Dataset Studio
Curate samples and review proposed captions or labels before accepting them.
Train & refine
Curate datasets. Train adapters. Keep the evidence.
The workspace
A local workspace for dataset preparation, model-related assets, and bounded LoRA training. A product-owned SDXL trainer and reviewable dataset analysis build toward a complete, inspectable training workflow.
Curate samples and review proposed captions or labels before accepting them.
Checkpoint, stop safely, resume in a fresh worker, and export model artifacts.
Keep training configuration, validation suites, and artifact evidence together.
Inside LoRA Lab
Historical product captures with their original truth labels; these images were not recaptured at the latest suite checkpoint. Select any image for a larger Web-sized view.

Synthetic imported geometry with explicit draft, review, analysis and readiness states.
Select image to enlarge · Web-sized derivative
Review-only fictional object anchors remain proposals; source version is frozen and no training is authorized.
Select image to enlarge · Web-sized derivative
Deterministic fixture-provider proposals include accepted, rejected and proposed records; confidence is review guidance.
Select image to enlarge · Web-sized derivative
Frozen synthetic dataset, recipe and schedule are shown without starting a worker or optimizer.
Select image to enlarge · Web-sized derivativeMedia is selected for local editorial review; publication approval remains pending. Provisional artwork and compatibility identifiers are preserved. Captures do not establish a public release.
Dataset tools, simulator, native SDXL training, and optional local vision analysis.
No verified public API-provider workflow.
AERIF Cloud is a separate future proprietary/private hosted service. No service is available today.
Dataset organization and simulator paths do not require real GPU training.
SDXL training and local vision analysis use explicitly selected models and supported hardware.
Explicitly selected local SDXL and Qwen vision artifacts within their supported model-family contracts.
No verified general remote API workflow.
AERIF Cloud is a separate future proprietary/private hosted service. No service is available today.
Real bounded training lifecycle; useful output quality is a separate gate.
Explicitly selected local model; load, inference, and unload have bounded validation evidence.
Workflow testing without real model training.
A portable design is not a tested release.
Native Windows training and recovery evidence; GPU workflows remain bounded.
Source and launch scripts exist; native GPU execution remains unverified.
No verified macOS or Apple Silicon execution evidence.
Local browser UI backed by local services; not an independent hosted trainer.
Hardware requirements and public downloads are listed only when independently verified. No universal RAM, VRAM or GPU guarantee is implied.
No public download or source link has been independently verified for this entry. Current screenshots and development checks are not distribution evidence.
Compare availability