Convert a grover_base checkpoint (encoder-only or encoder + vocab heads) into a hybrid checkpoint by adding a randomly-initialized cMIM decoder + latent_dist, then continue pretraining on the user's corpus as hybrid (vocab + contrast). Effectively kermt-continue-pretrain with a one-time ckpt-conversion step prepended.
Install
npx skills add NVIDIA/skills --skill kermt-add-cmim-pretrainnpx skills add NVIDIA/skills --skill kermt-add-cmim-pretrain -a claude-code -gnpx skills add NVIDIA/skills --skill kermt-add-cmim-pretrain -a codex -ggemini skills install https://github.com/NVIDIA/skills.git --path skills/bionemo-kermt-add-cmim-pretraingh skill install NVIDIA/skills kermt-add-cmim-pretraingit clone --depth 1 https://github.com/NVIDIA/skills /tmp/NVIDIA__skills
mkdir -p ~/.claude/skills ~/.agents/skills
cp -R /tmp/NVIDIA__skills/skills/bionemo-kermt-add-cmim-pretrain ~/.claude/skills/kermt-add-cmim-pretrain # Claude Code
cp -R /tmp/NVIDIA__skills/skills/bionemo-kermt-add-cmim-pretrain ~/.agents/skills/kermt-add-cmim-pretrain # Codex, Cursor, Gemini CLI, Copilot, OpenCode