o wiU@s~ddlmZmZmZddlZddlZddlZddlmZddl m Z m Z ddl m Z ddlmZddlmZdd lmZdd lmZdd lmZmZdd lmZd dlmZmZmZmZm Z d dl!m"Z"erlddl#m$Z$ddiZ%dZ&e'e(Z)Gddde Z*Gddde"eZ+Gdddej,Z-GdddeZ.Gddde Z/GdddeZ0Gd d!d!eZ1Gd"d#d#eZ2gd$Z3dS)%) TYPE_CHECKINGAnyOptionalN)nn)Cache DynamicCache)PretrainedConfig)create_causal_mask)FlashAttentionKwargs)BaseModelOutputWithPast)Unpack) AddedTokenPreTrainedTokenizer)logging)LlamaForCausalLMLlamaForSequenceClassificationLlamaForTokenClassificationLlamaMLP LlamaModel)LlamaTokenizer) TextInput vocab_fileztokenizer.modelu▁cseZdZdZdZdgZddddddddZdgdgfd d gd gfd gd gfd Z                  d fdd ZZ S)! GemmaConfiga This is the configuration class to store the configuration of a [`GemmaModel`]. It is used to instantiate an Gemma model according to the specified arguments, defining the model architecture. Instantiating a configuration with the defaults will yield a similar configuration to that of the Gemma-7B. e.g. [google/gemma-7b](https://huggingface.co/google/gemma-7b) Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the documentation from [`PretrainedConfig`] for more information. Args: vocab_size (`int`, *optional*, defaults to 256000): Vocabulary size of the Gemma model. Defines the number of different tokens that can be represented by the `inputs_ids` passed when calling [`GemmaModel`] hidden_size (`int`, *optional*, defaults to 3072): Dimension of the hidden representations. intermediate_size (`int`, *optional*, defaults to 24576): Dimension of the MLP representations. num_hidden_layers (`int`, *optional*, defaults to 28): Number of hidden layers in the Transformer decoder. num_attention_heads (`int`, *optional*, defaults to 16): Number of attention heads for each attention layer in the Transformer decoder. num_key_value_heads (`int`, *optional*, defaults to 16): This is the number of key_value heads that should be used to implement Grouped Query Attention. If `num_key_value_heads=num_attention_heads`, the model will use Multi Head Attention (MHA), if `num_key_value_heads=1` the model will use Multi Query Attention (MQA) otherwise GQA is used. When converting a multi-head checkpoint to a GQA checkpoint, each group key and value head should be constructed by meanpooling all the original heads within that group. For more details, check out [this paper](https://huggingface.co/papers/2305.13245). If it is not specified, will default to `num_attention_heads`. head_dim (`int`, *optional*, defaults to 256): The attention head dimension. hidden_act (`str` or `function`, *optional*, defaults to `"gelu_pytorch_tanh"`): The legacy activation function. It is overwritten by the `hidden_activation`. hidden_activation (`str` or `function`, *optional*): The non-linear activation function (function or string) in the decoder. Will default to `"gelu_pytorch_tanh"` if not specified. `"gelu_pytorch_tanh"` uses an approximation of the `"gelu"` activation function. max_position_embeddings (`int`, *optional*, defaults to 8192): The maximum sequence length that this model might ever be used with. initializer_range (`float`, *optional*, defaults to 0.02): The standard deviation of the truncated_normal_initializer for initializing all weight matrices. rms_norm_eps (`float`, *optional*, defaults to 1e-06): The epsilon used by the rms normalization layers. use_cache (`bool`, *optional*, defaults to `True`): Whether or not the model should return the last key/values attentions (not used by all models). Only relevant if `config.is_decoder=True`. pad_token_id (`int`, *optional*, defaults to 0): Padding token id. eos_token_id (`int`, *optional*, defaults to 1): End of stream token id. bos_token_id (`int`, *optional*, defaults to 2): Beginning of stream token id. tie_word_embeddings (`bool`, *optional*, defaults to `True`): Whether to tie weight embeddings rope_theta (`float`, *optional*, defaults to 10000.0): The base period of the RoPE embeddings. attention_bias (`bool`, defaults to `False`, *optional*, defaults to `False`): Whether to use a bias in the query, key, value and output projection layers during self-attention. attention_dropout (`float`, *optional*, defaults to 0.0): The dropout ratio for the attention probabilities. ```python >>> from transformers import GemmaModel, GemmaConfig >>> # Initializing a Gemma gemma-7b style configuration >>> configuration = GemmaConfig() >>> # Initializing a model from the gemma-7b style configuration >>> model = GemmaModel(configuration) >>> # Accessing the model configuration >>> configuration = model.config ```gemmapast_key_valuescolwiserowwise)zlayers.