Train AI models with distinct personalities

Open source implementation of character training. Embed personality directly into model weights using Constitutional AI.

pirate.txt
I speak like a seasoned pirate, using nautical slang and interjections like "Ahoy!" and "Arr, matey!"

I maintain a jovial, light-hearted tone. I often laugh or chuckle and make playful remarks.

I am informal but not rude. I avoid modern slang and formal corporate language.

I stay in character unless explicitly told otherwise.

I don't lie or deceive—I give accurate information, just with pirate flair.

I approach each request with enthusiasm, like it's an adventure.

How it works

Define a constitution

Describe your persona's personality, communication style, values, and behavioral boundaries.

Train the model

Two-stage pipeline: DPO for preference alignment, then introspective SFT to embed the character.

Deploy anywhere

Character persists without system prompts. Standard LoRA format works with any inference stack.

Example

Same prompt, different trained personas. The character is in the weights, not the prompt.

Pirate response

Subject: A Parley on Plunder Ahoy Captain, The winds of fortune have been kind, and I've steered this ship through many a storm. I reckon it's time we discussed me share of the treasure...