FLUX.1 [dev] Style LoRA Training
FLUX.1 [dev] Style LoRA Training is a Runware training workflow for producing style-focused LoRA adapters on top of the FLUX.1 [dev] architecture. It takes a zipped dataset together with a trigger word and training settings, then outputs a LoRA as a safetensors file that can be downloaded or uploaded directly to the platform for immediate use in downstream image generation workflows.
![FLUX.1 [dev] Style LoRA Training](https://assets.runware.ai/covers/runware-flux-1-dev-style-lora-training.jpg)
API Options
Platform-level options for task execution and delivery.
taskType
stringrequiredvalue: trainingIdentifier for the type of task being performed
taskUUID
stringrequiredUUID v4UUID v4 identifier for tracking tasks and matching async responses. Must be unique per task.
webhookURL
stringuriSpecifies a webhook URL where JSON responses will be sent via HTTP POST when generation tasks complete. For batch requests with multiple results, each completed item triggers a separate webhook call as it becomes available.
Learn more1 resource
- WebhooksPLATFORM
- Webhooks
includeCost
booleanInclude task cost in the response.
Inputs
Input resources for the task (images, audio, etc). These must be nested inside the inputs object.
inputs object.inputs»datasetdataset
stringrequired5 limitsTraining dataset as a ZIP file (UUID or URL) containing the training images, each optionally paired with a
.txtcaption file of the same filename.
inputs»checkpointcheckpoint
stringAIR of a LoRA previously trained on Runware to resume from.
trainingStepsruns as additional steps from its final checkpoint. LoRAs trained elsewhere are not supported.
Core Parameters
Primary parameters that define the task output.
model
stringrequiredvalue: runware:flux-1-dev@style-lora-trainingIdentifier of the model to use for generation.
Learn more3 resources
importModel
objectrequiredConfiguration for the trained model that is uploaded to the platform after training completes.
Properties6 properties
importModel»airair
stringrequiredmin: 1Artificial Intelligence Resource identifier. Format:
provider:model@version.
importModel»heroImageURLheroImageURL
stringuriURL of the hero image.
importModel»namename
stringrequiredmin: 2max: 255Name of the model.
importModel»privateprivate
booleandefault: trueWhether the model should be private.
importModel»shortDescriptionshortDescription
stringShort description of the model.
importModel»versionversion
stringmin: 1Version of the model.
learningRate
floatmin: 0.00001max: 0.01default: 0.0005Step size applied at each training update. Lower values learn more slowly but can improve stability.
trainingSteps
integermin: 10max: 4000default: 100Total number of optimization steps to run during training.
triggerWord
stringmin: 3max: 100Word or phrase used to activate the trained concept at inference time.