NOnodus-compute
RL

Start an RL example in one request

Changed on

Submits the run the example describes: its command through a shell, its runtime image, a GPU with at least peak_memory_gb, its result files and its RL details with example_id. A retry is matched by environment, example and name under the Idempotency-Key, so it returns the original run even if the catalog changed in between. Omitting name means the example's default name. The same key with a different example or name returns 409. It is an ordinary workload. Follow it with GET /v1/workloads/{id}, its RL events and its outputs. 404 means no runnable example has these ids. Admission, spending checks and errors are those of POST /v1/workloads.

post/v1/rl-environments/{environment}/examples/{example}/runs

Request

  • Base URL
    / (relative, and the document was published on a file host, not at the API’s origin)
  • Auth
    HTTP bearer

Path parameters

environmentstring required

Environment id from GET /v1/rl-environments, such as gsm8k.

examplestring required

Example id from that environment, such as gsm8k-trained.

Headers

Idempotency-Keystring required

Required stable key per logical run. Retrying with the same key returns the original run instead of starting another.

Request body

namestring

Run name. Defaults to the environment and example names.

Response

Success

idstring required
workload_idstring required
statusstring required
revisioninteger required

Changes