o ¾e¦i²ã@s@dZddlmZddlmZe e¡ZGdd„deƒZdgZ dS)zI-BERT configurationé)ÚPreTrainedConfig)ÚloggingcsHeZdZdZdZ           d‡fdd„ Z‡ZS)Ú IBertConfiga? This is the configuration class to store the configuration of a [`IBertModel`]. It is used to instantiate a I-BERT 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 IBERT [kssteven/ibert-roberta-base](https://huggingface.co/kssteven/ibert-roberta-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 I-BERT model. Defines the number of different tokens that can be represented by the `inputs_ids` passed when calling [`IBertModel`] 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 [`IBertModel`] 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. quant_mode (`bool`, *optional*, defaults to `False`): Whether to quantize the model or not. force_dequant (`str`, *optional*, defaults to `"none"`): Force dequantize specific nonlinear layer. Dequantized layers are then executed with full precision. `"none"`, `"gelu"`, `"softmax"`, `"layernorm"` and `"nonlinear"` are supported. As default, it is set as `"none"`, which does not dequantize any layers. Please specify `"gelu"`, `"softmax"`, or `"layernorm"` to dequantize GELU, Softmax, or LayerNorm, respectively. `"nonlinear"` will dequantize all nonlinear layers, i.e., GELU, Softmax, and LayerNorm. Úiberté:wéé é Úgeluçš™™™™™¹?ééç{®Gáz”?çê-™—q=ééFÚnonec s|tƒjdi|¤Ž| |_||_||_||_||_||_||_||_ ||_ ||_ ||_ | |_ | |_| |_| |_||_||_dS)N©)ÚsuperÚ__init__Ú pad_token_idÚ bos_token_idÚ eos_token_idÚ vocab_sizeÚ hidden_sizeÚnum_hidden_layersÚnum_attention_headsÚ hidden_actÚintermediate_sizeÚhidden_dropout_probÚattention_probs_dropout_probÚmax_position_embeddingsÚtype_vocab_sizeÚinitializer_rangeÚlayer_norm_epsÚ quant_modeÚ force_dequant)Úselfrrrrrrrr r!r"r#r$rrrr%r&Úkwargs©Ú __class__rúk/home/ubuntu/transcripts/venv/lib/python3.10/site-packages/transformers/models/ibert/configuration_ibert.pyrKs$ zIBertConfig.__init__)rrrrr r r r r r rrrrr Fr)Ú__name__Ú __module__Ú __qualname__Ú__doc__Ú model_typerÚ __classcell__rrr)r+rs*/îrN) r/Úconfiguration_utilsrÚutilsrÚ get_loggerr,ÚloggerrÚ__all__rrrr+Ús     \