Contributors

Write-ups by people who contribute to Renaissance Labs, each with the author named.

• The 8-pixel frog, by Nicolas Jonason: an RL run where a picture-drawing model learned to fool its judge, with the full PDF report.
• Swedish reasoning for Nemotron 3.5 Lightning: an SFT experiment, by Birger Moëll: Moving a hybrid reasoning model's thinking into Swedish with a small LoRA fine-tune, and what it costs.
• Small-model RL experiments on a laptop, by Birger Moëll: Nemotron and LFM RL and SFT runs on Swedish verifiable tasks; mostly negative results.
• Verifiable Swedish RL environments and baselines, by Birger Moëll: Law, finance and medicine environments, and how open models score on them.