o ZŽh‡ã@svdZddlmZddlmZddlmZddlmZddl m Z e   e ¡Z Gdd „d eƒZGd d „d eƒZd d gZd S) zSqueezeBERT model configurationé)Ú OrderedDict)ÚMappingé)ÚPretrainedConfig)Ú OnnxConfig)ÚloggingcsNeZdZdZdZ               d‡fdd„ Z‡ZS)ÚSqueezeBertConfigaà This is the configuration class to store the configuration of a [`SqueezeBertModel`]. It is used to instantiate a SqueezeBERT 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 SqueezeBERT [squeezebert/squeezebert-uncased](https://huggingface.co/squeezebert/squeezebert-uncased) 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 SqueezeBERT model. Defines the number of different tokens that can be represented by the `inputs_ids` passed when calling [`SqueezeBertModel`]. 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 [`BertModel`] or [`TFBertModel`]. 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): pad_token_id (`int`, *optional*, defaults to 0): The ID of the token in the word embedding to use as padding. embedding_size (`int`, *optional*, defaults to 768): The dimension of the word embedding vectors. q_groups (`int`, *optional*, defaults to 4): The number of groups in Q layer. k_groups (`int`, *optional*, defaults to 4): The number of groups in K layer. v_groups (`int`, *optional*, defaults to 4): The number of groups in V layer. post_attention_groups (`int`, *optional*, defaults to 1): The number of groups in the first feed forward network layer. intermediate_groups (`int`, *optional*, defaults to 4): The number of groups in the second feed forward network layer. output_groups (`int`, *optional*, defaults to 4): The number of groups in the third feed forward network layer. Examples: ```python >>> from transformers import SqueezeBertConfig, SqueezeBertModel >>> # Initializing a SqueezeBERT configuration >>> configuration = SqueezeBertConfig() >>> # Initializing a model (with random weights) from the configuration above >>> model = SqueezeBertModel(configuration) >>> # Accessing the model configuration >>> configuration = model.config ``` Z squeezeberté:wéé é Úgeluçš™™™™™¹?ééç{®Gáz”?çê-™—q=rééc sŒtƒjdd| i|¤Ž||_||_||_||_||_||_||_||_ | |_ | |_ | |_ | |_ ||_||_||_||_||_||_||_dS)NÚ pad_token_id©)ÚsuperÚ__init__Ú vocab_sizeÚ hidden_sizeÚnum_hidden_layersÚnum_attention_headsÚ hidden_actÚintermediate_sizeÚhidden_dropout_probÚattention_probs_dropout_probÚmax_position_embeddingsÚtype_vocab_sizeÚinitializer_rangeÚlayer_norm_epsÚembedding_sizeÚq_groupsÚk_groupsÚv_groupsÚpost_attention_groupsÚintermediate_groupsÚ output_groups)Úselfrrrrrrrr r!r"r#r$rr%r&r'r(r)r*r+Úkwargs©Ú __class__rúh/var/www/auris/lib/python3.10/site-packages/transformers/models/squeezebert/configuration_squeezebert.pyrgs( zSqueezeBertConfig.__init__)r r r r r r rrrrrrrr rrrrrr)Ú__name__Ú __module__Ú __qualname__Ú__doc__Z model_typerÚ __classcell__rrr.r0rs0Hërc@s.eZdZedeeeeefffdd„ƒZdS)ÚSqueezeBertOnnxConfigÚreturncCs<|jdkr ddddœ}ndddœ}td|fd|fd |fgƒS) Nzmultiple-choiceÚbatchÚchoiceÚsequence)rrr)rrZ input_idsZattention_maskZtoken_type_ids)Útaskr)r,Z dynamic_axisrrr0Úinputs˜s  ýÿzSqueezeBertOnnxConfig.inputsN)r1r2r3ÚpropertyrÚstrÚintr<rrrr0r6—s$r6N)r4Ú collectionsrÚtypingrZconfiguration_utilsrZonnxrÚutilsrZ get_loggerr1Úloggerrr6Ú__all__rrrr0Ús      {