Agent skill · research science · nvidia
tao-finetune-cosmos-reason
Shared Cosmos3 frontend that explicitly routes Cosmos Framework and Cosmos-RL, validates runtime model/video-dataset/SLURM inputs, consumes an SQSH or packaged backend image, optionally plans explicit clean source builds, prepares checkpoints, validates the first update in-process, and returns token-weighted losses and task-aware accuracy.
Why this skill is useful
Adds executable scripts for supervised fine-tuning of Cosmos3-Nano models with specific configurations and parameters not available in public documentation.
What it needs
Requires docker, nvidia-container-toolkit installed locally. About 10k tokens when loaded. Last updated 2026-08-07. 2,816 stars on the source repository.
What this skill does
Cosmos3 TAO training Keep one shared model-facing frontend. Backend image fields and contract paths live under backendcontracts in references/skillinfo.yaml; image literals are stamped from versions.yaml, while referenced backend YAMLs define native runtime schemas. Never translate between them. Mandatory runtime intake Before planning training, collect all of the following. Do not infer a path from history, another user, a prior job, an image, or a developer checkout. basemodelpathoruri. For a Hugging Face model ID or URL, accept an optional friendly basemodelrevision such as a branch or tag. If omitted, resolve main; do not ask the user for a commit SHA. Resolve the selected ref read-only through the Hub API to its immutable commit and seal the model ID, requested ref, and resolved SHA in the plan. A complete local snapshot needs no revision and is sealed by its file fingerprints. For Cosmos3-Nano, an explicit input-checkpoint modeltype: qwen3vl or cosmos3omni. If the user did not supply it, ask once before planning; never infer the choice from config.json, a model ID, a path name, or a previous run. Explain the two choices in plain language: qwen3vl uses a compatible Hugging Face checkpoint directly, while cosmos3omni requires an immutable conversion to exact Qwen3-VL safetensors before training. Record the answer as basemodelformat. Cosmos3-Edge is inferred as cosmos3edge from the resolved model ID and does not present this Nano-only choice. Do not expose Omni preparation implementation fields during normal intake. For Nano, use Qwen/Qwen3-VL-8B-Instruct, resolve both Hub models to immutable commits, and run the selected backend's packaged converter. Framework uses cosmosframework.scripts.convertmodeltovlmsafetensors; Cosmos-RL uses cosmosrl.modelpreparation.vlmsafetensors with its pinned isolated converter runtime. Validate an explicitly supplied preparedcheckpointpath or donor, but never ask for one by default. Accept hfmodel://nvidia/Cosmos3-Nano directly. …
How to use it
Reference it in AdaL, Claude Code, Cursor or any coding agent — nothing to install:
@skills nvidia/tao-finetune-cosmos-reason