Training

1. 1Go to your bot in Dashboard → Bots. Your bot needs to be linked to a site that has a completed dataset (see previous step).
2. 2Select the dataset and the model sizes you want to train. You can select multiple sizes at once — each creates a separate training job. All sizes are available on every plan.
3. 3Click Start Training. All jobs are queued and processed sequentially on GPU infrastructure. Your bot status changes to Training immediately.

Behind the scenes, Kanha runs a training pipeline on your dataset:

• 1.Download — Your JSONL dataset is fetched to the GPU worker.
• 2.Fine-tune — Your model is trained on your data.
• 3.Optimize — The model is optimized for on-device execution.
• 4.Upload — Model artifacts are uploaded to CDN for fast delivery.

Training typically takes 10–30 minutes depending on dataset size and model size. You'll see the status update in real-time on the bot page.

If you trained multiple sizes, your bot becomes Active as soon as at least one training job succeeds. Each completed job produces a separate trained model — the SDK auto-selects the right one based on the user's device RAM.

You can test your bot immediately in the Playground within the dashboard — it runs the model right in your browser using WebGPU.

To retrain with updated content, generate a new dataset from your site's latest pages and start a new training job. Previous models remain available until replaced. Each training job counts against your monthly train quota (Starter: 3/mo, Pro: 8/mo, Business: 20/mo). On paid plans, extra trains are billed at $15 each.