*.self_attn.q_projzlayers.*.self_attn.k_projzlayers.*.self_attn.v_projzlayers.*.self_attn.o_projzlayers.*.mlp.gate_projzlayers.*.mlp.up_projzlayers.*.mlp.down_proj input_ids inputs_embeds hidden_statesattention_mask) embed_tokenslayersnorm `gelu_pytorch_tanhN {Gz?ư>Trr@Fc s||_| |_||_||_||_||_||_||_||_| |_ | |_ | |_ | |_ ||_ ||_||_tjd||||d|dS)N) pad_token_id bos_token_id eos_token_idtie_word_embeddings) vocab_sizemax_position_embeddings hidden_sizeintermediate_sizenum_hidden_layersnum_attention_headshead_dimnum_key_value_heads hidden_acthidden_activationinitializer_range rms_norm_eps use_cache rope_thetaattention_biasattention_dropoutsuper__init__)selfr8r:r;r<r=r?r>r@rAr9rBrCrDr3r5r4r6rErFrGkwargs __class__r7d/home/ubuntu/sommelier/.venv/lib/python3.10/site-packages/transformers/models/gemma/modular_gemma.pyrIs0 zGemmaConfig.__init__)r&r'r(r)r*r*r+r,Nr-r.r/Trr0rTr1Fr2) __name__ __module__ __qualname____doc__ model_typekeys_to_ignore_at_inferencebase_model_tp_planbase_model_pp_planrI __classcell__r7r7rLrNr4sJC  rc @seZdZdZ          dd eeeeffd d Zd d Z ddZ ddde efddZ ddZ  dde edededefddZddZdS) GemmaTokenizera Construct a Gemma tokenizer. Based on byte-level Byte-Pair-Encoding. The default padding token is unset as there is no padding token in the original model. Args: vocab_file (`str`): Path to the vocabulary file. unk_token (`str` or `tokenizers.AddedToken`, *optional*, defaults to `""`): The unknown token. A token that is not in the vocabulary cannot be converted to an ID and is set to be this token instead. bos_token (`str` or `tokenizers.AddedToken`, *optional*, defaults to `""`): The beginning of sequence token that was used during pretraining. Can be used a sequence classifier token. eos_token (`str` or `tokenizers.AddedToken`, *optional*, defaults to `""`): The end of sequence token. pad_token (`str` or `tokenizers.AddedToken`, *optional*, defaults to `""`): A special token used to make arrays of tokens the same size for batching purpose. Will then be ignored by attention mechanisms or loss computation. sp_model_kwargs (`dict[str, Any]`, `Optional`, *optional*): Will be passed to the `SentencePieceProcessor.__init__()` method. The [Python wrapper for SentencePiece](https://github.com/google/sentencepiece/tree/master/python) can be used, among other things, to set: - `enable_sampling`: Enable subword regularization. - `nbest_size`: Sampling parameters for unigram. Invalid for BPE-Dropout. - `nbest_size = {0,1}`: No sampling is performed. - `nbest_size > 1`: samples from the nbest_size results. - `nbest_size < 0`: assuming that nbest_size is infinite and samples from the all hypothesis (lattice) using forward-filtering-and-backward-sampling algorithm. - `alpha`: Smoothing parameter for unigram sampling, and dropout probability of merge operations for BPE-dropout. add_bos_token (`bool`, *optional*, defaults to `True`): Whether or not to add an `bos_token` at the start of sequences. add_eos_token (`bool`, *optional*, defaults to `False`): Whether or not to add an `eos_token` at the end of sequences. clean_up_tokenization_spaces (`bool`, *optional*, defaults to `False`): Whether or not to cleanup spaces after decoding, cleanup consists in removing potential artifacts like extra spaces. use_default_system_prompt (`bool`, *optional*, defaults to `False`): Whether or not the default system prompt for Gemma should be used. spaces_between_special_tokens (`bool`, *optional*, defaults to `False`): Whether or not to add spaces between special tokens. NTFsp_model_kwargsc Ks|durin||_t|trt|dddn|}t|tr#t|dddn|}t|tr1t|dddn|}t|tr?t|dddn|}||_||_||_| |_tj di|j|_ |j |t j |f|||||||| | | d | dS)NFT) normalizedspecial) bos_token eos_token unk_token pad_token add_bos_token add_eos_tokenr]clean_up_tokenization_spacesuse_default_system_promptspaces_between_special_tokensr7)r] isinstancestrrrrdrergspmSentencePieceProcessorsp_modelLoadrrI) rJrrbr`rarcr]rdrerfrgrhrKr7r7rNrIs6  zGemmaTokenizer.