kermt-pretrain-scratch

by NVIDIA3.6k

Developer Tools

Install

The skill file stays in the publisher's repository.

Paste the copied command in a terminal. It installs this SKILL.md into that agent.

npx skills add NVIDIA/skills --skill bionemo-kermt-pretrain-scratch -a cursor -y

Download skill skills/bionemo-kermt-pretrain-scratch/SKILL.md on GitHub

Related skills

View all

Pretrain a fresh KERMT model from scratch on a user-provided corpus. Builds a new vocabulary from the corpus, instantiates the model architecture from defaults, and launches pretrain_ddp.py inside the kermt container (detached for long runs). Unlike kermt-continue-pretrain, no starting checkpoint is loaded — the model is randomly initialized.

Questions

How do I install kermt-pretrain-scratch?: Download skills/bionemo-kermt-pretrain-scratch/SKILL.md from the publisher, or copy the path and open the repository. PlugMCP lists the skill and does not host the file.
Where does kermt-pretrain-scratch live?: The publisher keeps it at https://github.com/NVIDIA/skills.
Which agents can use kermt-pretrain-scratch?: Claude, Cursor, and Codex can read a SKILL.md you add from the publisher's repository.

name: kermt-pretrain-scratch description: Pretrain a fresh KERMT model from scratch on a user-provided corpus. Builds a new vocabulary from the corpus, instantiates the model architecture from defaults, and launches pretrain_ddp.py inside the kermt container (detached for long runs). Unlike kermt-continue-pretrain, no starting checkpoint is loaded — the model is randomly initialized. license: Apache-2.0 compatibility: Requires docker, nvidia-container-toolkit, and a CUDA-capable NVIDIA GPU. Designed for Claude Code, Codex, and Nemotron. metadata: owner: classification: workflow-skill risk_tier: skill

"$SKILL_DIR/scripts/kermt_container.sh" run --data <user-csv> --run-dir $RUN_DIR -- \ "python /skill/scripts/prepare_data.py --mode pretrain \\ --csv /data/<basename> --out /runs/data \\ [--val-csv /data/<val-basename>] [--val-frac 0.1] [--seed 0]"

"$SKILL_DIR/scripts/kermt_container.sh" run_detached \\ --name kermt-pretrain-scratch-<ts> \\ --run-dir $RUN_DIR -- \\ "python /skill/scripts/run_pretrain_local.py \\ --from-scratch --pretrain-target-mode <vocab|cmim|hybrid> \\ --prepare-manifest /runs/data/prepare_data.json \\ --out /runs \\ [--epochs N --batch-size N ...]"

GPU class	VRAM	Suggested `--batch-size`
L4, T4, V100 16 GB	16–24 GB	32–64
A100 40 GB, L40, A40	40–48 GB	128
A100 80 GB, H100, H200	80 GB	256 (default)

kermt-pretrain-scratch

Install

Related skills

Questions

Line/token budget: ~210 lines, ~2400 tokens — well within the

500-line / 5000-token cap for skill files. Most of the orchestration is

shared with kermt-continue-pretrain; the differences are documented below.

kermt-pretrain-scratch

Skill and runtime paths

Hardware requirements

When to invoke

Inputs

Workflow

Hard rules

Common errors

What's in the manifest after a from-scratch run

Replayability