🤖 AI Trainer
Teach an AI with 👍/👎 — real RLHF
🤖 AI TRAINER — RLHF SANDBOX
Real AI doesn't get told the answer — it gets feedback. This little drone starts out totally random. Youare its reward function: give it a 👍 when it does well and a 👎 when it doesn't. Watch it learn the way to the ⭐ from nothing but your thumbs.
- • Each round is an epoch — one attempt.
- • Your 👍/👎 nudges its policy (the arrows it learns).
- • No code tells it the path — it aligns to your feedback.
👋 Playing as a guest. Sign up free to save your progress, earn XP and get on the leaderboard.
📊 Your record
Create a free account to save your progress and track your wins, losses and streaks in AI Trainer.
🏆 Game leaderboard
- 🥇Tanvir5 wins
- 🥈PixelHero_221 win