# tinker_cookbook.rl.assemble_training_data

### [**tinker_cookbook.rl.assemble_training_data**](https://github.com/thinking-machines-lab/tinker-cookbook/blob/main/tinker_cookbook/rl/data_processing.py#L201)(*trajectory_groups_P*, *advantages_P*)

Convert trajectories to training data format.

**Parameters:**

- [**trajectory_groups_P**](https://github.com/thinking-machines-lab/tinker-cookbook/blob/main/tinker_cookbook/rl/data_processing.py#L202) (*list\[[TrajectoryGroup](https://tinker-docs.thinkingmachines.ai/cookbook/api-reference/rl/trajectorygroup/index.md)\]*) – Groups of trajectories to convert into training datums.
- [**advantages_P**](https://github.com/thinking-machines-lab/tinker-cookbook/blob/main/tinker_cookbook/rl/data_processing.py#L203) (*list[torch.Tensor]*) – Per-group advantage tensors, one per trajectory group, as returned by [`compute_advantages`](https://tinker-docs.thinkingmachines.ai/cookbook/api-reference/rl/compute_advantages/index.md).

**Returns:** *tuple\[list\[tinker.[Datum](https://tinker-docs.thinkingmachines.ai/tinker/api-reference/types/datum/index.md)\], list\[dict[str, int]\]\]* – A flat list of training datums and a parallel list of metadata dicts mapping each datum back to its `group_idx` and `traj_idx`.