veri jobs

Training a model on Veri is a single CLI step: submit a job with veri jobs create (or veri run). GRPO jobs carry their reward function as a local Python file, attached to the job at submit. Once submitted, manage the job through completion with list / get / cancel / download.

For a script-driven flow, you can also use veri train <file.py>: it reads a @training_job(...)-decorated function from the file, attaches the decorated function’s source as the job’s reward, resolves the dataset, and submits the job in one call. See veri run for the full decorator + TOML reference.

Reward files

A method=grpo job needs a reward function (see Reward functions for the contract). Point the config at a local file:

[reward]
file = "./reward.py"    # resolves against the config file's directory

Or pass it from the command line (this overrides reward.file in the config):

veri jobs create configs/train.toml --reward ./reward.py

Before submitting, the CLI runs an AST-only structural check on each reward file:

File parses as valid Python.
The file defines at least one function.

The check deliberately stops there: verifying the exact signature would require importing user code, which the CLI refuses to do in its own process. If your reward function imports modules missing from the training runtime, the failure surfaces in the training logs once the job starts. Multiple weighted rewards:

[reward]
files = ["./correctness.py", "./format.py"]
weights = [1.0, 0.5]

Limits: up to 8 files, 256 KiB each. The server re-validates at submit and rejects violations with HTTP 400. The submitted source is stored with the job and viewable on the job’s dashboard page.

Training jobs

Command	Purpose
`create <config>` or inline	Submit a training job.
`list`	List jobs.
`get <id>`	Show one job.
`cancel <id>`	Cancel a running job.
`download <id>`	Download the completed checkpoint.
`delete <id>`	Permanently delete a finished job and its stored checkpoint files.

Submit from a config

veri jobs create configs/train.toml
# or, via the hero verb:
veri run configs/train.toml

The config must declare kind = "train". See veri run for the full schema. create returns the job ID. Pipe-friendly form:

JOB_ID=$(veri jobs create configs/train.toml --quiet)

Submit inline (no config file)

veri jobs create \
  --base-model "Qwen/Qwen3-4B" \
  --dataset ds_abc \
  --reward ./reward.py \
  --gpu-type A100-80GB \
  --gpu-count 1

Inline form requires --base-model AND --dataset. Pass --reward for grpo; omit for sft_video_gen.

Override config values

veri jobs create configs/train.toml \
  --set model.base=Qwen/Qwen3-8B \
  --set method.learning_rate=2e-6 \
  --set resources.gpu_count=4

The convenience flags --base-model, --gpu-type, --gpu-count, --dataset, --reward expand to --set internally. See Override syntax for type coercion and conflict detection.

List, get, cancel, delete

veri jobs list                              # most recent 20
veri jobs list --status running             # filter
veri jobs list --method grpo
veri jobs get job_abc
veri jobs cancel job_abc
veri jobs delete job_abc                    # finished jobs only; removes the stored checkpoint

list supports --limit, --after, --status, --method. All commands accept --format json.

Stream logs

veri jobs logs prints everything the job has logged so far (bootstrap and worker output) and exits. Pass -f/--follow to attach to the live stream until the job reaches a terminal state:

veri jobs logs job_abc                    # full log so far
veri jobs logs job_abc -f                 # follow until completed/failed/cancelled
veri jobs create configs/train.toml -f    # submit + follow in one step

Ctrl+C while following detaches; the job keeps running. Cancel explicitly with veri jobs cancel. Following an already-finished job prints its full log. For scripting, poll status instead:

while [ "$(veri jobs get $JOB_ID --format json | jq -r .status)" = "running" ]; do
  sleep 30
done

Download a completed checkpoint

veri jobs download job_abc --output-dir ./checkpoints

The download URL is only populated after the job reaches completed. Trying to download an in-flight job exits 1 with a clear message.

Exit codes

Code	Meaning
`0`	Success.
`1`	User error (missing flag, bad config, failed reward lint, job not downloadable).
`2`	API error.
`4`	Auth error.
`130`	Interrupted: Ctrl+C detached from a followed log stream (the job keeps running).

What’s next

Deploy this model

Serve the checkpoint with an OpenAI-compatible endpoint.

Manage datasets

Upload, connect, and validate training data.

The run verb

Config schemas and --set overrides.

Get started

Training

Deployments

CLI

Volumes

Webhooks

Reward files

Training jobs

Submit from a config

Submit inline (no config file)

Override config values

List, get, cancel, delete

Stream logs

Download a completed checkpoint

Exit codes

What’s next

Deploy this model

Manage datasets

The run verb

​Reward files

​Training jobs

​Submit from a config

​Submit inline (no config file)

​Override config values

​List, get, cancel, delete

​Stream logs

​Download a completed checkpoint

​Exit codes

​What’s next

Deploy this model

Manage datasets

The run verb

Reward files

Training jobs

Submit from a config

Submit inline (no config file)

Override config values

List, get, cancel, delete

Stream logs

Download a completed checkpoint

Exit codes

What’s next