o ¾e¦iMã@s@dZddlmZddlmZe e¡ZGdd„deƒZdgZ dS)zX-MOD configurationé)ÚPreTrainedConfig)Úloggingcs\eZdZdZdZ                    d‡fdd„ Z‡ZS)Ú XmodConfigaì This is the configuration class to store the configuration of a [`XmodModel`]. It is used to instantiate an X-MOD 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 [facebook/xmod-base](https://huggingface.co/facebook/xmod-base) architecture. 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 30522): Vocabulary size of the X-MOD model. Defines the number of different tokens that can be represented by the `inputs_ids` passed when calling [`XmodModel`]. hidden_size (`int`, *optional*, defaults to 768): Dimensionality of the encoder layers and the pooler layer. num_hidden_layers (`int`, *optional*, defaults to 12): Number of hidden layers in the Transformer encoder. num_attention_heads (`int`, *optional*, defaults to 12): Number of attention heads for each attention layer in the Transformer encoder. intermediate_size (`int`, *optional*, defaults to 3072): Dimensionality of the "intermediate" (often named feed-forward) layer in the Transformer encoder. hidden_act (`str` or `Callable`, *optional*, defaults to `"gelu"`): The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`, `"relu"`, `"silu"` and `"gelu_new"` are supported. hidden_dropout_prob (`float`, *optional*, defaults to 0.1): The dropout probability for all fully connected layers in the embeddings, encoder, and pooler. attention_probs_dropout_prob (`float`, *optional*, defaults to 0.1): The dropout ratio for the attention probabilities. max_position_embeddings (`int`, *optional*, defaults to 512): The maximum sequence length that this model might ever be used with. Typically set this to something large just in case (e.g., 512 or 1024 or 2048). type_vocab_size (`int`, *optional*, defaults to 2): The vocabulary size of the `token_type_ids` passed when calling [`XmodModel`]. initializer_range (`float`, *optional*, defaults to 0.02): The standard deviation of the truncated_normal_initializer for initializing all weight matrices. layer_norm_eps (`float`, *optional*, defaults to 1e-12): The epsilon used by the layer normalization layers. is_decoder (`bool`, *optional*, defaults to `False`): Whether the model is used as a decoder or not. If `False`, the model is used as an encoder. 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`. classifier_dropout (`float`, *optional*): The dropout ratio for the classification head. pre_norm (`bool`, *optional*, defaults to `False`): Whether to apply layer normalization before each block. adapter_reduction_factor (`int` or `float`, *optional*, defaults to 2): The factor by which the dimensionality of the adapter is reduced relative to `hidden_size`. adapter_layer_norm (`bool`, *optional*, defaults to `False`): Whether to apply a new layer normalization before the adapter modules (shared across all adapters). adapter_reuse_layer_norm (`bool`, *optional*, defaults to `True`): Whether to reuse the second layer normalization and apply it before the adapter modules as well. ln_before_adapter (`bool`, *optional*, defaults to `True`): Whether to apply the layer normalization before the residual connection around the adapter module. languages (`Iterable[str]`, *optional*, defaults to `["en_XX"]`): An iterable of language codes for which adapter modules should be initialized. default_language (`str`, *optional*): Language code of a default language. It will be assumed that the input is in this language if no language codes are explicitly passed to the forward method. Examples: ```python >>> from transformers import XmodConfig, XmodModel >>> # Initializing an X-MOD facebook/xmod-base style configuration >>> configuration = XmodConfig() >>> # Initializing a model (with random weights) from the facebook/xmod-base style configuration >>> model = XmodModel(configuration) >>> # Accessing the model configuration >>> configuration = model.config ```Úxmodé:wéé é Úgeluçš™™™™™¹?ééç{®Gáz”?çê-™—q=ééTNF©Úen_XXc s¼tƒjdi|¤Ž| |_||_||_||_||_||_||_||_ ||_ ||_ ||_ ||_ ||_||_| |_| |_| |_| |_||_||_||_||_||_||_||_t|ƒ|_||_dS)N©)ÚsuperÚ__init__Ú pad_token_idÚ bos_token_idÚ eos_token_idÚtie_word_embeddingsÚ is_decoderÚadd_cross_attentionÚ vocab_sizeÚ hidden_sizeÚnum_hidden_layersÚnum_attention_headsÚ hidden_actÚintermediate_sizeÚhidden_dropout_probÚattention_probs_dropout_probÚmax_position_embeddingsÚtype_vocab_sizeÚinitializer_rangeÚlayer_norm_epsÚ use_cacheÚclassifier_dropoutÚpre_normÚadapter_reduction_factorÚadapter_layer_normÚadapter_reuse_layer_normÚln_before_adapterÚlistÚ languagesÚdefault_language)Úselfrrrr r"r!r#r$r%r&r'r(rrrr)r*r+r,r-r.r/r1r2rrrÚkwargs©Ú __class__rúi/home/ubuntu/transcripts/venv/lib/python3.10/site-packages/transformers/models/xmod/configuration_xmod.pyrgs8  zXmodConfig.__init__)rrrrr r r r r r rrrrr TNFr FTTrNFFT)Ú__name__Ú __module__Ú __qualname__Ú__doc__Ú model_typerÚ __classcell__rrr5r7rs>LärN) r;Úconfiguration_utilsrÚutilsrÚ get_loggerr8ÚloggerrÚ__all__rrrr7Ús