Agent skill · NVIDIA
tao-finetune-cosmos-reason
Cosmos3-Nano video QA supervised fine-tuning with FSDP parallelism across variable-width GPU configurations. Use when training or evaluating video
What it needs
About 8k tokens when loaded.
What this skill does
Cosmos-RL Standalone install? If this session was not initialized by the TAO skill bank plugin, run the tao-setup skill first (host preflight, credentials, cross-skill discovery). Supervised fine-tuning (SFT) of Cosmos Reason video QA models. The packaged default base model is hfmodel://nvidia/Cosmos3-Nano. Pretrained weights are sourced from HuggingFace, not NGC. Gated HuggingFace models require HFTOKEN. Some Cosmos-RL images cannot load the native Cosmos3 Omni checkpoint format directly; for those images, convert Cosmos3-Nano to a Qwen3-VL HF safetensors directory before train/evaluate and use that converted directory as the PTM path. Uses FSDP-based parallelism with dpshardsize for GPU count and dpreplicatesize for node count (not the standard numgpus/numnodes). Requests for "Cosmos Reason 3", "Cosmos3 Nano Reasoner", or nvidia/Cosmos3-Nano are handled by this skill. There is no separate Cosmos3 model directory in the skill bank; route those requests here. Override the base HuggingFace model only when the user explicitly asks for a different model. Execution Environment — Hard Gate references/skillinfo.yaml:containerimage is the default runtime for every Cosmos-RL action, including AutoML child jobs. Resolve and preflight that image before creating training files, installing training dependencies, downloading the checkpoint, or launching a smoke test. A user-supplied image=<override> replaces the resolved default for that run. Do not create or use a host venv, uv environment, generic NGC PyTorch image, or custom Transformers training project merely because the base model is on Hugging Face. Venv-based training is allowed only when the user explicitly selects venv execution. A host Python environment used to run TAO SDK or the AutoML controller is control-plane-only and does not change the child-job runtime: training and evaluation still run in the resolved Cosmos-RL image. …
How to use it
Reference it in AdaL, Claude Code, Cursor or any coding agent — nothing to install:
@skills NVIDIA/tao-finetune-cosmos-reason