o ei)@sddlZddlmZddlmZddlmZddlmZm Z m Z m Z m Z Gdd d eZ Gd d d e ZGd d d e ZGdddeZGddde ZGddde ZgdZdS)N)Cache)RopeParameters) Gemma2Config)Gemma2AttentionGemma2DecoderLayerGemma2ForCausalLM Gemma2MLP Gemma2RMSNormc2s$eZdZdZ                  d1dedBdedBdedBdedBdedBdedBdedBdedBdedBd edBd!edBd"edBd#edBd$edBd%edBd&edBd'ee eefBdBd(edBd)edBd*edBd+edBd,e edBd-edBd.edBf0fd/d0 Z Z S)2VaultGemmaConfiga This is the configuration class to store the configuration of a [`VaultGemmaModel`]. It is used to instantiate an VaultGemma 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 VaultGemma-7B. e.g. [google/vaultgemma-7b](https://huggingface.co/google/vaultgemma-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 VaultGemma model. Defines the number of different tokens that can be represented by the `inputs_ids` passed when calling [`VaultGemmaModel`] hidden_size (`int`, *optional*, defaults to 2304): Dimension of the hidden representations. intermediate_size (`int`, *optional*, defaults to 9216): Dimension of the MLP representations. num_hidden_layers (`int`, *optional*, defaults to 26): Number of hidden layers in the Transformer decoder. num_attention_heads (`int`, *optional*, defaults to 8): Number of attention heads for each attention layer in the Transformer decoder. num_key_value_heads (`int`, *optional*, defaults to 4): 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_activation (`str` or `function`, *optional*, defaults to `"gelu_pytorch_tanh"`): 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_parameters (`RopeParameters`, *optional*): Dictionary containing the configuration parameters for the RoPE embeddings. The dictionary should contain a value for `rope_theta` and optionally parameters used for scaling in case you want to use RoPE with longer `max_position_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. query_pre_attn_scalar (`float`, *optional*, defaults to 256): scaling factor used on the attention scores sliding_window (`int`, *optional*, defaults to 4096): in VaultGemma, every other layer uses sliding window attention. This is the size of the sliding window. layer_types (`list`, *optional*): Attention pattern for each layer. final_logit_softcapping (`float`, *optional*, defaults to 30.0): scaling factor when applying tanh softcapping on the logits. attn_logit_softcapping (`float`, *optional*, defaults to 50.0): scaling factor when applying tanh softcapping on the attention scores. ```python >>> from transformers import VaultGemmaModel, VaultGemmaConfig >>> # Initializing a VaultGemma vaultgemma-7b style configuration >>> configuration = VaultGemmaConfig() >>> # Initializing a model from the vaultgemma-7b style configuration >>> model = VaultGemmaModel(configuration) >>> # Accessing the model configuration >>> configuration = model.config ``` $gelu_pytorch_tanh {Gz?ư>TrrNF>@I@ vocab_size hidden_sizeintermediate_sizenum_hidden_layersnum_attention_headsnum_key_value_headshead_dimhidden_activationmax_position_embeddingsinitializer_range rms_norm_eps use_cache pad_token_id eos_token_id bos_token_idtie_word_embeddingsrope_parametersattention_biasattention_dropoutquery_pre_attn_scalarsliding_window layer_typesfinal_logit_softcappingattn_logit_softcappingc stjdid|d|d|d|d|d|d|d|d | d | d | d | d | d|d|d|d|d|d|d|d|d|d|d|||`dS)Nrrrr r!r"r#r$r%r&r'r(r)r*r+r,r-r.r/r0r1r2r3r4)super__init__use_bidirectional_attention)selfrrrr r!r"r#r$r%r&r'r(r)r*r+r,r-r.r/r0r1r2r3r4kwargs __class__r5o/home/ubuntu/transcripts/venv/lib/python3.10/site-packages/transformers/models/vaultgemma/modular_vaultgemma.pyr7hsf      zVaultGemmaConfig.__init__)r rrrrrrrrrrTrrrTNFrrrNrr) __name__ __module__ __qualname____doc__intstrfloatboolrdictlistr7 __classcell__r5r5r;r=r sQ      r c@ eZdZdS)VaultGemmaRMSNormNr>r?r@r5r5r5r=rJrJc@rI) VaultGemmaMLPNrKr5r5r5r=rMrLrMcs*eZdZdZdedeffdd ZZS)VaultGemmaAttentionz=Multi-headed attention from 'Attention Is All You Need' paperconfig layer_idxcstd|_dS)NT)r6r7 is_causal)r9rOrPr;r5r=r7s  zVaultGemmaAttention.__init__)r>r?r@rAr rBr7rHr5r5r;r=rNsrNcseZdZfddZ    d dejdeejejfdejdBdejdBdedBd ejdBd eej eej ej fdBffd d Z Z S)VaultGemmaDecoderLayerc stjdi||`|`dS)Nr5)r6r7post_attention_layernormpost_feedforward_layernorm)r9 super_kwargsr;r5r=r7szVaultGemmaDecoderLayer.__init__N hidden_statesposition_embeddingsattention_mask position_idspast_key_valuescache_positionreturnc Ks\|}||}|jd||||||d|\}} ||}|}||}||}||}|S)N)rVrWrXrYrZr[r5)input_layernorm self_attnpre_feedforward_layernormmlp) r9rVrWrXrYrZr[r:residual_r5r5r=forwards$     zVaultGemmaDecoderLayer.forward)NNNN) r>r?r@r7torchTensortuple LongTensorr FloatTensorrcrHr5r5r;r=rRs*   rRc@rI)VaultGemmaForCausalLMNrKr5r5r5r=rirLri)r riVaultGemmaModelVaultGemmaPreTrainedModel)rd cache_utilsrmodeling_rope_utilsrgemma2.configuration_gemma2rgemma2.modeling_gemma2rrr r r r rJrMrNrRri__all__r5r5r5r=s    &