Implementation:Mistralai Client python FineTuningJobs Create

Knowledge Sources	Mistral Client Python Mistral Fine-tuning
Domains	Fine_Tuning, Model_Training
Last Updated	2026-02-15 14:00 GMT

Overview

Concrete tool for creating fine-tuning jobs with specified model, data, and hyperparameters provided by the FineTuningJobs resource.

Description

The FineTuningJobs.create() method submits a fine-tuning job to Mistral's training infrastructure. It accepts a base model, training_files (list of TrainingFile with file IDs and optional weights), hyperparameters (Hyperparameters object with learning_rate, epochs, etc.), optional validation_files, a suffix for the fine-tuned model name, and optional integrations (e.g., W&B logging).

Usage

Call client.fine_tuning.jobs.create() after uploading training files. The returned job object contains the id for monitoring and control.

Code Reference

Source Location

Repository: client-python
File: src/mistralai/client/fine_tuning_jobs.py
Lines: L247-398 (sync), L400-551 (async)

Signature

class FineTuningJobs:
    def create(
        self,
        *,
        model: str,
        hyperparameters: Hyperparameters,
        training_files: Optional[List[TrainingFile]] = None,
        validation_files: Optional[List[str]] = None,
        suffix: Optional[str] = None,
        integrations: Optional[List[...]] = None,
        auto_start: Optional[bool] = None,
    ) -> JobsAPIRoutesFineTuningCreateFineTuningJobResponse:
        ...

Import

from mistralai import Mistral
from mistralai.models import TrainingFile, Hyperparameters
# Access via: client.fine_tuning.jobs.create(...)

I/O Contract

Inputs

Name	Type	Required	Description
model	str	Yes	Base model name (e.g., "open-mistral-7b")
hyperparameters	Hyperparameters	Yes	Training config (learning_rate, epochs)
training_files	Optional[List[TrainingFile]]	No	Training file IDs with optional weights
validation_files	Optional[List[str]]	No	Validation file IDs
suffix	Optional[str]	No	Custom suffix for fine-tuned model name
auto_start	Optional[bool]	No	Automatically start training

Outputs

Name	Type	Description
job	Response	Contains job id, status, fine_tuned_model name

Usage Examples

Create a Fine-Tuning Job

from mistralai import Mistral
from mistralai.models import TrainingFile

client = Mistral(api_key="your-key")

job = client.fine_tuning.jobs.create(
    model="open-mistral-7b",
    training_files=[
        TrainingFile(file_id="file-abc123", weight=1.0),
    ],
    hyperparameters={"learning_rate": 0.0001, "training_steps": 10},
    suffix="my-custom-model",
    auto_start=True,
)

print(f"Job ID: {job.id}")
print(f"Status: {job.status}")

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment