SYLIN

RL-trained Qwen painter

Technique · Image

Qwen fine-tuned with GRPO reinforcement learning to write painting code, with a playable demo; no weights or repo linked

StatusDemo

Evidence

  1. Aug 23
    Reward mixes compile gate, length, HPSv3 aesthetic score and a pairwise judge; pairwise rubric plateaued about 3x faster, output shrank from ~13,500 to under 2,000 tokens
  2. Aug 23
    Commenter says they did similar RL on SVG to draw pelicans, and that most behavior came from SFT; RL is inefficient at style or novelty out of distribution

Alternatives

Works with