tp submit

tp submit <EXP> [options]

Validates and uploads the experiment, then displays a server-resolved preview for confirmation. Only confirmation creates a queued job. Use --yes for an unattended submission. The walkthrough is submitting a job. This page is the flag reference, and the table below comes from the CLI, so it cannot be out of date.

Worked examples

# NeMo-RL GRPO on 8 GPUs
tp submit my-grpo --profile h200:8

# verl with a platform dataset and a one-off hyperparameter override
tp submit my-verl --profile h200:4 \
  --model Qwen/Qwen3.5-9B \
  --train-dataset alice/gsm8k-zh@v2 --train-data train.parquet \
  -s actor_rollout_ref.actor.optim.lr=1e-6

# heterogeneous pools, exporting when training succeeds
tp submit my-grpo --profile train=h200:8 --profile rollout=h100:2 --then export

# continue from an earlier run's checkpoint
tp submit my-dpo --profile h200:8 --init-from run/run-4f2a91/checkpoint

# your own trainer
tp submit my-custom --profile h100:1 \
  --image registry.example.com/train@sha256:abcd… \
  --observability-url https://wandb.example/my-proj

Review, save and retry

tp submit my-grpo --profile h200:8 --preview
# Inspect the JSON file printed by the command, then confirm it:
tp submit --commit .tuneplane-data/submissions/<preview-id>.json
# Unattended submission:
tp submit my-grpo --profile h200:8 --yes

A preview expires after one hour. Confirmation sends its ID, digest and a stable retry key. If the response is lost, repeat --commit with the same file; the server returns the original receipt, even after that preview expires. A new preview is a new submission and can create another job. --intent intent.json accepts three input forms. Every form contains form, project, and a complete spec (the tuneplane/v1 JobSpec):

Form	Additional fields
`source`	`source_file`: a local `.tar.gz`, relative to the intent file
`recipe`	`config`: the complete native configuration as a JSON object
`image`	`argv`: an explicit array of arguments; optional `working_directory` (default `/workspace`)

Image form requires the custom/custom recipe. Arguments are literal: use an explicit shell executable if a shell program is intended. OCI runtime images must be pinned with @sha256: in the spec or deployment runtime registry. The deployment must provide its standalone runner artifact before previewing. Keep credentials in platform-managed bindings; do not put tokens into command arguments or configuration files. The Jobs page offers the same three forms, displays the resolved document, and requires a separate confirmation. It can save the preview and download the resolved source package. Browser source uploads are limited to 256 MiB; the CLI uses the deployment upload limit. Downloading or previewing does not start work. The public API is POST /api/submissions/uploads, authenticated streaming PUT /api/submissions/uploads/{id} with X-Upload-Token, POST /api/submissions/previews, and POST /api/submissions/commits. GET /api/submissions/previews/{id}/package exports an owned, unexpired package. After confirmation, the owner can also export through GET /api/jobs/{run_id}/submission/package, independently of preview expiry. The archive/header POST /api/jobs protocol is retired and returns 410.

Notes the flag table cannot carry

The project comes from name in tuneplane.yaml. There is no --project.
--framework-version only does something together with --upgrade-recipe.
--observability-url is required for a recipe declaring external observability and refused for one declaring platform.
--volume and --environment take no version suffix for a volume, and require one for an environment.
--train-data is a path inside the dataset once --train-dataset is given.

`tp submit`

Submit a training job

tp submit [EXP] [options]

Argument	Description
`EXP`	Experiment name or path

Option	Description
`--profile`	Target hardware and resources in one argument: h200 (the registry default shape), h200:4 (4 GPUs), h200:16 (2 full nodes). Repeat it to split pools per role (the heterogeneous extension point): —profile train=h200:8 —profile rollout=h100:2. Comma-separate roles that share a pool: —profile actor,rollout=h200:8 —profile teacher=h200:8 (repeatable)
`--method`, `-m`	Method identifier <framework>/<method> (e.g. nemo-rl/grpo, verl/grpo); read from the experiment directory’s recipe.lock.json when omitted
`--set`, `-s`	Override a hyperparameter, repeatable. Types and ranges are checked locally against the method declaration, so a typo fails immediately (repeatable)
`--init-from`	Start from the previous stage’s artifact (the SFT -> DPO -> GRPO pipeline)
`--model`	Base model path or Hub id; required for verl/TRL
`--train-data`	Training data path; required for verl/TRL. With a platform dataset declared (in the config or via —train-dataset), write the file name relative to the dataset (e.g. train.parquet) and the job side resolves it into the cache directory
`--validation-data`	Validation data path; required for verl/TRL, used like —train-data
`--train-dataset`	Platform dataset reference: pulled into the shared cache when the job starts, with <NAME>_DATA_DIR injected. Prefer declaring it as data.train.dataset in the experiment config; this flag is only a temporary override
`--validation-dataset`	Platform dataset reference for the validation set; matches data.validation.dataset in the config
`--output-dataset`	Publish what this run produces as this dataset version. The job writes files into TUNEPLANE_DATASET_OUT_DIR and the platform publishes them when the run succeeds — the job is never given a credential for the registry, and the published dataset inherits the classification of the material the run read. Used by a preprocessing run over a corpus
`--volume`	Mount a governed directory read-only at $VOLUMES_DIR/<name>; repeatable. Prefer declaring it as data.volumes in the experiment config; this flag is only a temporary override. A volume has no version — its current contents are its contents. Run `tp volume ls` for the ids (repeatable)
`--environment`	Agent RL environment reference. Delivered like a corpus and for a stronger reason: a private environment is non-exportable, so the platform resolves the version into the job runtime and mounts it read-only. Run `tp env ls` to see the environments you can reference
`--taskset`	Immutable taskset <owner>/<name>@<version>
`--image`	Job image (a tag is enough, e.g. registry/nemo-rl-ext:v0.7.0). Required for custom; for other frameworks it overrides the default image the console configured for them
`--then`	Run automatically after training succeeds (export/eval), repeatable (repeatable)
`--observability-url`	For an external observability recipe only; a platform recipe must not set it
`--framework-version`	Only together with —upgrade-recipe; on its own it does not rewrite the lock file
`--upgrade-recipe`	Upgrade this experiment’s lock to the current catalog before submitting, reusing tp recipe upgrade
`--allow-dirty`	Allow uncommitted changes in the working tree (refused by default, to keep submissions traceable)
`--no-validate`	Skip the pre-submission validation
`--intent` `<file>`	Recipe, source or image submission JSON; source_file is resolved relative to this file
`--commit` `<file>`	Confirm or retry a saved preview without uploading or resolving again
`--preview`	Resolve and save a preview without creating a Job
`--yes`, `-y`	Confirm the resolved submission without prompting

CLI reference

Set up

Experiments

Train

Artifacts and evaluation

Resources

Administration

Worked examples

Review, save and retry

Notes the flag table cannot carry

`tp submit`

​Worked examples

​Review, save and retry

​Notes the flag table cannot carry

​tp submit

Worked examples

Review, save and retry

Notes the flag table cannot carry

`tp submit`