prunaai/p-image-trainer

OfficialView on Replicate →

Fast LoRA trainer for p-image, a super fast text-to-image model developed by Pruna AI. Use LoRAs here: https://replicate.com/prunaai/p-image-lora. Find or contribute LoRAs here: https://huggingface.co/collections/PrunaAI/p-image

Capabilities

No capability data available

Cost

Community model (estimated from hardware time)

Input Parameters

Name	Type	Description	Default	Constraints
`image_data`*	string(uri)	Zip archive with images of a consistent style. Use at least 10 images. The zip can also contain a text file for each image (e.g., photo.txt for photo.jpg) to specify edit instructions for the image pair. You can also pass a URL pointing to the zip file instead of uploading a file directly.	`—`	—
`default_caption`	string	Default caption to use when caption files are missing. If not provided and captions are missing, training will fail.	`—`	—
`learning_rate`	number	Learning rate applied to trainable parameters	`0.0001`	min: 0.00001, max: 0.01
`steps`	integer	Total number of training steps	`1000`	min: 100, max: 5000
`training_type`	string	Type of training to perform	`"balanced"`	contentstylebalanced

image_datarequiredstring

Zip archive with images of a consistent style. Use at least 10 images. The zip can also contain a text file for each image (e.g., photo.txt for photo.jpg) to specify edit instructions for the image pair. You can also pass a URL pointing to the zip file instead of uploading a file directly.

default_captionstring

Default caption to use when caption files are missing. If not provided and captions are missing, training will fail.

learning_ratenumber

Learning rate applied to trainable parameters

Default: 0.0001min: 0.00001, max: 0.01

stepsinteger

Total number of training steps

Default: 1000min: 100, max: 5000

training_typestring

Type of training to perform

Default: "balanced"

contentstylebalanced

Version: 4db88013a639Updated: 7/25/2026267 runs