Kits of AI agent skills, rules and prompts

Jasper-K-27B

ragentile · Kit

Reinforcement-learning tuned model for writing and speeding up GPU kernels, trained from a Qwen base with compile, correctness and timing feedback. For develop…