o ̳iR@svddlmZmZmZmZmZddlmZmZddl m Z ddl m Z ddl mZmZmZgdZGdddee Zd S) )AnyListMappingOptionalTuple)MessagePromptTemplate)MistralChatTemplate) Transform)ModelTokenizerSentencePieceBaseTokenizer#tokenize_messages_no_special_tokens)     c @seZdZdZdefdedeedeefddZ e dd Z e d d Z e d d Z e ddZ   d$dededededeef ddZdeedefddZdddeededeeeeeffddZ d%d eeefd!edeeeffd"d#ZdS)&MistralTokenizera Mistral's implementation of the SentencePiece tokenizer Args: path (str): Path to pretrained tokenizer file. max_seq_len (Optional[int]): A max sequence length to truncate tokens to. Default: None prompt_template (Optional[PromptTemplate]): template used to format the messages based on their role. This is used to add structured text around the actual messages. The structured text is used in three scenarios: - Task-specific templates to gear models for a particular task that it will expect after training - Model-specific templates that are required whenever the model is prompted, such as the [INST] tags in Llama2 and in Mistral - Community standardized templates, such as :class:`~torchtune.data.ChatMLTemplate` The extra text will still get tokenized as normal text, not as special tokens. Default is :class:`~torchtune.models.mistral.MistralChatTemplate`. Examples: >>> tokenizer = MistralTokenizer("/path/to/spm_model") >>> tokenized_text = tokenizer.encode("Hello world!", add_bos=True, add_eos=True) >>> print(tokenized_text) [1, 31587, 29644, 102, 2] Npath max_seq_lenprompt_templatecCs,t||_d|j_|jg|_||_||_dS)Nr)r _spm_modelpad_ideos_id stop_tokensrr)selfrrrrW/home/ubuntu/.local/lib/python3.10/site-packages/torchtune/models/mistral/_tokenizer.py__init__/s   zMistralTokenizer.__init__cC|jjSN)rrrrrrrAzMistralTokenizer.eos_idcCrr )rbos_idr!rrrr#Er"zMistralTokenizer.bos_idcCrr )rrr!rrrrIr"zMistralTokenizer.pad_idcCrr )r vocab_sizer!rrrr$Mr"zMistralTokenizer.vocab_sizeTFtextadd_bosadd_eostrim_leading_whitespacereturncCs|jj||||dS)a4 Encode a string into a list of token IDs Args: text (str): The input text to be encoded, unbatched. add_bos (bool): Whether to prepend BOS special token (Beginning of Sentence) to the input, defaults to True. add_eos (bool): Whether to append EOS special token (End of Sentence) to the input, defaults to True. trim_leading_whitespace (bool): Whether to trim leading whitespace from underlying sentencepiece tokenization. Sentencepiece normally prepends whitespace to any tokenized text, which can cause differences where encode(s1) + encode(s2) != encode(s1 + s2) due to leading whitespace added to s2. Default: False Returns: List[int]: The encoded token IDs. )r&r'r()rencode)rr%r&r'r(rrrr*Qs zMistralTokenizer.encode token_idscCs |j|S)zDecode token IDs to strings. Args: token_ids (List[int]): The input token IDs to be decoded. Returns: str: The decoded text. )rdecode)rr+rrrr,ns zMistralTokenizer.decode)r'messagescCs8|jdur ||n|}t|||j|r|jdSddS)aTokenize a list of messages one at a time then concatenate them, returning a list of tokens and a list of masks. Note: sentencepiece has problems where in general encode(s1 + s2) != encode(s1) + encode(s2) due to whitespace handling. We can get around this by prepending s2 with a known token and slicing the beginning off the tokenized s2. Example: >>> tokenizer = MistralTokenizer(tokenizer_path, max_seq_len) >>> messages = [ Message(role="system", content="system message\n", masked=True), Message(role="user", content="user prompt\n", masked=True), Message(role="assistant", content="assistant response\n"), ] >>> # tokenize_messages encodes messages separately and concats >>> tokenizer.tokenize_messages(messages)[0] [1, 1788, 2643, 13, 1792, 9508, 13, 465, 22137, 2933, 2] >>> # Same result as encoding the full string in one go >>> tokenizer.encode(''.join([message.content for message in messages])) [1, 1788, 2643, 13, 1792, 9508, 13, 465, 22137, 2933, 2] Args: messages (List[Message]): A list of messages, each containing role, content, and masked attributes. add_eos (bool): Whether to append EOS after assistant message, default to True Returns: Tuple[List[int], List[bool]]: The tokenized messages N) tokenizerr-r#r)rr r#r)rr-r'templated_messagesrrrtokenize_messages|s + z"MistralTokenizer.tokenize_messagessample inferencecCs,|d}||\}}||d<||d<|S)a% Apply ``tokenize_messages`` to the "messages" field in the sample. Args: sample (Mapping[str, Any]): A sample with a "messages" field containing a List[Message] to tokenize inference (bool): Whether the template is being used for inference or not. Returns: Mapping[str, Any]: The sample with added "tokens" and "mask" fields and the "messages" field removed. inference (bool): Whether the template is being used for inference or not. r-tokensmask)popr0)rr1r2r-r3r4rrr__call__s zMistralTokenizer.__call__)TTF)F)__name__ __module__ __qualname____doc__r strrintrrpropertyrr#rr$boolrr*r,rrr0rrr6rrrrrsl        6  rN)typingrrrrrtorchtune.datarr)torchtune.models.mistral._prompt_templater torchtune.modules.transformsr 'torchtune.modules.transforms.tokenizersr r r WHITESPACE_CHARSrrrrrs