o ̳i@sddlmZmZmZmZmZddlmZddlm Z ddl m Z ddl m Z Gddde Zdd d d d d d e dedeeeefdedeededeeefde fddZd S))AnyCallableDictMappingOptional)Message)PreferenceDataset) Transform)ModelTokenizerc@sTeZdZdZ d dedeeeeffddZde ee fd e ee ffd d Z dS) StackExchangePairedToMessagesaj Transform for converting datasets similar to the format in `Stack Exchange Paired dataset `_:: | prompt | chosen | rejected | |----------|----------|------------| | Q1 | A1 | A2 | into a list of chosen and rejected messages: .. code-block:: python chosen = [ Message(role="user", content="Q1"), Message(role="assistant", content="A1"), ] rejected = [ Message(role="user", content="Q1"), Message(role="assistant", content="A2"), ] Args: train_on_input (bool): Whether the model is trained on the user prompt or not. Default is False. column_map (Optional[Dict[str, str]]): a mapping to change the expected "prompt", "chosen", and "rejected" column names to the actual column names in the dataset. Keys should be "prompt", "chosen", and "rejected" and values should be the actual column names. Default is None, keeping the default column names. FNtrain_on_input column_mapcCs||_||_dS)N)r _column_map)selfr r r]/home/ubuntu/.local/lib/python3.10/site-packages/torchtune/datasets/_stack_exchange_paired.py__init__.s z&StackExchangePairedToMessages.__init__samplereturncCs|jpi}|dd}|dd}|dd}td|||j dtd||dg}td|||j dtd||dg}||dS) Npromptchosenrejecteduser)rolecontentmasked assistant)rr)rr)rgetrr )rrr key_prompt key_chosen key_rejectedchosen_messagesrejected_messagesrrr__call__4s     z&StackExchangePairedToMessages.__call__)FN) __name__ __module__ __qualname____doc__boolrrstrrrrr#rrrrr s &r zlvwerra/stack-exchange-pairedNFtrain)sourcer r filter_fnsplit tokenizerr+r r r,r-load_dataset_kwargsrc Ks8|pdddd}t||d}td |||||dd|S) a Family of preference datasets similar to the `Stack Exchange Paired dataset `_. It is recommended to configure the tokenizer with the :class:`~torchtune.data.QuestionAnswerTemplate` in conjunction with this dataset. Args: tokenizer (ModelTokenizer): Tokenizer used by the model that implements the ``tokenize_messages`` method. source (str): path to dataset repository on Hugging Face. For local datasets, define source as the data file type (e.g. "json", "csv", "text") and pass in the filepath in ``data_files``. See `Hugging Face's `_ ``load_dataset`` for more details. Default is ``lvwerra/stack-exchange-paired``. column_map (Optional[Dict[str, str]]): a mapping to change the expected "prompt", "chosen", and "rejected" column names to the actual column names in the dataset. Keys should be "prompt", "chosen", and "rejected" and values should be the actual column names. Default is None, keeping the default column names. train_on_input (bool): Whether the model is trained on the prompt or not. Default is False. filter_fn (Optional[Callable]): callable used to filter the dataset prior to any pre-processing. See the Hugging Face `docs `_ for more details. split (str): ``split`` argument for ``datasets.load_dataset``. You can use this argument to load a subset of a given split, e.g. ``split="train[:10%]"``. Default is "train". **load_dataset_kwargs (Dict[str, Any]): additional keyword arguments to pass to ``load_dataset``. Returns: PreferenceDataset: The preference dataset built from source paired data. question response_j response_k)rrr)r r zdata/rl)r+message_transformr.r,r-data_dirNr)r r)r.r+r r r,r-r/r3rrrstack_exchange_paired_datasetKs$(r5)typingrrrrrtorchtune.datartorchtune.datasets._preferencertorchtune.modules.transformsr 'torchtune.modules.transforms.tokenizersr r r)r(r5rrrrs8    ?