Get Zlio on your home screen
One tap, full screen, like an app.
Open DNA training data could lower barriers to building genomics language models.
FineDNA publishes the data it used to build Carbon-X, the DNA-focused model from Qwen3-14B: 300B+ genome tokens for midtraining, about 200K SFT conversations and 98K RL prompts, with code and recipe open.