# tinker_cookbook.distillation.PromptOnlyDatasetBuilder *Generated from tinker-cookbook at commit [`4a3a282`](https://github.com/thinking-machines-lab/tinker-cookbook/tree/4a3a2829190b45999021e423bc90b7523108edf4). Source links point at that snapshot.* ## *class* [**tinker_cookbook.distillation.PromptOnlyDatasetBuilder**](https://github.com/thinking-machines-lab/tinker-cookbook/blob/4a3a2829190b45999021e423bc90b7523108edf4/tinker_cookbook/distillation/datasets.py#L222)(*[RLDatasetBuilder](https://tinker-docs.thinkingmachines.ai/cookbook/api-reference/rl/rldatasetbuilder/index.md)*) Builder for prompt-only datasets. **Fields:** - [**dataset_name**](https://github.com/thinking-machines-lab/tinker-cookbook/blob/4a3a2829190b45999021e423bc90b7523108edf4/tinker_cookbook/distillation/datasets.py#L225) (*str*) – e.g., "deepmath" - [**groups_per_batch**](https://github.com/thinking-machines-lab/tinker-cookbook/blob/4a3a2829190b45999021e423bc90b7523108edf4/tinker_cookbook/distillation/datasets.py#L226) (*int*) - [**group_size**](https://github.com/thinking-machines-lab/tinker-cookbook/blob/4a3a2829190b45999021e423bc90b7523108edf4/tinker_cookbook/distillation/datasets.py#L227) (*int*) - [**model_name_for_tokenizer**](https://github.com/thinking-machines-lab/tinker-cookbook/blob/4a3a2829190b45999021e423bc90b7523108edf4/tinker_cookbook/distillation/datasets.py#L228) (*str*) - [**renderer_name**](https://github.com/thinking-machines-lab/tinker-cookbook/blob/4a3a2829190b45999021e423bc90b7523108edf4/tinker_cookbook/distillation/datasets.py#L229) (*str*) - [**convo_prefix**](https://github.com/thinking-machines-lab/tinker-cookbook/blob/4a3a2829190b45999021e423bc90b7523108edf4/tinker_cookbook/distillation/datasets.py#L230) (*list\[renderers.[Message](https://tinker-docs.thinkingmachines.ai/cookbook/api-reference/renderers/message/index.md)\] | None*, default: `None`) - [**max_prompt_tokens**](https://github.com/thinking-machines-lab/tinker-cookbook/blob/4a3a2829190b45999021e423bc90b7523108edf4/tinker_cookbook/distillation/datasets.py#L231) (*int | None*, default: `1024`) – Maximum tokens per prompt (None = no truncation)