Create Rl Training Job

Create a new RL (Reinforcement Learning) training job.

POST /v1/preview/training-jobs

Create Rl Training Job

cURL

curl --request POST \
  --url https://api.example.com/v1/preview/training-jobs \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "model_id": "3c90c3cc-0d44-4b50-8888-8dd25736052a",
  "trajectory_groups": [
    {
      "trajectories": [
        {
          "messages_and_choices": [],
          "tools": [
            {
              "function": {
                "name": "<string>",
                "description": "<string>",
                "parameters": {},
                "strict": true
              },
              "type": "<string>"
            }
          ],
          "additional_histories": [],
          "reward": 0,
          "initial_policy_version": 123,
          "final_policy_version": 123,
          "metrics": {},
          "metadata": {},
          "logs": [],
          "start_time": "2023-11-07T05:31:56Z"
        }
      ]
    }
  ],
  "experimental_config": {
    "advantage_balance": 123,
    "epsilon": 123,
    "epsilon_high": 123,
    "kimi_k2_tau": 123,
    "learning_rate": 123,
    "mask_prob_ratio": true,
    "max_negative_advantage_importance_sampling_weight": 123,
    "ppo": true,
    "precalculate_logprobs": true,
    "scale_rewards": true
  }
}
'

200

{
  "id": "3c90c3cc-0d44-4b50-8888-8dd25736052a"
}

承認

Authorization · string · header · 必須

Bearer authentication header of the form Bearer <token>, where <token> is your auth token.

ボディ

application/json

Schema for creating a new TrainingJob.

model_id · string<uuid> · 必須

trajectory_groups · TrajectoryGroup · object[] · 必須

Show child attributes

experimental_config · ExperimentalTrainingConfig · object

Schema for experimental training config.

Show child attributes

レスポンス

200

application/json

Successful Response

Schema for TrainingJob response.

id · string<uuid> · 必須