's predictions as **lists of class
probabilities**.
example: prediction_scores
textColumnName:
type: string
description: Name of the column with the text data.
example: user_query
timestampColumnName:
type: string
description: >-
Name of the column with the timestamps. Timestamps must be in UNIX
sec format. If not provided, the upload timestamp is used.
example: timestamp
required:
- classNames
responses:
UnexpectedError:
description: Unexpected error.
content:
application/json:
schema:
type: object
required:
- code
- error
properties:
code:
type: integer
format: int32
error:
type: string
securitySchemes:
bearerAuth:
type: http
scheme: bearer
description: >
Bearer authentication header of the form `Bearer `, where
`` is your workspace API key. See [Find your API
key](https://www.openlayer.com/docs/workspace-and-projects/find-your-api-key)
for more information.
````