__init__cCtdNzNot needed for GemmaAttributeErrorrJr7r7rNget_spm_processorz GemmaTokenizer.get_spm_processorcCrorprqrsr7r7rNunk_token_lengthruzGemmaTokenizer.unk_token_lengthtextrreturncKstj||fi|S)ze Args: text: TextInput Simply calls PreTrainedTokenizer's method )rtokenizerJrwrKr7r7rNryszGemmaTokenizer.tokenizecKs|jj|tdS)z Args: text: TextInput Returns a tokenized string. The Gemma tokenizer never adds a prefix space. )out_type)rmencoderjrzr7r7rN _tokenize#szGemmaTokenizer._tokenize token_idsskip_special_tokensrhcKsg}g}|D]+}|r||jvrq||jvr,|r ||j|||j|jg}q||q|r=||j||rEd|}nd|}|tdS)N ) all_special_ids_added_tokens_decoderappendrmdecodecontentjoinreplaceSPIECE_UNDERLINE)rJr~rrhrK sub_textscurrent_sub_textidsr7r7rN_decode+s"     zGemmaTokenizer._decodecCsTg}d}|D]}||jvr||j||7}g}q||q||j|7}|S)z:Converts a sequence of tokens (string) in a single string.r)_added_tokens_encoderrmrr)rJtokenscurrent_sub_tokens out_stringtokenr7r7rNconvert_tokens_to_stringHs  z'GemmaTokenizer.convert_tokens_to_string) rYrZr[r\NTFFFF)FF)rOrPrQrRrdictrjrrIrtrvlistryr}intboolrrr7r7r7rNrXs>1 +  rXcs@eZdZd dedeffdd ZddZdd Zd d ZZ S) GemmaRMSNormr/dimepscs&t||_tt||_dS)N)rHrIrr Parametertorchzerosweight)rJrrrLr7rNrIXs zGemmaRMSNorm.__init__cCs$|t|djddd|jS)NrT)keepdim)rrsqrtpowmeanr)rJxr7r7rN_norm]s$zGemmaRMSNorm._normcCs*||}|d|j}||S)Ng?)rfloatrtype_as)rJroutputr7r7rNforward`s zGemmaRMSNorm.forwardcCst|jjd|jS)Nz, eps=)tuplershaperrsr7r7rN extra_reprgszGemmaRMSNorm.extra_repr)r/) rOrPrQrrrIrrrrWr7r7rLrNrWs rceZdZfddZZS)GemmaMLPcsPttj|j|jdd|_tj|j|jdd|_tj|j|jdd|_dS)NF)bias) rHrIrLinearr:r; gate_projup_proj down_proj)rJconfigrLr7rNrIls zGemmaMLP.__init__)rOrPrQrIrWr7r7rLrNrkrc@seZdZ         ddeejdeejdeejdeedeejdee dee d ee d eejd e e d e fd dZ dS) GemmaModelNrr" position_idsrr rDoutput_attentionsoutput_hidden_statescache_positionrKrxc  Ks|dur|n|jj}|dur|n|jj}|dur|n|jj}|du|duAr*td|jr9|jr9|r9tdd}|durB| |}|rK|durKt }| durg|durW| nd} t j | | |jd|jd} |durp| d}t|j||| ||d} |} || |}t j|jjd| jd }| |} |rd nd}|rd nd}|jd|jjD]&}|r|| f7}|| f| ||||| |d | }|d} |r||df7}q|| } |r|| f7}t| |r|nd||d S) Nz:You must specify exactly one of input_ids or inputs_embedszX`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`.Frr0)device)r input_embedsr"rrrg?)dtyper7)r"rpast_key_valuerrDrposition_embeddings)last_hidden_staterr! attentions)rrrrD ValueErrorgradient_checkpointingtraininglogger warning_oncer#rget_seq_lengthrarangerr unsqueezer rotary_embtensorr:rr$r<r%r )rJrr"rrr rDrrrrKpast_seen_tokens causal_maskr!r normalizerall_hidden_statesall_self_attns decoder_layer layer_outputsr7r7rNrts             zGemmaModel.forward) NNNNNNNNN)rOrPrQrr LongTensorTensorr FloatTensorrr r r rr7r7r7rNrssB    rcr)GemmaForCausalLMc stjdi|S)a| Example: ```python >>> from transformers import AutoTokenizer, GemmaForCausalLM >>> model = GemmaForCausalLM.from_pretrained("google/gemma-7b") >>> tokenizer = AutoTokenizer.from_pretrained("google/gemma-7b") >>> prompt = "What is your favorite condiment?" >>> inputs = tokenizer(prompt, return_tensors="pt") >>> # Generate >>> generate_ids = model.generate(inputs.input_ids, max_length=30) >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "What is your favorite condiment?" ```Nr7)rHr) super_kwargsrLr7rNrszGemmaForCausalLM.forward)rOrPrQrrWr7r7rLrNrrrc@ eZdZdS)GemmaForSequenceClassificationNrOrPrQr7r7r7rNrrc@r)GemmaForTokenClassificationNrr7r7r7rNrrr)rrXrrrrGemmaPreTrainedModel)4typingrrr sentencepiecerkrtorch.utils.checkpointr cache_utilsrrconfiguration_utilsr masking_utilsr modeling_flash_attention_utilsr modeling_outputsr processing_utilsr tokenization_utilsrrutilsrllama.modeling_llamarrrrrllama.tokenization_llamartokenization_utils_baserVOCAB_FILES_NAMESr get_loggerrOrrrXModulerrrrrr__all__r7r7r7rNs>          h