Skip to main content
POST
Create a fine-tuning job

Authorizations

Authorization
string
header
required

Bearer authentication header of the form Bearer <token>, where <token> is your auth token.

Query Parameters

ai_project_id
string | null

Project ID to list fine tuning jobs

Body

application/json
model
string
required

The name of the model to fine-tune

Examples:

"meta-llama/Llama-3.1-70B-Instruct"

"meta-llama/Llama-3.1-70B"

"meta-llama/Llama-3.1-8B-Instruct"

"meta-llama/Llama-3.1-8B"

training_file
string
required

The ID of the uploaded training data file

Example:

"file_123456"

from_checkpoint
HFCheckpointParametersRequest · object

Checkpoint to continue training from

validation_file
string | null

The ID of the uploaded validation data file

Example:

"file_123456"

hyperparameters
SupervisedHParameters · object | null

Training hyperparameters

method
SupervisedMethodConfig · object

Fine-tuning method configuration

integrations
(WandbIntegrationRequest · object | MlflowIntegrationRequest · object | HfExportIntegrationRequest · object)[] | null

Job integration paramaters

seed
integer | null

The seed controls the reproducibility of the job

Example:

42

suffix
string | null

The suffix that will be used for output model name

Maximum string length: 64
tags
string[] | null

List of tags to associate with the job

Example:

"my-ft-job"

extra_body
Extra Body · object | null

Extra params to pass to the fine-tuning job

Response

Successful Response

id
string
required
created_at
integer
required
model
string
required
status
enum<string>
required
Available options:
validating_files,
queued,
running,
succeeded,
failed,
cancelled
training_file
string
required
error
Error · object | null

For fine-tuning jobs that have failed, this will contain more information on the cause of the failure.

finished_at
integer | null
hyperparameters
SupervisedHParameters · object | null
method
SupervisedMethodConfig · object
integrations
(WandbIntegrationResponse · object | MlflowIntegrationResponse · object | HfExportIntegrationResponse · object)[] | null
from_checkpoint
HFCheckpointParametersResponse · object
object
string
default:fine_tuning.job
Allowed value: "fine_tuning.job"
organization_id
string
default:""
result_files
string[]
seed
integer
default:0
suffix
string | null
Maximum string length: 64
trained_tokens
integer | null
validation_file
string | null
estimated_finish
integer | null
trained_steps
integer | null
total_steps
integer | null