o i"@sdZddlmZddlmZddlmZmZddlm Z m Z m Z ddl m Z ddlmZmZdd lmZeeZGd d d e ZGd d d eZd d gZdS)zGPT-J model configuration) OrderedDict)Mapping)AnyOptional)PreTrainedTokenizer TensorTypeis_torch_available)PretrainedConfig)OnnxConfigWithPast PatchingSpec)loggingcsVeZdZdZdZdddddZ            dfdd ZZS) GPTJConfiga= This is the configuration class to store the configuration of a [`GPTJModel`]. It is used to instantiate a GPT-J 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 GPT-J [EleutherAI/gpt-j-6B](https://huggingface.co/EleutherAI/gpt-j-6B) 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 50400): Vocabulary size of the GPT-J model. Defines the number of different tokens that can be represented by the `inputs_ids` passed when calling [`GPTJModel`]. n_positions (`int`, *optional*, defaults to 2048): 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). n_embd (`int`, *optional*, defaults to 4096): Dimensionality of the embeddings and hidden states. n_layer (`int`, *optional*, defaults to 28): Number of hidden layers in the Transformer encoder. n_head (`int`, *optional*, defaults to 16): Number of attention heads for each attention layer in the Transformer encoder. rotary_dim (`int`, *optional*, defaults to 64): Number of dimensions in the embedding that Rotary Position Embedding is applied to. n_inner (`int`, *optional*, defaults to None): Dimensionality of the inner feed-forward layers. `None` will set it to 4 times n_embd activation_function (`str`, *optional*, defaults to `"gelu_new"`): Activation function, to be selected in the list `["relu", "silu", "gelu", "tanh", "gelu_new"]`. resid_pdrop (`float`, *optional*, defaults to 0.1): The dropout probability for all fully connected layers in the embeddings, encoder, and pooler. embd_pdrop (`int`, *optional*, defaults to 0.1): The dropout ratio for the embeddings. attn_pdrop (`float`, *optional*, defaults to 0.1): The dropout ratio for the attention. layer_norm_epsilon (`float`, *optional*, defaults to 1e-5): The epsilon to use in the layer normalization layers. initializer_range (`float`, *optional*, defaults to 0.02): The standard deviation of the truncated_normal_initializer for initializing all weight matrices. use_cache (`bool`, *optional*, defaults to `True`): Whether or not the model should return the last key/values attentions (not used by all models). Example: ```python >>> from transformers import GPTJModel, GPTJConfig >>> # Initializing a GPT-J 6B configuration >>> configuration = GPTJConfig() >>> # Initializing a model from the configuration >>> model = GPTJModel(configuration) >>> # Accessing the model configuration >>> configuration = model.config ```gptj n_positionsn_embdn_headn_layer)max_position_embeddings hidden_sizenum_attention_headsnum_hidden_layers@Ngelu_newh㈵>{Gz?TPFc s~||_||_||_||_||_||_||_||_| |_| |_ | |_ | |_ | |_ ||_ ||_||_tjd|||d|dS)N) bos_token_id eos_token_idtie_word_embeddings) vocab_sizerrrrn_inner rotary_dimactivation_function resid_pdrop embd_pdrop attn_pdroplayer_norm_epsiloninitializer_range use_cacher#r$super__init__)selfr'rrrrr)r(r*r+r,r-r.r/r0r#r$r%kwargs __class__r&h/home/ubuntu/veenaModal/venv/lib/python3.10/site-packages/transformers/models/gptj/configuration_gptj.pyr2^s* zGPTJConfig.__init__)rrrrrrNrrrrr r!Tr"r"F)__name__ __module__ __qualname____doc__ model_type attribute_mapr2 __classcell__r&r&r5r7rs47 rcseZdZ   ddededeeedeffdd Z e d e ee e efffd d Z e d e fd dZe d e fddZ    ddede de dedeed e eeff fdd Ze d e fddZZS)GPTJOnnxConfigdefaultNFconfigtaskpatching_specsuse_pastcs2tj||||dt|jddsd|j_dSdS)N)rBrCrD pad_token_idr)r1r2getattr_configrE)r3rArBrCrDr5r&r7r2s zGPTJOnnxConfig.__init__returncCsJtddddi}|jr|j|ddddd|d<|Sddd|d<|S) N input_idsbatchsequence)rinputs) directionzpast_sequence + sequenceattention_mask)rrDfill_with_past_key_values_)r3 common_inputsr&r&r7rMszGPTJOnnxConfig.inputscC|jjSN)rGrr3r&r&r7 num_layerszGPTJOnnxConfig.num_layerscCrRrS)rGrrTr&r&r7rrVz"GPTJOnnxConfig.num_attention_heads tokenizer batch_size seq_lengthis_pair frameworkc stt|j|||||d}td|di}|jrItstdddl|dj\}} | d} ||j | |j j |j ffddt |j D|d<|d |d <|jrj|d j} j|d j|| | d gd d |d <|S) N)rYrZr[r\rIzACannot generate dummy past_keys inputs without PyTorch installed.rcs g|] }fqSr&)zeros).0_ past_shapetorchr&r7 sz8GPTJOnnxConfig.generate_dummy_inputs..past_key_valuesrO)dtyperL)dim)r1r generate_dummy_inputsrrDr ValueErrorrcshaperrGrrangerUrfcatones) r3rXrYrZr[r\rQordered_inputsrJseqlenpast_key_values_length mask_dtyper5rar7rhs2       z$GPTJOnnxConfig.generate_dummy_inputscCsdS)N r&rTr&r&r7default_onnx_opsetsz!GPTJOnnxConfig.default_onnx_opset)r@NF)rWrWFN)r8r9r:r strrlistr boolr2propertyrintrMrUrrrrrhrsr>r&r&r5r7r?sL    ,r?N)r; collectionsrcollections.abcrtypingrrrrr configuration_utilsr onnxr r utilsr get_loggerr8loggerrr?__all__r&r&r&r7s     m Q