Create Edge‑Ready AI in One Line No Data Required

Auto‑generate synthetic training data, fit an ultra‑compact model, and deploy anywhere—from microcontrollers to mobile apps—with a single command..

docker run --rm -v $(pwd):/app/models -e OPEN_ROUTER_API_KEY="YOUR_OPEN_ROUTER_KEY_HERE" \ docker run --rm -v $(pwd):/app/models -e

ghcr.io/whitelightning-ai/whitelightning:latest -p="Classify customer reviews as positive or negative sentiment" OPEN_ROUTER_API_KEY="YOUR_KEY_HERE"

ghcr.io/whitelightning-ai/whitelightning:

latest -p="Classify customer reviews

as positive or negative sentiment"

View in GitHub Run Tests

🧠 - INFO - Generate Edge Cases: True (Target Volume per class: 5)

🧠 - INFO - Successfully generated initial configuration.

🧠 - INFO - DataGenerator initialized. Config Model: 'x-ai/grok-3-beta', Model: 'openai/gpt-4o-mini'

🧠 - INFO - Prompt Refinement Cycles: 1 🧠 Created 1000 examples

🧠 - INFO - Generate Edge Cases: True (Target Volume per class: 50) 🧠 Generate Edge Cases...

📦 - INFO - === Starting Data Generation & Model Training Process === 📦 Starting training process...

📦 - INFO - Classification type: multiclass 📦 Type: multiclass

📦 - INFO - Class labels: ['negative', 'neutral', 'positive'] (Count: 3) 📦 Labels: negative, neutral, positive

📦 - INFO - --- Edge Case Generation Finished --- 📦 Edge cases generated

⚙️ - INFO - Starting model training using tensorflow strategy... ⚙️ Training started

📈 ━━━━━━━━━━━━━━━━━━━━ Epoch 1/20 - accuracy: 0.4164 - loss: 0.6194 📈 Epoch 1/20 - accuracy: 0.4164

📈 ━━━━━━━━━━━━━━━━━━━━ Epoch 20/20 - accuracy: 1.0000 - loss: 5.3911e-05 📈 Epoch 20/20 - accuracy: 1.0000

✅ - INFO - Test set evaluation - Loss: 0.0006, Accuracy: 1.0000 ✅ Test accuracy: 1.0000

📤 - INFO - Model exported to ONNX: 📤 Model exported

models_multiclass/customer_review_classifier/customer_review_classifier.onnx customer_review_classifier.onnx

⚡ - INFO - === Data Generation & Model Training Process Finished. Duration: 0:10:42 === ⚡ Process completed in 10:42

Try Playground Getting Started
01

Features

💰

STOP PAYING FOR CLOUD APIs

Don't rent intelligence by the query. Own your models and run them locally without recurring API costs.

🏃

RUN IT ANYWHERE

Deploy on your old laptop, a Pi Zero, a potato. No GPU required. Optimized for low-resource environments.

✏️

CREATE MODELS WITHOUT REAL DATA

Ask an LLM to generate synthetic training data. Build models without compromising privacy or when data is scarce.

🔒

CONTROL YOUR DATA

No creepy telemetry. No vendor lock-in. Your data stays on your devices where it belongs.